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
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
|
/* SPDX-License-Identifier: LGPL-2.1+ */
/***
This file is part of systemd.
Copyright 2010 Lennart Poettering
***/
#include <errno.h>
#include <fcntl.h>
#include <linux/kd.h>
#include <signal.h>
#include <stdio_ext.h>
#include <string.h>
#include <sys/epoll.h>
#include <sys/inotify.h>
#include <sys/ioctl.h>
#include <sys/reboot.h>
#include <sys/timerfd.h>
#include <sys/wait.h>
#include <unistd.h>
#if HAVE_AUDIT
#include <libaudit.h>
#endif
#include "sd-daemon.h"
#include "sd-messages.h"
#include "sd-path.h"
#include "alloc-util.h"
#include "audit-fd.h"
#include "boot-timestamps.h"
#include "bus-common-errors.h"
#include "bus-error.h"
#include "bus-kernel.h"
#include "bus-util.h"
#include "clean-ipc.h"
#include "clock-util.h"
#include "dbus-job.h"
#include "dbus-manager.h"
#include "dbus-unit.h"
#include "dbus.h"
#include "dirent-util.h"
#include "env-util.h"
#include "escape.h"
#include "exec-util.h"
#include "execute.h"
#include "exit-status.h"
#include "fd-util.h"
#include "fileio.h"
#include "fs-util.h"
#include "hashmap.h"
#include "io-util.h"
#include "label.h"
#include "locale-setup.h"
#include "log.h"
#include "macro.h"
#include "manager.h"
#include "missing.h"
#include "mkdir.h"
#include "parse-util.h"
#include "path-lookup.h"
#include "path-util.h"
#include "process-util.h"
#include "ratelimit.h"
#include "rm-rf.h"
#include "signal-util.h"
#include "special.h"
#include "stat-util.h"
#include "string-table.h"
#include "string-util.h"
#include "strv.h"
#include "strxcpyx.h"
#include "terminal-util.h"
#include "time-util.h"
#include "transaction.h"
#include "umask-util.h"
#include "unit-name.h"
#include "user-util.h"
#include "util.h"
#include "virt.h"
#include "watchdog.h"
#define NOTIFY_RCVBUF_SIZE (8*1024*1024)
#define CGROUPS_AGENT_RCVBUF_SIZE (8*1024*1024)
/* Initial delay and the interval for printing status messages about running jobs */
#define JOBS_IN_PROGRESS_WAIT_USEC (5*USEC_PER_SEC)
#define JOBS_IN_PROGRESS_PERIOD_USEC (USEC_PER_SEC / 3)
#define JOBS_IN_PROGRESS_PERIOD_DIVISOR 3
/* If there are more than 1K bus messages queue across our API and direct busses, then let's not add more on top until
* the queue gets more empty. */
#define MANAGER_BUS_BUSY_THRESHOLD 1024LU
/* How many units and jobs to process of the bus queue before returning to the event loop. */
#define MANAGER_BUS_MESSAGE_BUDGET 100U
static int manager_dispatch_notify_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata);
static int manager_dispatch_cgroups_agent_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata);
static int manager_dispatch_signal_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata);
static int manager_dispatch_time_change_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata);
static int manager_dispatch_idle_pipe_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata);
static int manager_dispatch_user_lookup_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata);
static int manager_dispatch_jobs_in_progress(sd_event_source *source, usec_t usec, void *userdata);
static int manager_dispatch_run_queue(sd_event_source *source, void *userdata);
static int manager_dispatch_sigchld(sd_event_source *source, void *userdata);
static int manager_run_environment_generators(Manager *m);
static int manager_run_generators(Manager *m);
static void manager_watch_jobs_in_progress(Manager *m) {
usec_t next;
int r;
assert(m);
/* We do not want to show the cylon animation if the user
* needs to confirm service executions otherwise confirmation
* messages will be screwed by the cylon animation. */
if (!manager_is_confirm_spawn_disabled(m))
return;
if (m->jobs_in_progress_event_source)
return;
next = now(CLOCK_MONOTONIC) + JOBS_IN_PROGRESS_WAIT_USEC;
r = sd_event_add_time(
m->event,
&m->jobs_in_progress_event_source,
CLOCK_MONOTONIC,
next, 0,
manager_dispatch_jobs_in_progress, m);
if (r < 0)
return;
(void) sd_event_source_set_description(m->jobs_in_progress_event_source, "manager-jobs-in-progress");
}
#define CYLON_BUFFER_EXTRA (2*STRLEN(ANSI_RED) + STRLEN(ANSI_HIGHLIGHT_RED) + 2*STRLEN(ANSI_NORMAL))
static void draw_cylon(char buffer[], size_t buflen, unsigned width, unsigned pos) {
char *p = buffer;
assert(buflen >= CYLON_BUFFER_EXTRA + width + 1);
assert(pos <= width+1); /* 0 or width+1 mean that the center light is behind the corner */
if (pos > 1) {
if (pos > 2)
p = mempset(p, ' ', pos-2);
if (log_get_show_color())
p = stpcpy(p, ANSI_RED);
*p++ = '*';
}
if (pos > 0 && pos <= width) {
if (log_get_show_color())
p = stpcpy(p, ANSI_HIGHLIGHT_RED);
*p++ = '*';
}
if (log_get_show_color())
p = stpcpy(p, ANSI_NORMAL);
if (pos < width) {
if (log_get_show_color())
p = stpcpy(p, ANSI_RED);
*p++ = '*';
if (pos < width-1)
p = mempset(p, ' ', width-1-pos);
if (log_get_show_color())
strcpy(p, ANSI_NORMAL);
}
}
void manager_flip_auto_status(Manager *m, bool enable) {
assert(m);
if (enable) {
if (m->show_status == SHOW_STATUS_AUTO)
manager_set_show_status(m, SHOW_STATUS_TEMPORARY);
} else {
if (m->show_status == SHOW_STATUS_TEMPORARY)
manager_set_show_status(m, SHOW_STATUS_AUTO);
}
}
static void manager_print_jobs_in_progress(Manager *m) {
_cleanup_free_ char *job_of_n = NULL;
Iterator i;
Job *j;
unsigned counter = 0, print_nr;
char cylon[6 + CYLON_BUFFER_EXTRA + 1];
unsigned cylon_pos;
char time[FORMAT_TIMESPAN_MAX], limit[FORMAT_TIMESPAN_MAX] = "no limit";
uint64_t x;
assert(m);
assert(m->n_running_jobs > 0);
manager_flip_auto_status(m, true);
print_nr = (m->jobs_in_progress_iteration / JOBS_IN_PROGRESS_PERIOD_DIVISOR) % m->n_running_jobs;
HASHMAP_FOREACH(j, m->jobs, i)
if (j->state == JOB_RUNNING && counter++ == print_nr)
break;
/* m->n_running_jobs must be consistent with the contents of m->jobs,
* so the above loop must have succeeded in finding j. */
assert(counter == print_nr + 1);
assert(j);
cylon_pos = m->jobs_in_progress_iteration % 14;
if (cylon_pos >= 8)
cylon_pos = 14 - cylon_pos;
draw_cylon(cylon, sizeof(cylon), 6, cylon_pos);
m->jobs_in_progress_iteration++;
if (m->n_running_jobs > 1) {
if (asprintf(&job_of_n, "(%u of %u) ", counter, m->n_running_jobs) < 0)
job_of_n = NULL;
}
format_timespan(time, sizeof(time), now(CLOCK_MONOTONIC) - j->begin_usec, 1*USEC_PER_SEC);
if (job_get_timeout(j, &x) > 0)
format_timespan(limit, sizeof(limit), x - j->begin_usec, 1*USEC_PER_SEC);
manager_status_printf(m, STATUS_TYPE_EPHEMERAL, cylon,
"%sA %s job is running for %s (%s / %s)",
strempty(job_of_n),
job_type_to_string(j->type),
unit_description(j->unit),
time, limit);
}
static int have_ask_password(void) {
_cleanup_closedir_ DIR *dir;
struct dirent *de;
dir = opendir("/run/systemd/ask-password");
if (!dir) {
if (errno == ENOENT)
return false;
else
return -errno;
}
FOREACH_DIRENT_ALL(de, dir, return -errno) {
if (startswith(de->d_name, "ask."))
return true;
}
return false;
}
static int manager_dispatch_ask_password_fd(sd_event_source *source,
int fd, uint32_t revents, void *userdata) {
Manager *m = userdata;
assert(m);
(void) flush_fd(fd);
m->have_ask_password = have_ask_password();
if (m->have_ask_password < 0)
/* Log error but continue. Negative have_ask_password
* is treated as unknown status. */
log_error_errno(m->have_ask_password, "Failed to list /run/systemd/ask-password: %m");
return 0;
}
static void manager_close_ask_password(Manager *m) {
assert(m);
m->ask_password_event_source = sd_event_source_unref(m->ask_password_event_source);
m->ask_password_inotify_fd = safe_close(m->ask_password_inotify_fd);
m->have_ask_password = -EINVAL;
}
static int manager_check_ask_password(Manager *m) {
int r;
assert(m);
if (!m->ask_password_event_source) {
assert(m->ask_password_inotify_fd < 0);
mkdir_p_label("/run/systemd/ask-password", 0755);
m->ask_password_inotify_fd = inotify_init1(IN_NONBLOCK|IN_CLOEXEC);
if (m->ask_password_inotify_fd < 0)
return log_error_errno(errno, "inotify_init1() failed: %m");
if (inotify_add_watch(m->ask_password_inotify_fd, "/run/systemd/ask-password", IN_CREATE|IN_DELETE|IN_MOVE) < 0) {
log_error_errno(errno, "Failed to add watch on /run/systemd/ask-password: %m");
manager_close_ask_password(m);
return -errno;
}
r = sd_event_add_io(m->event, &m->ask_password_event_source,
m->ask_password_inotify_fd, EPOLLIN,
manager_dispatch_ask_password_fd, m);
if (r < 0) {
log_error_errno(errno, "Failed to add event source for /run/systemd/ask-password: %m");
manager_close_ask_password(m);
return -errno;
}
(void) sd_event_source_set_description(m->ask_password_event_source, "manager-ask-password");
/* Queries might have been added meanwhile... */
manager_dispatch_ask_password_fd(m->ask_password_event_source,
m->ask_password_inotify_fd, EPOLLIN, m);
}
return m->have_ask_password;
}
static int manager_watch_idle_pipe(Manager *m) {
int r;
assert(m);
if (m->idle_pipe_event_source)
return 0;
if (m->idle_pipe[2] < 0)
return 0;
r = sd_event_add_io(m->event, &m->idle_pipe_event_source, m->idle_pipe[2], EPOLLIN, manager_dispatch_idle_pipe_fd, m);
if (r < 0)
return log_error_errno(r, "Failed to watch idle pipe: %m");
(void) sd_event_source_set_description(m->idle_pipe_event_source, "manager-idle-pipe");
return 0;
}
static void manager_close_idle_pipe(Manager *m) {
assert(m);
m->idle_pipe_event_source = sd_event_source_unref(m->idle_pipe_event_source);
safe_close_pair(m->idle_pipe);
safe_close_pair(m->idle_pipe + 2);
}
static int manager_setup_time_change(Manager *m) {
int r;
/* We only care for the cancellation event, hence we set the
* timeout to the latest possible value. */
struct itimerspec its = {
.it_value.tv_sec = TIME_T_MAX,
};
assert(m);
assert_cc(sizeof(time_t) == sizeof(TIME_T_MAX));
if (m->test_run_flags)
return 0;
/* Uses TFD_TIMER_CANCEL_ON_SET to get notifications whenever
* CLOCK_REALTIME makes a jump relative to CLOCK_MONOTONIC */
m->time_change_fd = timerfd_create(CLOCK_REALTIME, TFD_NONBLOCK|TFD_CLOEXEC);
if (m->time_change_fd < 0)
return log_error_errno(errno, "Failed to create timerfd: %m");
if (timerfd_settime(m->time_change_fd, TFD_TIMER_ABSTIME|TFD_TIMER_CANCEL_ON_SET, &its, NULL) < 0) {
log_debug_errno(errno, "Failed to set up TFD_TIMER_CANCEL_ON_SET, ignoring: %m");
m->time_change_fd = safe_close(m->time_change_fd);
return 0;
}
r = sd_event_add_io(m->event, &m->time_change_event_source, m->time_change_fd, EPOLLIN, manager_dispatch_time_change_fd, m);
if (r < 0)
return log_error_errno(r, "Failed to create time change event source: %m");
(void) sd_event_source_set_description(m->time_change_event_source, "manager-time-change");
log_debug("Set up TFD_TIMER_CANCEL_ON_SET timerfd.");
return 0;
}
static int enable_special_signals(Manager *m) {
_cleanup_close_ int fd = -1;
assert(m);
if (m->test_run_flags)
return 0;
/* Enable that we get SIGINT on control-alt-del. In containers
* this will fail with EPERM (older) or EINVAL (newer), so
* ignore that. */
if (reboot(RB_DISABLE_CAD) < 0 && !IN_SET(errno, EPERM, EINVAL))
log_warning_errno(errno, "Failed to enable ctrl-alt-del handling: %m");
fd = open_terminal("/dev/tty0", O_RDWR|O_NOCTTY|O_CLOEXEC);
if (fd < 0) {
/* Support systems without virtual console */
if (fd != -ENOENT)
log_warning_errno(errno, "Failed to open /dev/tty0: %m");
} else {
/* Enable that we get SIGWINCH on kbrequest */
if (ioctl(fd, KDSIGACCEPT, SIGWINCH) < 0)
log_warning_errno(errno, "Failed to enable kbrequest handling: %m");
}
return 0;
}
#define RTSIG_IF_AVAILABLE(signum) (signum <= SIGRTMAX ? signum : -1)
static int manager_setup_signals(Manager *m) {
struct sigaction sa = {
.sa_handler = SIG_DFL,
.sa_flags = SA_NOCLDSTOP|SA_RESTART,
};
sigset_t mask;
int r;
assert(m);
assert_se(sigaction(SIGCHLD, &sa, NULL) == 0);
/* We make liberal use of realtime signals here. On
* Linux/glibc we have 30 of them (with the exception of Linux
* on hppa, see below), between SIGRTMIN+0 ... SIGRTMIN+30
* (aka SIGRTMAX). */
assert_se(sigemptyset(&mask) == 0);
sigset_add_many(&mask,
SIGCHLD, /* Child died */
SIGTERM, /* Reexecute daemon */
SIGHUP, /* Reload configuration */
SIGUSR1, /* systemd/upstart: reconnect to D-Bus */
SIGUSR2, /* systemd: dump status */
SIGINT, /* Kernel sends us this on control-alt-del */
SIGWINCH, /* Kernel sends us this on kbrequest (alt-arrowup) */
SIGPWR, /* Some kernel drivers and upsd send us this on power failure */
SIGRTMIN+0, /* systemd: start default.target */
SIGRTMIN+1, /* systemd: isolate rescue.target */
SIGRTMIN+2, /* systemd: isolate emergency.target */
SIGRTMIN+3, /* systemd: start halt.target */
SIGRTMIN+4, /* systemd: start poweroff.target */
SIGRTMIN+5, /* systemd: start reboot.target */
SIGRTMIN+6, /* systemd: start kexec.target */
/* ... space for more special targets ... */
SIGRTMIN+13, /* systemd: Immediate halt */
SIGRTMIN+14, /* systemd: Immediate poweroff */
SIGRTMIN+15, /* systemd: Immediate reboot */
SIGRTMIN+16, /* systemd: Immediate kexec */
/* ... space for more immediate system state changes ... */
SIGRTMIN+20, /* systemd: enable status messages */
SIGRTMIN+21, /* systemd: disable status messages */
SIGRTMIN+22, /* systemd: set log level to LOG_DEBUG */
SIGRTMIN+23, /* systemd: set log level to LOG_INFO */
SIGRTMIN+24, /* systemd: Immediate exit (--user only) */
/* .. one free signal here ... */
/* Apparently Linux on hppa had fewer RT signals until v3.18,
* SIGRTMAX was SIGRTMIN+25, and then SIGRTMIN was lowered,
* see commit v3.17-7614-g1f25df2eff.
*
* We cannot unconditionally make use of those signals here,
* so let's use a runtime check. Since these commands are
* accessible by different means and only really a safety
* net, the missing functionality on hppa shouldn't matter.
*/
RTSIG_IF_AVAILABLE(SIGRTMIN+26), /* systemd: set log target to journal-or-kmsg */
RTSIG_IF_AVAILABLE(SIGRTMIN+27), /* systemd: set log target to console */
RTSIG_IF_AVAILABLE(SIGRTMIN+28), /* systemd: set log target to kmsg */
RTSIG_IF_AVAILABLE(SIGRTMIN+29), /* systemd: set log target to syslog-or-kmsg (obsolete) */
/* ... one free signal here SIGRTMIN+30 ... */
-1);
assert_se(sigprocmask(SIG_SETMASK, &mask, NULL) == 0);
m->signal_fd = signalfd(-1, &mask, SFD_NONBLOCK|SFD_CLOEXEC);
if (m->signal_fd < 0)
return -errno;
r = sd_event_add_io(m->event, &m->signal_event_source, m->signal_fd, EPOLLIN, manager_dispatch_signal_fd, m);
if (r < 0)
return r;
(void) sd_event_source_set_description(m->signal_event_source, "manager-signal");
/* Process signals a bit earlier than the rest of things, but later than notify_fd processing, so that the
* notify processing can still figure out to which process/service a message belongs, before we reap the
* process. Also, process this before handling cgroup notifications, so that we always collect child exit
* status information before detecting that there's no process in a cgroup. */
r = sd_event_source_set_priority(m->signal_event_source, SD_EVENT_PRIORITY_NORMAL-6);
if (r < 0)
return r;
if (MANAGER_IS_SYSTEM(m))
return enable_special_signals(m);
return 0;
}
static void manager_sanitize_environment(Manager *m) {
assert(m);
/* Let's remove some environment variables that we need ourselves to communicate with our clients */
strv_env_unset_many(
m->environment,
"EXIT_CODE",
"EXIT_STATUS",
"INVOCATION_ID",
"JOURNAL_STREAM",
"LISTEN_FDNAMES",
"LISTEN_FDS",
"LISTEN_PID",
"MAINPID",
"MANAGERPID",
"NOTIFY_SOCKET",
"REMOTE_ADDR",
"REMOTE_PORT",
"SERVICE_RESULT",
"WATCHDOG_PID",
"WATCHDOG_USEC",
NULL);
/* Let's order the environment alphabetically, just to make it pretty */
strv_sort(m->environment);
}
static int manager_default_environment(Manager *m) {
assert(m);
if (MANAGER_IS_SYSTEM(m)) {
/* The system manager always starts with a clean
* environment for its children. It does not import
* the kernel's or the parents' exported variables.
*
* The initial passed environment is untouched to keep
* /proc/self/environ valid; it is used for tagging
* the init process inside containers. */
m->environment = strv_new("PATH=" DEFAULT_PATH,
NULL);
/* Import locale variables LC_*= from configuration */
locale_setup(&m->environment);
} else
/* The user manager passes its own environment
* along to its children. */
m->environment = strv_copy(environ);
if (!m->environment)
return -ENOMEM;
manager_sanitize_environment(m);
return 0;
}
static int manager_setup_prefix(Manager *m) {
struct table_entry {
uint64_t type;
const char *suffix;
};
static const struct table_entry paths_system[_EXEC_DIRECTORY_TYPE_MAX] = {
[EXEC_DIRECTORY_RUNTIME] = { SD_PATH_SYSTEM_RUNTIME, NULL },
[EXEC_DIRECTORY_STATE] = { SD_PATH_SYSTEM_STATE_PRIVATE, NULL },
[EXEC_DIRECTORY_CACHE] = { SD_PATH_SYSTEM_STATE_CACHE, NULL },
[EXEC_DIRECTORY_LOGS] = { SD_PATH_SYSTEM_STATE_LOGS, NULL },
[EXEC_DIRECTORY_CONFIGURATION] = { SD_PATH_SYSTEM_CONFIGURATION, NULL },
};
static const struct table_entry paths_user[_EXEC_DIRECTORY_TYPE_MAX] = {
[EXEC_DIRECTORY_RUNTIME] = { SD_PATH_USER_RUNTIME, NULL },
[EXEC_DIRECTORY_STATE] = { SD_PATH_USER_CONFIGURATION, NULL },
[EXEC_DIRECTORY_CACHE] = { SD_PATH_USER_STATE_CACHE, NULL },
[EXEC_DIRECTORY_LOGS] = { SD_PATH_USER_CONFIGURATION, "log" },
[EXEC_DIRECTORY_CONFIGURATION] = { SD_PATH_USER_CONFIGURATION, NULL },
};
const struct table_entry *p;
ExecDirectoryType i;
int r;
assert(m);
if (MANAGER_IS_SYSTEM(m))
p = paths_system;
else
p = paths_user;
for (i = 0; i < _EXEC_DIRECTORY_TYPE_MAX; i++) {
r = sd_path_home(p[i].type, p[i].suffix, &m->prefix[i]);
if (r < 0)
return r;
}
return 0;
}
static int manager_setup_run_queue(Manager *m) {
int r;
assert(m);
assert(!m->run_queue_event_source);
r = sd_event_add_defer(m->event, &m->run_queue_event_source, manager_dispatch_run_queue, m);
if (r < 0)
return r;
r = sd_event_source_set_priority(m->run_queue_event_source, SD_EVENT_PRIORITY_IDLE);
if (r < 0)
return r;
r = sd_event_source_set_enabled(m->run_queue_event_source, SD_EVENT_OFF);
if (r < 0)
return r;
(void) sd_event_source_set_description(m->run_queue_event_source, "manager-run-queue");
return 0;
}
static int manager_setup_sigchld_event_source(Manager *m) {
int r;
assert(m);
assert(!m->sigchld_event_source);
r = sd_event_add_defer(m->event, &m->sigchld_event_source, manager_dispatch_sigchld, m);
if (r < 0)
return r;
r = sd_event_source_set_priority(m->sigchld_event_source, SD_EVENT_PRIORITY_NORMAL-7);
if (r < 0)
return r;
r = sd_event_source_set_enabled(m->sigchld_event_source, SD_EVENT_OFF);
if (r < 0)
return r;
(void) sd_event_source_set_description(m->sigchld_event_source, "manager-sigchld");
return 0;
}
int manager_new(UnitFileScope scope, unsigned test_run_flags, Manager **_m) {
_cleanup_(manager_freep) Manager *m = NULL;
int r;
assert(_m);
assert(IN_SET(scope, UNIT_FILE_SYSTEM, UNIT_FILE_USER));
m = new0(Manager, 1);
if (!m)
return -ENOMEM;
m->unit_file_scope = scope;
m->exit_code = _MANAGER_EXIT_CODE_INVALID;
m->default_timer_accuracy_usec = USEC_PER_MINUTE;
m->default_memory_accounting = MEMORY_ACCOUNTING_DEFAULT;
m->default_tasks_accounting = true;
m->default_tasks_max = UINT64_MAX;
m->default_timeout_start_usec = DEFAULT_TIMEOUT_USEC;
m->default_timeout_stop_usec = DEFAULT_TIMEOUT_USEC;
m->default_restart_usec = DEFAULT_RESTART_USEC;
#if ENABLE_EFI
if (MANAGER_IS_SYSTEM(m) && detect_container() <= 0)
boot_timestamps(m->timestamps + MANAGER_TIMESTAMP_USERSPACE,
m->timestamps + MANAGER_TIMESTAMP_FIRMWARE,
m->timestamps + MANAGER_TIMESTAMP_LOADER);
#endif
/* Prepare log fields we can use for structured logging */
if (MANAGER_IS_SYSTEM(m)) {
m->unit_log_field = "UNIT=";
m->unit_log_format_string = "UNIT=%s";
m->invocation_log_field = "INVOCATION_ID=";
m->invocation_log_format_string = "INVOCATION_ID=%s";
} else {
m->unit_log_field = "USER_UNIT=";
m->unit_log_format_string = "USER_UNIT=%s";
m->invocation_log_field = "USER_INVOCATION_ID=";
m->invocation_log_format_string = "USER_INVOCATION_ID=%s";
}
m->idle_pipe[0] = m->idle_pipe[1] = m->idle_pipe[2] = m->idle_pipe[3] = -1;
m->pin_cgroupfs_fd = m->notify_fd = m->cgroups_agent_fd = m->signal_fd = m->time_change_fd =
m->dev_autofs_fd = m->private_listen_fd = m->cgroup_inotify_fd =
m->ask_password_inotify_fd = -1;
m->user_lookup_fds[0] = m->user_lookup_fds[1] = -1;
m->current_job_id = 1; /* start as id #1, so that we can leave #0 around as "null-like" value */
m->have_ask_password = -EINVAL; /* we don't know */
m->first_boot = -1;
m->test_run_flags = test_run_flags;
/* Reboot immediately if the user hits C-A-D more often than 7x per 2s */
RATELIMIT_INIT(m->ctrl_alt_del_ratelimit, 2 * USEC_PER_SEC, 7);
r = manager_default_environment(m);
if (r < 0)
return r;
r = hashmap_ensure_allocated(&m->units, &string_hash_ops);
if (r < 0)
return r;
r = hashmap_ensure_allocated(&m->jobs, NULL);
if (r < 0)
return r;
r = hashmap_ensure_allocated(&m->cgroup_unit, &path_hash_ops);
if (r < 0)
return r;
r = hashmap_ensure_allocated(&m->watch_bus, &string_hash_ops);
if (r < 0)
return r;
r = manager_setup_prefix(m);
if (r < 0)
return r;
m->udev = udev_new();
if (!m->udev)
return -ENOMEM;
r = sd_event_default(&m->event);
if (r < 0)
return r;
r = manager_setup_run_queue(m);
if (r < 0)
return r;
if (test_run_flags == MANAGER_TEST_RUN_MINIMAL) {
m->cgroup_root = strdup("");
if (!m->cgroup_root)
return -ENOMEM;
} else {
r = manager_setup_signals(m);
if (r < 0)
return r;
r = manager_setup_cgroup(m);
if (r < 0)
return r;
r = manager_setup_time_change(m);
if (r < 0)
return r;
r = manager_setup_sigchld_event_source(m);
if (r < 0)
return r;
}
if (MANAGER_IS_SYSTEM(m) && test_run_flags == 0) {
r = mkdir_label("/run/systemd/units", 0755);
if (r < 0 && r != -EEXIST)
return r;
}
m->taint_usr =
!in_initrd() &&
dir_is_empty("/usr") > 0;
/* Note that we do not set up the notify fd here. We do that after deserialization,
* since they might have gotten serialized across the reexec. */
*_m = TAKE_PTR(m);
return 0;
}
static int manager_setup_notify(Manager *m) {
int r;
if (m->test_run_flags)
return 0;
if (m->notify_fd < 0) {
_cleanup_close_ int fd = -1;
union sockaddr_union sa = {
.sa.sa_family = AF_UNIX,
};
static const int one = 1;
/* First free all secondary fields */
m->notify_socket = mfree(m->notify_socket);
m->notify_event_source = sd_event_source_unref(m->notify_event_source);
fd = socket(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
if (fd < 0)
return log_error_errno(errno, "Failed to allocate notification socket: %m");
fd_inc_rcvbuf(fd, NOTIFY_RCVBUF_SIZE);
m->notify_socket = strappend(m->prefix[EXEC_DIRECTORY_RUNTIME], "/systemd/notify");
if (!m->notify_socket)
return log_oom();
(void) mkdir_parents_label(m->notify_socket, 0755);
(void) unlink(m->notify_socket);
strncpy(sa.un.sun_path, m->notify_socket, sizeof(sa.un.sun_path)-1);
r = bind(fd, &sa.sa, SOCKADDR_UN_LEN(sa.un));
if (r < 0)
return log_error_errno(errno, "bind(%s) failed: %m", sa.un.sun_path);
r = setsockopt(fd, SOL_SOCKET, SO_PASSCRED, &one, sizeof(one));
if (r < 0)
return log_error_errno(errno, "SO_PASSCRED failed: %m");
m->notify_fd = TAKE_FD(fd);
log_debug("Using notification socket %s", m->notify_socket);
}
if (!m->notify_event_source) {
r = sd_event_add_io(m->event, &m->notify_event_source, m->notify_fd, EPOLLIN, manager_dispatch_notify_fd, m);
if (r < 0)
return log_error_errno(r, "Failed to allocate notify event source: %m");
/* Process notification messages a bit earlier than SIGCHLD, so that we can still identify to which
* service an exit message belongs. */
r = sd_event_source_set_priority(m->notify_event_source, SD_EVENT_PRIORITY_NORMAL-8);
if (r < 0)
return log_error_errno(r, "Failed to set priority of notify event source: %m");
(void) sd_event_source_set_description(m->notify_event_source, "manager-notify");
}
return 0;
}
static int manager_setup_cgroups_agent(Manager *m) {
static const union sockaddr_union sa = {
.un.sun_family = AF_UNIX,
.un.sun_path = "/run/systemd/cgroups-agent",
};
int r;
/* This creates a listening socket we receive cgroups agent messages on. We do not use D-Bus for delivering
* these messages from the cgroups agent binary to PID 1, as the cgroups agent binary is very short-living, and
* each instance of it needs a new D-Bus connection. Since D-Bus connections are SOCK_STREAM/AF_UNIX, on
* overloaded systems the backlog of the D-Bus socket becomes relevant, as not more than the configured number
* of D-Bus connections may be queued until the kernel will start dropping further incoming connections,
* possibly resulting in lost cgroups agent messages. To avoid this, we'll use a private SOCK_DGRAM/AF_UNIX
* socket, where no backlog is relevant as communication may take place without an actual connect() cycle, and
* we thus won't lose messages.
*
* Note that PID 1 will forward the agent message to system bus, so that the user systemd instance may listen
* to it. The system instance hence listens on this special socket, but the user instances listen on the system
* bus for these messages. */
if (m->test_run_flags)
return 0;
if (!MANAGER_IS_SYSTEM(m))
return 0;
r = cg_unified_controller(SYSTEMD_CGROUP_CONTROLLER);
if (r < 0)
return log_error_errno(r, "Failed to determine whether unified cgroups hierarchy is used: %m");
if (r > 0) /* We don't need this anymore on the unified hierarchy */
return 0;
if (m->cgroups_agent_fd < 0) {
_cleanup_close_ int fd = -1;
/* First free all secondary fields */
m->cgroups_agent_event_source = sd_event_source_unref(m->cgroups_agent_event_source);
fd = socket(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
if (fd < 0)
return log_error_errno(errno, "Failed to allocate cgroups agent socket: %m");
fd_inc_rcvbuf(fd, CGROUPS_AGENT_RCVBUF_SIZE);
(void) unlink(sa.un.sun_path);
/* Only allow root to connect to this socket */
RUN_WITH_UMASK(0077)
r = bind(fd, &sa.sa, SOCKADDR_UN_LEN(sa.un));
if (r < 0)
return log_error_errno(errno, "bind(%s) failed: %m", sa.un.sun_path);
m->cgroups_agent_fd = fd;
fd = -1;
}
if (!m->cgroups_agent_event_source) {
r = sd_event_add_io(m->event, &m->cgroups_agent_event_source, m->cgroups_agent_fd, EPOLLIN, manager_dispatch_cgroups_agent_fd, m);
if (r < 0)
return log_error_errno(r, "Failed to allocate cgroups agent event source: %m");
/* Process cgroups notifications early, but after having processed service notification messages or
* SIGCHLD signals, so that a cgroup running empty is always just the last safety net of notification,
* and we collected the metadata the notification and SIGCHLD stuff offers first. Also see handling of
* cgroup inotify for the unified cgroup stuff. */
r = sd_event_source_set_priority(m->cgroups_agent_event_source, SD_EVENT_PRIORITY_NORMAL-4);
if (r < 0)
return log_error_errno(r, "Failed to set priority of cgroups agent event source: %m");
(void) sd_event_source_set_description(m->cgroups_agent_event_source, "manager-cgroups-agent");
}
return 0;
}
static int manager_setup_user_lookup_fd(Manager *m) {
int r;
assert(m);
/* Set up the socket pair used for passing UID/GID resolution results from forked off processes to PID
* 1. Background: we can't do name lookups (NSS) from PID 1, since it might involve IPC and thus activation,
* and we might hence deadlock on ourselves. Hence we do all user/group lookups asynchronously from the forked
* off processes right before executing the binaries to start. In order to be able to clean up any IPC objects
* created by a unit (see RemoveIPC=) we need to know in PID 1 the used UID/GID of the executed processes,
* hence we establish this communication channel so that forked off processes can pass their UID/GID
* information back to PID 1. The forked off processes send their resolved UID/GID to PID 1 in a simple
* datagram, along with their unit name, so that we can share one communication socket pair among all units for
* this purpose.
*
* You might wonder why we need a communication channel for this that is independent of the usual notification
* socket scheme (i.e. $NOTIFY_SOCKET). The primary difference is about trust: data sent via the $NOTIFY_SOCKET
* channel is only accepted if it originates from the right unit and if reception was enabled for it. The user
* lookup socket OTOH is only accessible by PID 1 and its children until they exec(), and always available.
*
* Note that this function is called under two circumstances: when we first initialize (in which case we
* allocate both the socket pair and the event source to listen on it), and when we deserialize after a reload
* (in which case the socket pair already exists but we still need to allocate the event source for it). */
if (m->user_lookup_fds[0] < 0) {
/* Free all secondary fields */
safe_close_pair(m->user_lookup_fds);
m->user_lookup_event_source = sd_event_source_unref(m->user_lookup_event_source);
if (socketpair(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC, 0, m->user_lookup_fds) < 0)
return log_error_errno(errno, "Failed to allocate user lookup socket: %m");
(void) fd_inc_rcvbuf(m->user_lookup_fds[0], NOTIFY_RCVBUF_SIZE);
}
if (!m->user_lookup_event_source) {
r = sd_event_add_io(m->event, &m->user_lookup_event_source, m->user_lookup_fds[0], EPOLLIN, manager_dispatch_user_lookup_fd, m);
if (r < 0)
return log_error_errno(errno, "Failed to allocate user lookup event source: %m");
/* Process even earlier than the notify event source, so that we always know first about valid UID/GID
* resolutions */
r = sd_event_source_set_priority(m->user_lookup_event_source, SD_EVENT_PRIORITY_NORMAL-11);
if (r < 0)
return log_error_errno(errno, "Failed to set priority ot user lookup event source: %m");
(void) sd_event_source_set_description(m->user_lookup_event_source, "user-lookup");
}
return 0;
}
static unsigned manager_dispatch_cleanup_queue(Manager *m) {
Unit *u;
unsigned n = 0;
assert(m);
while ((u = m->cleanup_queue)) {
assert(u->in_cleanup_queue);
unit_free(u);
n++;
}
return n;
}
enum {
GC_OFFSET_IN_PATH, /* This one is on the path we were traveling */
GC_OFFSET_UNSURE, /* No clue */
GC_OFFSET_GOOD, /* We still need this unit */
GC_OFFSET_BAD, /* We don't need this unit anymore */
_GC_OFFSET_MAX
};
static void unit_gc_mark_good(Unit *u, unsigned gc_marker) {
Unit *other;
Iterator i;
void *v;
u->gc_marker = gc_marker + GC_OFFSET_GOOD;
/* Recursively mark referenced units as GOOD as well */
HASHMAP_FOREACH_KEY(v, other, u->dependencies[UNIT_REFERENCES], i)
if (other->gc_marker == gc_marker + GC_OFFSET_UNSURE)
unit_gc_mark_good(other, gc_marker);
}
static void unit_gc_sweep(Unit *u, unsigned gc_marker) {
Unit *other;
bool is_bad;
Iterator i;
void *v;
assert(u);
if (IN_SET(u->gc_marker - gc_marker,
GC_OFFSET_GOOD, GC_OFFSET_BAD, GC_OFFSET_UNSURE, GC_OFFSET_IN_PATH))
return;
if (u->in_cleanup_queue)
goto bad;
if (!unit_may_gc(u))
goto good;
u->gc_marker = gc_marker + GC_OFFSET_IN_PATH;
is_bad = true;
HASHMAP_FOREACH_KEY(v, other, u->dependencies[UNIT_REFERENCED_BY], i) {
unit_gc_sweep(other, gc_marker);
if (other->gc_marker == gc_marker + GC_OFFSET_GOOD)
goto good;
if (other->gc_marker != gc_marker + GC_OFFSET_BAD)
is_bad = false;
}
if (u->refs_by_target) {
const UnitRef *ref;
LIST_FOREACH(refs_by_target, ref, u->refs_by_target) {
unit_gc_sweep(ref->source, gc_marker);
if (ref->source->gc_marker == gc_marker + GC_OFFSET_GOOD)
goto good;
if (ref->source->gc_marker != gc_marker + GC_OFFSET_BAD)
is_bad = false;
}
}
if (is_bad)
goto bad;
/* We were unable to find anything out about this entry, so
* let's investigate it later */
u->gc_marker = gc_marker + GC_OFFSET_UNSURE;
unit_add_to_gc_queue(u);
return;
bad:
/* We definitely know that this one is not useful anymore, so
* let's mark it for deletion */
u->gc_marker = gc_marker + GC_OFFSET_BAD;
unit_add_to_cleanup_queue(u);
return;
good:
unit_gc_mark_good(u, gc_marker);
}
static unsigned manager_dispatch_gc_unit_queue(Manager *m) {
unsigned n = 0, gc_marker;
Unit *u;
assert(m);
/* log_debug("Running GC..."); */
m->gc_marker += _GC_OFFSET_MAX;
if (m->gc_marker + _GC_OFFSET_MAX <= _GC_OFFSET_MAX)
m->gc_marker = 1;
gc_marker = m->gc_marker;
while ((u = m->gc_unit_queue)) {
assert(u->in_gc_queue);
unit_gc_sweep(u, gc_marker);
LIST_REMOVE(gc_queue, m->gc_unit_queue, u);
u->in_gc_queue = false;
n++;
if (IN_SET(u->gc_marker - gc_marker,
GC_OFFSET_BAD, GC_OFFSET_UNSURE)) {
if (u->id)
log_unit_debug(u, "Collecting.");
u->gc_marker = gc_marker + GC_OFFSET_BAD;
unit_add_to_cleanup_queue(u);
}
}
return n;
}
static unsigned manager_dispatch_gc_job_queue(Manager *m) {
unsigned n = 0;
Job *j;
assert(m);
while ((j = m->gc_job_queue)) {
assert(j->in_gc_queue);
LIST_REMOVE(gc_queue, m->gc_job_queue, j);
j->in_gc_queue = false;
n++;
if (!job_may_gc(j))
continue;
log_unit_debug(j->unit, "Collecting job.");
(void) job_finish_and_invalidate(j, JOB_COLLECTED, false, false);
}
return n;
}
static void manager_clear_jobs_and_units(Manager *m) {
Unit *u;
assert(m);
while ((u = hashmap_first(m->units)))
unit_free(u);
manager_dispatch_cleanup_queue(m);
assert(!m->load_queue);
assert(!m->run_queue);
assert(!m->dbus_unit_queue);
assert(!m->dbus_job_queue);
assert(!m->cleanup_queue);
assert(!m->gc_unit_queue);
assert(!m->gc_job_queue);
assert(hashmap_isempty(m->jobs));
assert(hashmap_isempty(m->units));
m->n_on_console = 0;
m->n_running_jobs = 0;
}
Manager* manager_free(Manager *m) {
UnitType c;
int i;
ExecDirectoryType dt;
if (!m)
return NULL;
manager_clear_jobs_and_units(m);
for (c = 0; c < _UNIT_TYPE_MAX; c++)
if (unit_vtable[c]->shutdown)
unit_vtable[c]->shutdown(m);
/* If we reexecute ourselves, we keep the root cgroup around */
manager_shutdown_cgroup(m, m->exit_code != MANAGER_REEXECUTE);
lookup_paths_flush_generator(&m->lookup_paths);
bus_done(m);
exec_runtime_vacuum(m);
hashmap_free(m->exec_runtime_by_id);
dynamic_user_vacuum(m, false);
hashmap_free(m->dynamic_users);
hashmap_free(m->units);
hashmap_free(m->units_by_invocation_id);
hashmap_free(m->jobs);
hashmap_free(m->watch_pids);
hashmap_free(m->watch_bus);
set_free(m->startup_units);
set_free(m->failed_units);
sd_event_source_unref(m->signal_event_source);
sd_event_source_unref(m->sigchld_event_source);
sd_event_source_unref(m->notify_event_source);
sd_event_source_unref(m->cgroups_agent_event_source);
sd_event_source_unref(m->time_change_event_source);
sd_event_source_unref(m->jobs_in_progress_event_source);
sd_event_source_unref(m->run_queue_event_source);
sd_event_source_unref(m->user_lookup_event_source);
sd_event_source_unref(m->sync_bus_names_event_source);
safe_close(m->signal_fd);
safe_close(m->notify_fd);
safe_close(m->cgroups_agent_fd);
safe_close(m->time_change_fd);
safe_close_pair(m->user_lookup_fds);
manager_close_ask_password(m);
manager_close_idle_pipe(m);
udev_unref(m->udev);
sd_event_unref(m->event);
free(m->notify_socket);
lookup_paths_free(&m->lookup_paths);
strv_free(m->environment);
hashmap_free(m->cgroup_unit);
set_free_free(m->unit_path_cache);
free(m->switch_root);
free(m->switch_root_init);
for (i = 0; i < _RLIMIT_MAX; i++)
m->rlimit[i] = mfree(m->rlimit[i]);
assert(hashmap_isempty(m->units_requiring_mounts_for));
hashmap_free(m->units_requiring_mounts_for);
hashmap_free(m->uid_refs);
hashmap_free(m->gid_refs);
for (dt = 0; dt < _EXEC_DIRECTORY_TYPE_MAX; dt++)
m->prefix[dt] = mfree(m->prefix[dt]);
return mfree(m);
}
void manager_enumerate(Manager *m) {
UnitType c;
assert(m);
/* Let's ask every type to load all units from disk/kernel
* that it might know */
for (c = 0; c < _UNIT_TYPE_MAX; c++) {
if (!unit_type_supported(c)) {
log_debug("Unit type .%s is not supported on this system.", unit_type_to_string(c));
continue;
}
if (!unit_vtable[c]->enumerate)
continue;
unit_vtable[c]->enumerate(m);
}
manager_dispatch_load_queue(m);
}
static void manager_coldplug(Manager *m) {
Iterator i;
Unit *u;
char *k;
int r;
assert(m);
/* Then, let's set up their initial state. */
HASHMAP_FOREACH_KEY(u, k, m->units, i) {
/* ignore aliases */
if (u->id != k)
continue;
r = unit_coldplug(u);
if (r < 0)
log_warning_errno(r, "We couldn't coldplug %s, proceeding anyway: %m", u->id);
}
}
static void manager_build_unit_path_cache(Manager *m) {
char **i;
int r;
assert(m);
set_free_free(m->unit_path_cache);
m->unit_path_cache = set_new(&path_hash_ops);
if (!m->unit_path_cache) {
r = -ENOMEM;
goto fail;
}
/* This simply builds a list of files we know exist, so that
* we don't always have to go to disk */
STRV_FOREACH(i, m->lookup_paths.search_path) {
_cleanup_closedir_ DIR *d = NULL;
struct dirent *de;
d = opendir(*i);
if (!d) {
if (errno != ENOENT)
log_warning_errno(errno, "Failed to open directory %s, ignoring: %m", *i);
continue;
}
FOREACH_DIRENT(de, d, r = -errno; goto fail) {
char *p;
p = strjoin(streq(*i, "/") ? "" : *i, "/", de->d_name);
if (!p) {
r = -ENOMEM;
goto fail;
}
r = set_consume(m->unit_path_cache, p);
if (r < 0)
goto fail;
}
}
return;
fail:
log_warning_errno(r, "Failed to build unit path cache, proceeding without: %m");
m->unit_path_cache = set_free_free(m->unit_path_cache);
}
static void manager_distribute_fds(Manager *m, FDSet *fds) {
Iterator i;
Unit *u;
assert(m);
HASHMAP_FOREACH(u, m->units, i) {
if (fdset_size(fds) <= 0)
break;
if (!UNIT_VTABLE(u)->distribute_fds)
continue;
UNIT_VTABLE(u)->distribute_fds(u, fds);
}
}
static bool manager_dbus_is_running(Manager *m, bool deserialized) {
Unit *u;
assert(m);
/* This checks whether the dbus instance we are supposed to expose our APIs on is up. We check both the socket
* and the service unit. If the 'deserialized' parameter is true we'll check the deserialized state of the unit
* rather than the current one. */
if (m->test_run_flags != 0)
return false;
/* If we are in the user instance, and the env var is already set for us, then this means D-Bus is ran
* somewhere outside of our own logic. Let's use it */
if (MANAGER_IS_USER(m) && getenv("DBUS_SESSION_BUS_ADDRESS"))
return true;
u = manager_get_unit(m, SPECIAL_DBUS_SOCKET);
if (!u)
return false;
if ((deserialized ? SOCKET(u)->deserialized_state : SOCKET(u)->state) != SOCKET_RUNNING)
return false;
u = manager_get_unit(m, SPECIAL_DBUS_SERVICE);
if (!u)
return false;
if (!IN_SET((deserialized ? SERVICE(u)->deserialized_state : SERVICE(u)->state), SERVICE_RUNNING, SERVICE_RELOAD))
return false;
return true;
}
int manager_startup(Manager *m, FILE *serialization, FDSet *fds) {
int r;
assert(m);
/* If we are running in test mode, we still want to run the generators,
* but we should not touch the real generator directories. */
r = lookup_paths_init(&m->lookup_paths, m->unit_file_scope,
m->test_run_flags ? LOOKUP_PATHS_TEMPORARY_GENERATED : 0,
NULL);
if (r < 0)
return r;
r = manager_run_environment_generators(m);
if (r < 0)
return r;
dual_timestamp_get(m->timestamps + MANAGER_TIMESTAMP_GENERATORS_START);
r = manager_run_generators(m);
dual_timestamp_get(m->timestamps + MANAGER_TIMESTAMP_GENERATORS_FINISH);
if (r < 0)
return r;
/* If this is the first boot, and we are in the host system, then preset everything */
if (m->first_boot > 0 &&
MANAGER_IS_SYSTEM(m) &&
!m->test_run_flags) {
r = unit_file_preset_all(UNIT_FILE_SYSTEM, 0, NULL, UNIT_FILE_PRESET_ENABLE_ONLY, NULL, 0);
if (r < 0)
log_full_errno(r == -EEXIST ? LOG_NOTICE : LOG_WARNING, r,
"Failed to populate /etc with preset unit settings, ignoring: %m");
else
log_info("Populated /etc with preset unit settings.");
}
lookup_paths_reduce(&m->lookup_paths);
manager_build_unit_path_cache(m);
/* If we will deserialize make sure that during enumeration
* this is already known, so we increase the counter here
* already */
if (serialization)
m->n_reloading++;
/* First, enumerate what we can from all config files */
dual_timestamp_get(m->timestamps + MANAGER_TIMESTAMP_UNITS_LOAD_START);
manager_enumerate(m);
dual_timestamp_get(m->timestamps + MANAGER_TIMESTAMP_UNITS_LOAD_FINISH);
/* Second, deserialize if there is something to deserialize */
if (serialization) {
r = manager_deserialize(m, serialization, fds);
if (r < 0)
return log_error_errno(r, "Deserialization failed: %m");
}
/* Any fds left? Find some unit which wants them. This is
* useful to allow container managers to pass some file
* descriptors to us pre-initialized. This enables
* socket-based activation of entire containers. */
manager_distribute_fds(m, fds);
/* We might have deserialized the notify fd, but if we didn't
* then let's create the bus now */
r = manager_setup_notify(m);
if (r < 0)
/* No sense to continue without notifications, our children would fail anyway. */
return r;
r = manager_setup_cgroups_agent(m);
if (r < 0)
/* Likewise, no sense to continue without empty cgroup notifications. */
return r;
r = manager_setup_user_lookup_fd(m);
if (r < 0)
/* This shouldn't fail, except if things are really broken. */
return r;
/* Let's set up our private bus connection now, unconditionally */
(void) bus_init_private(m);
/* If we are in --user mode also connect to the system bus now */
if (MANAGER_IS_USER(m))
(void) bus_init_system(m);
/* Let's connect to the bus now, but only if the unit is supposed to be up */
if (manager_dbus_is_running(m, !!serialization)) {
(void) bus_init_api(m);
if (MANAGER_IS_SYSTEM(m))
(void) bus_init_system(m);
}
/* Now that we are connected to all possible busses, let's deserialize who is tracking us. */
(void) bus_track_coldplug(m, &m->subscribed, false, m->deserialized_subscribed);
m->deserialized_subscribed = strv_free(m->deserialized_subscribed);
/* Third, fire things up! */
manager_coldplug(m);
/* Release any dynamic users no longer referenced */
dynamic_user_vacuum(m, true);
exec_runtime_vacuum(m);
/* Release any references to UIDs/GIDs no longer referenced, and destroy any IPC owned by them */
manager_vacuum_uid_refs(m);
manager_vacuum_gid_refs(m);
if (serialization) {
assert(m->n_reloading > 0);
m->n_reloading--;
/* Let's wait for the UnitNew/JobNew messages being
* sent, before we notify that the reload is
* finished */
m->send_reloading_done = true;
}
return 0;
}
int manager_add_job(Manager *m, JobType type, Unit *unit, JobMode mode, sd_bus_error *e, Job **_ret) {
int r;
Transaction *tr;
assert(m);
assert(type < _JOB_TYPE_MAX);
assert(unit);
assert(mode < _JOB_MODE_MAX);
if (mode == JOB_ISOLATE && type != JOB_START)
return sd_bus_error_setf(e, SD_BUS_ERROR_INVALID_ARGS, "Isolate is only valid for start.");
if (mode == JOB_ISOLATE && !unit->allow_isolate)
return sd_bus_error_setf(e, BUS_ERROR_NO_ISOLATION, "Operation refused, unit may not be isolated.");
log_unit_debug(unit, "Trying to enqueue job %s/%s/%s", unit->id, job_type_to_string(type), job_mode_to_string(mode));
type = job_type_collapse(type, unit);
tr = transaction_new(mode == JOB_REPLACE_IRREVERSIBLY);
if (!tr)
return -ENOMEM;
r = transaction_add_job_and_dependencies(tr, type, unit, NULL, true, false,
IN_SET(mode, JOB_IGNORE_DEPENDENCIES, JOB_IGNORE_REQUIREMENTS),
mode == JOB_IGNORE_DEPENDENCIES, e);
if (r < 0)
goto tr_abort;
if (mode == JOB_ISOLATE) {
r = transaction_add_isolate_jobs(tr, m);
if (r < 0)
goto tr_abort;
}
r = transaction_activate(tr, m, mode, e);
if (r < 0)
goto tr_abort;
log_unit_debug(unit,
"Enqueued job %s/%s as %u", unit->id,
job_type_to_string(type), (unsigned) tr->anchor_job->id);
if (_ret)
*_ret = tr->anchor_job;
transaction_free(tr);
return 0;
tr_abort:
transaction_abort(tr);
transaction_free(tr);
return r;
}
int manager_add_job_by_name(Manager *m, JobType type, const char *name, JobMode mode, sd_bus_error *e, Job **ret) {
Unit *unit = NULL; /* just to appease gcc, initialization is not really necessary */
int r;
assert(m);
assert(type < _JOB_TYPE_MAX);
assert(name);
assert(mode < _JOB_MODE_MAX);
r = manager_load_unit(m, name, NULL, NULL, &unit);
if (r < 0)
return r;
assert(unit);
return manager_add_job(m, type, unit, mode, e, ret);
}
int manager_add_job_by_name_and_warn(Manager *m, JobType type, const char *name, JobMode mode, Job **ret) {
_cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL;
int r;
assert(m);
assert(type < _JOB_TYPE_MAX);
assert(name);
assert(mode < _JOB_MODE_MAX);
r = manager_add_job_by_name(m, type, name, mode, &error, ret);
if (r < 0)
return log_warning_errno(r, "Failed to enqueue %s job for %s: %s", job_mode_to_string(mode), name, bus_error_message(&error, r));
return r;
}
int manager_propagate_reload(Manager *m, Unit *unit, JobMode mode, sd_bus_error *e) {
int r;
Transaction *tr;
assert(m);
assert(unit);
assert(mode < _JOB_MODE_MAX);
assert(mode != JOB_ISOLATE); /* Isolate is only valid for start */
tr = transaction_new(mode == JOB_REPLACE_IRREVERSIBLY);
if (!tr)
return -ENOMEM;
/* We need an anchor job */
r = transaction_add_job_and_dependencies(tr, JOB_NOP, unit, NULL, false, false, true, true, e);
if (r < 0)
goto tr_abort;
/* Failure in adding individual dependencies is ignored, so this always succeeds. */
transaction_add_propagate_reload_jobs(tr, unit, tr->anchor_job, mode == JOB_IGNORE_DEPENDENCIES, e);
r = transaction_activate(tr, m, mode, e);
if (r < 0)
goto tr_abort;
transaction_free(tr);
return 0;
tr_abort:
transaction_abort(tr);
transaction_free(tr);
return r;
}
Job *manager_get_job(Manager *m, uint32_t id) {
assert(m);
return hashmap_get(m->jobs, UINT32_TO_PTR(id));
}
Unit *manager_get_unit(Manager *m, const char *name) {
assert(m);
assert(name);
return hashmap_get(m->units, name);
}
static int manager_dispatch_target_deps_queue(Manager *m) {
Unit *u;
unsigned k;
int r = 0;
static const UnitDependency deps[] = {
UNIT_REQUIRED_BY,
UNIT_REQUISITE_OF,
UNIT_WANTED_BY,
UNIT_BOUND_BY
};
assert(m);
while ((u = m->target_deps_queue)) {
assert(u->in_target_deps_queue);
LIST_REMOVE(target_deps_queue, u->manager->target_deps_queue, u);
u->in_target_deps_queue = false;
for (k = 0; k < ELEMENTSOF(deps); k++) {
Unit *target;
Iterator i;
void *v;
HASHMAP_FOREACH_KEY(v, target, u->dependencies[deps[k]], i) {
r = unit_add_default_target_dependency(u, target);
if (r < 0)
return r;
}
}
}
return r;
}
unsigned manager_dispatch_load_queue(Manager *m) {
Unit *u;
unsigned n = 0;
assert(m);
/* Make sure we are not run recursively */
if (m->dispatching_load_queue)
return 0;
m->dispatching_load_queue = true;
/* Dispatches the load queue. Takes a unit from the queue and
* tries to load its data until the queue is empty */
while ((u = m->load_queue)) {
assert(u->in_load_queue);
unit_load(u);
n++;
}
m->dispatching_load_queue = false;
/* Dispatch the units waiting for their target dependencies to be added now, as all targets that we know about
* should be loaded and have aliases resolved */
(void) manager_dispatch_target_deps_queue(m);
return n;
}
int manager_load_unit_prepare(
Manager *m,
const char *name,
const char *path,
sd_bus_error *e,
Unit **_ret) {
_cleanup_(unit_freep) Unit *cleanup_ret = NULL;
Unit *ret;
UnitType t;
int r;
assert(m);
assert(name || path);
assert(_ret);
/* This will prepare the unit for loading, but not actually
* load anything from disk. */
if (path && !is_path(path))
return sd_bus_error_setf(e, SD_BUS_ERROR_INVALID_ARGS, "Path %s is not absolute.", path);
if (!name)
name = basename(path);
t = unit_name_to_type(name);
if (t == _UNIT_TYPE_INVALID || !unit_name_is_valid(name, UNIT_NAME_PLAIN|UNIT_NAME_INSTANCE)) {
if (unit_name_is_valid(name, UNIT_NAME_TEMPLATE))
return sd_bus_error_setf(e, SD_BUS_ERROR_INVALID_ARGS, "Unit name %s is missing the instance name.", name);
return sd_bus_error_setf(e, SD_BUS_ERROR_INVALID_ARGS, "Unit name %s is not valid.", name);
}
ret = manager_get_unit(m, name);
if (ret) {
*_ret = ret;
return 1;
}
ret = cleanup_ret = unit_new(m, unit_vtable[t]->object_size);
if (!ret)
return -ENOMEM;
if (path) {
ret->fragment_path = strdup(path);
if (!ret->fragment_path)
return -ENOMEM;
}
r = unit_add_name(ret, name);
if (r < 0)
return r;
unit_add_to_load_queue(ret);
unit_add_to_dbus_queue(ret);
unit_add_to_gc_queue(ret);
*_ret = ret;
cleanup_ret = NULL;
return 0;
}
int manager_load_unit(
Manager *m,
const char *name,
const char *path,
sd_bus_error *e,
Unit **_ret) {
int r;
assert(m);
assert(_ret);
/* This will load the service information files, but not actually
* start any services or anything. */
r = manager_load_unit_prepare(m, name, path, e, _ret);
if (r != 0)
return r;
manager_dispatch_load_queue(m);
*_ret = unit_follow_merge(*_ret);
return 0;
}
void manager_dump_jobs(Manager *s, FILE *f, const char *prefix) {
Iterator i;
Job *j;
assert(s);
assert(f);
HASHMAP_FOREACH(j, s->jobs, i)
job_dump(j, f, prefix);
}
void manager_dump_units(Manager *s, FILE *f, const char *prefix) {
Iterator i;
Unit *u;
const char *t;
assert(s);
assert(f);
HASHMAP_FOREACH_KEY(u, t, s->units, i)
if (u->id == t)
unit_dump(u, f, prefix);
}
void manager_dump(Manager *m, FILE *f, const char *prefix) {
ManagerTimestamp q;
assert(m);
assert(f);
for (q = 0; q < _MANAGER_TIMESTAMP_MAX; q++) {
char buf[FORMAT_TIMESTAMP_MAX];
if (dual_timestamp_is_set(m->timestamps + q))
fprintf(f, "%sTimestamp %s: %s\n",
strempty(prefix),
manager_timestamp_to_string(q),
format_timestamp(buf, sizeof(buf), m->timestamps[q].realtime));
}
manager_dump_units(m, f, prefix);
manager_dump_jobs(m, f, prefix);
}
int manager_get_dump_string(Manager *m, char **ret) {
_cleanup_free_ char *dump = NULL;
_cleanup_fclose_ FILE *f = NULL;
size_t size;
int r;
assert(m);
assert(ret);
f = open_memstream(&dump, &size);
if (!f)
return -errno;
(void) __fsetlocking(f, FSETLOCKING_BYCALLER);
manager_dump(m, f, NULL);
r = fflush_and_check(f);
if (r < 0)
return r;
f = safe_fclose(f);
*ret = TAKE_PTR(dump);
return 0;
}
void manager_clear_jobs(Manager *m) {
Job *j;
assert(m);
while ((j = hashmap_first(m->jobs)))
/* No need to recurse. We're cancelling all jobs. */
job_finish_and_invalidate(j, JOB_CANCELED, false, false);
}
static int manager_dispatch_run_queue(sd_event_source *source, void *userdata) {
Manager *m = userdata;
Job *j;
assert(source);
assert(m);
while ((j = m->run_queue)) {
assert(j->installed);
assert(j->in_run_queue);
job_run_and_invalidate(j);
}
if (m->n_running_jobs > 0)
manager_watch_jobs_in_progress(m);
if (m->n_on_console > 0)
manager_watch_idle_pipe(m);
return 1;
}
static unsigned manager_dispatch_dbus_queue(Manager *m) {
unsigned n = 0, budget;
Unit *u;
Job *j;
assert(m);
if (m->dispatching_dbus_queue)
return 0;
/* Anything to do at all? */
if (!m->dbus_unit_queue && !m->dbus_job_queue && !m->send_reloading_done && !m->queued_message)
return 0;
/* Do we have overly many messages queued at the moment? If so, let's not enqueue more on top, let's sit this
* cycle out, and process things in a later cycle when the queues got a bit emptier. */
if (manager_bus_n_queued_write(m) > MANAGER_BUS_BUSY_THRESHOLD)
return 0;
/* Only process a certain number of units/jobs per event loop iteration. Even if the bus queue wasn't overly
* full before this call we shouldn't increase it in size too wildly in one step, and we shouldn't monopolize
* CPU time with generating these messages. Note the difference in counting of this "budget" and the
* "threshold" above: the "budget" is decreased only once per generated message, regardless how many
* busses/direct connections it is enqueued on, while the "threshold" is applied to each queued instance of bus
* message, i.e. if the same message is enqueued to five busses/direct connections it will be counted five
* times. This difference in counting ("references" vs. "instances") is primarily a result of the fact that
* it's easier to implement it this way, however it also reflects the thinking that the "threshold" should put
* a limit on used queue memory, i.e. space, while the "budget" should put a limit on time. Also note that
* the "threshold" is currently chosen much higher than the "budget". */
budget = MANAGER_BUS_MESSAGE_BUDGET;
m->dispatching_dbus_queue = true;
while (budget > 0 && (u = m->dbus_unit_queue)) {
assert(u->in_dbus_queue);
bus_unit_send_change_signal(u);
n++, budget--;
}
while (budget > 0 && (j = m->dbus_job_queue)) {
assert(j->in_dbus_queue);
bus_job_send_change_signal(j);
n++, budget--;
}
m->dispatching_dbus_queue = false;
if (budget > 0 && m->send_reloading_done) {
m->send_reloading_done = false;
bus_manager_send_reloading(m, false);
n++, budget--;
}
if (budget > 0 && m->queued_message) {
bus_send_queued_message(m);
n++;
}
return n;
}
static int manager_dispatch_cgroups_agent_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata) {
Manager *m = userdata;
char buf[PATH_MAX+1];
ssize_t n;
n = recv(fd, buf, sizeof(buf), 0);
if (n < 0)
return log_error_errno(errno, "Failed to read cgroups agent message: %m");
if (n == 0) {
log_error("Got zero-length cgroups agent message, ignoring.");
return 0;
}
if ((size_t) n >= sizeof(buf)) {
log_error("Got overly long cgroups agent message, ignoring.");
return 0;
}
if (memchr(buf, 0, n)) {
log_error("Got cgroups agent message with embedded NUL byte, ignoring.");
return 0;
}
buf[n] = 0;
manager_notify_cgroup_empty(m, buf);
(void) bus_forward_agent_released(m, buf);
return 0;
}
static void manager_invoke_notify_message(
Manager *m,
Unit *u,
const struct ucred *ucred,
const char *buf,
FDSet *fds) {
assert(m);
assert(u);
assert(ucred);
assert(buf);
if (u->notifygen == m->notifygen) /* Already invoked on this same unit in this same iteration? */
return;
u->notifygen = m->notifygen;
if (UNIT_VTABLE(u)->notify_message) {
_cleanup_strv_free_ char **tags = NULL;
tags = strv_split(buf, NEWLINE);
if (!tags) {
log_oom();
return;
}
UNIT_VTABLE(u)->notify_message(u, ucred, tags, fds);
} else if (DEBUG_LOGGING) {
_cleanup_free_ char *x = NULL, *y = NULL;
x = ellipsize(buf, 20, 90);
if (x)
y = cescape(x);
log_unit_debug(u, "Got notification message \"%s\", ignoring.", strnull(y));
}
}
static int manager_dispatch_notify_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata) {
_cleanup_fdset_free_ FDSet *fds = NULL;
Manager *m = userdata;
char buf[NOTIFY_BUFFER_MAX+1];
struct iovec iovec = {
.iov_base = buf,
.iov_len = sizeof(buf)-1,
};
union {
struct cmsghdr cmsghdr;
uint8_t buf[CMSG_SPACE(sizeof(struct ucred)) +
CMSG_SPACE(sizeof(int) * NOTIFY_FD_MAX)];
} control = {};
struct msghdr msghdr = {
.msg_iov = &iovec,
.msg_iovlen = 1,
.msg_control = &control,
.msg_controllen = sizeof(control),
};
struct cmsghdr *cmsg;
struct ucred *ucred = NULL;
_cleanup_free_ Unit **array_copy = NULL;
Unit *u1, *u2, **array;
int r, *fd_array = NULL;
unsigned n_fds = 0;
bool found = false;
ssize_t n;
assert(m);
assert(m->notify_fd == fd);
if (revents != EPOLLIN) {
log_warning("Got unexpected poll event for notify fd.");
return 0;
}
n = recvmsg(m->notify_fd, &msghdr, MSG_DONTWAIT|MSG_CMSG_CLOEXEC|MSG_TRUNC);
if (n < 0) {
if (IN_SET(errno, EAGAIN, EINTR))
return 0; /* Spurious wakeup, try again */
/* If this is any other, real error, then let's stop processing this socket. This of course means we
* won't take notification messages anymore, but that's still better than busy looping around this:
* being woken up over and over again but being unable to actually read the message off the socket. */
return log_error_errno(errno, "Failed to receive notification message: %m");
}
CMSG_FOREACH(cmsg, &msghdr) {
if (cmsg->cmsg_level == SOL_SOCKET && cmsg->cmsg_type == SCM_RIGHTS) {
fd_array = (int*) CMSG_DATA(cmsg);
n_fds = (cmsg->cmsg_len - CMSG_LEN(0)) / sizeof(int);
} else if (cmsg->cmsg_level == SOL_SOCKET &&
cmsg->cmsg_type == SCM_CREDENTIALS &&
cmsg->cmsg_len == CMSG_LEN(sizeof(struct ucred))) {
ucred = (struct ucred*) CMSG_DATA(cmsg);
}
}
if (n_fds > 0) {
assert(fd_array);
r = fdset_new_array(&fds, fd_array, n_fds);
if (r < 0) {
close_many(fd_array, n_fds);
log_oom();
return 0;
}
}
if (!ucred || !pid_is_valid(ucred->pid)) {
log_warning("Received notify message without valid credentials. Ignoring.");
return 0;
}
if ((size_t) n >= sizeof(buf) || (msghdr.msg_flags & MSG_TRUNC)) {
log_warning("Received notify message exceeded maximum size. Ignoring.");
return 0;
}
/* As extra safety check, let's make sure the string we get doesn't contain embedded NUL bytes. We permit one
* trailing NUL byte in the message, but don't expect it. */
if (n > 1 && memchr(buf, 0, n-1)) {
log_warning("Received notify message with embedded NUL bytes. Ignoring.");
return 0;
}
/* Make sure it's NUL-terminated. */
buf[n] = 0;
/* Increase the generation counter used for filtering out duplicate unit invocations. */
m->notifygen++;
/* Notify every unit that might be interested, which might be multiple. */
u1 = manager_get_unit_by_pid_cgroup(m, ucred->pid);
u2 = hashmap_get(m->watch_pids, PID_TO_PTR(ucred->pid));
array = hashmap_get(m->watch_pids, PID_TO_PTR(-ucred->pid));
if (array) {
size_t k = 0;
while (array[k])
k++;
array_copy = newdup(Unit*, array, k+1);
if (!array_copy)
log_oom();
}
/* And now invoke the per-unit callbacks. Note that manager_invoke_notify_message() will handle duplicate units
* make sure we only invoke each unit's handler once. */
if (u1) {
manager_invoke_notify_message(m, u1, ucred, buf, fds);
found = true;
}
if (u2) {
manager_invoke_notify_message(m, u2, ucred, buf, fds);
found = true;
}
if (array_copy)
for (size_t i = 0; array_copy[i]; i++) {
manager_invoke_notify_message(m, array_copy[i], ucred, buf, fds);
found = true;
}
if (!found)
log_warning("Cannot find unit for notify message of PID "PID_FMT", ignoring.", ucred->pid);
if (fdset_size(fds) > 0)
log_warning("Got extra auxiliary fds with notification message, closing them.");
return 0;
}
static void manager_invoke_sigchld_event(
Manager *m,
Unit *u,
const siginfo_t *si) {
assert(m);
assert(u);
assert(si);
/* Already invoked the handler of this unit in this iteration? Then don't process this again */
if (u->sigchldgen == m->sigchldgen)
return;
u->sigchldgen = m->sigchldgen;
log_unit_debug(u, "Child "PID_FMT" belongs to %s.", si->si_pid, u->id);
unit_unwatch_pid(u, si->si_pid);
if (UNIT_VTABLE(u)->sigchld_event)
UNIT_VTABLE(u)->sigchld_event(u, si->si_pid, si->si_code, si->si_status);
}
static int manager_dispatch_sigchld(sd_event_source *source, void *userdata) {
Manager *m = userdata;
siginfo_t si = {};
int r;
assert(source);
assert(m);
/* First we call waitd() for a PID and do not reap the zombie. That way we can still access /proc/$PID for it
* while it is a zombie. */
if (waitid(P_ALL, 0, &si, WEXITED|WNOHANG|WNOWAIT) < 0) {
if (errno != ECHILD)
log_error_errno(errno, "Failed to peek for child with waitid(), ignoring: %m");
goto turn_off;
}
if (si.si_pid <= 0)
goto turn_off;
if (IN_SET(si.si_code, CLD_EXITED, CLD_KILLED, CLD_DUMPED)) {
_cleanup_free_ Unit **array_copy = NULL;
_cleanup_free_ char *name = NULL;
Unit *u1, *u2, **array;
(void) get_process_comm(si.si_pid, &name);
log_debug("Child "PID_FMT" (%s) died (code=%s, status=%i/%s)",
si.si_pid, strna(name),
sigchld_code_to_string(si.si_code),
si.si_status,
strna(si.si_code == CLD_EXITED
? exit_status_to_string(si.si_status, EXIT_STATUS_FULL)
: signal_to_string(si.si_status)));
/* Increase the generation counter used for filtering out duplicate unit invocations */
m->sigchldgen++;
/* And now figure out the unit this belongs to, it might be multiple... */
u1 = manager_get_unit_by_pid_cgroup(m, si.si_pid);
u2 = hashmap_get(m->watch_pids, PID_TO_PTR(si.si_pid));
array = hashmap_get(m->watch_pids, PID_TO_PTR(-si.si_pid));
if (array) {
size_t n = 0;
/* Cound how many entries the array has */
while (array[n])
n++;
/* Make a copy of the array so that we don't trip up on the array changing beneath us */
array_copy = newdup(Unit*, array, n+1);
if (!array_copy)
log_oom();
}
/* Finally, execute them all. Note that u1, u2 and the array might contain duplicates, but
* that's fine, manager_invoke_sigchld_event() will ensure we only invoke the handlers once for
* each iteration. */
if (u1)
manager_invoke_sigchld_event(m, u1, &si);
if (u2)
manager_invoke_sigchld_event(m, u2, &si);
if (array_copy)
for (size_t i = 0; array_copy[i]; i++)
manager_invoke_sigchld_event(m, array_copy[i], &si);
}
/* And now, we actually reap the zombie. */
if (waitid(P_PID, si.si_pid, &si, WEXITED) < 0) {
log_error_errno(errno, "Failed to dequeue child, ignoring: %m");
return 0;
}
return 0;
turn_off:
/* All children processed for now, turn off event source */
r = sd_event_source_set_enabled(m->sigchld_event_source, SD_EVENT_OFF);
if (r < 0)
return log_error_errno(r, "Failed to disable SIGCHLD event source: %m");
return 0;
}
static void manager_start_target(Manager *m, const char *name, JobMode mode) {
_cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL;
int r;
log_debug("Activating special unit %s", name);
r = manager_add_job_by_name(m, JOB_START, name, mode, &error, NULL);
if (r < 0)
log_error("Failed to enqueue %s job: %s", name, bus_error_message(&error, r));
}
static void manager_handle_ctrl_alt_del(Manager *m) {
/* If the user presses C-A-D more than
* 7 times within 2s, we reboot/shutdown immediately,
* unless it was disabled in system.conf */
if (ratelimit_test(&m->ctrl_alt_del_ratelimit) || m->cad_burst_action == EMERGENCY_ACTION_NONE)
manager_start_target(m, SPECIAL_CTRL_ALT_DEL_TARGET, JOB_REPLACE_IRREVERSIBLY);
else
emergency_action(m, m->cad_burst_action, NULL,
"Ctrl-Alt-Del was pressed more than 7 times within 2s");
}
static int manager_dispatch_signal_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata) {
Manager *m = userdata;
ssize_t n;
struct signalfd_siginfo sfsi;
int r;
assert(m);
assert(m->signal_fd == fd);
if (revents != EPOLLIN) {
log_warning("Got unexpected events from signal file descriptor.");
return 0;
}
n = read(m->signal_fd, &sfsi, sizeof(sfsi));
if (n != sizeof(sfsi)) {
if (n >= 0) {
log_warning("Truncated read from signal fd (%zu bytes), ignoring!", n);
return 0;
}
if (IN_SET(errno, EINTR, EAGAIN))
return 0;
/* We return an error here, which will kill this handler,
* to avoid a busy loop on read error. */
return log_error_errno(errno, "Reading from signal fd failed: %m");
}
log_received_signal(sfsi.ssi_signo == SIGCHLD ||
(sfsi.ssi_signo == SIGTERM && MANAGER_IS_USER(m))
? LOG_DEBUG : LOG_INFO,
&sfsi);
switch (sfsi.ssi_signo) {
case SIGCHLD:
r = sd_event_source_set_enabled(m->sigchld_event_source, SD_EVENT_ON);
if (r < 0)
log_warning_errno(r, "Failed to enable SIGCHLD event source, ignoring: %m");
break;
case SIGTERM:
if (MANAGER_IS_SYSTEM(m)) {
/* This is for compatibility with the original sysvinit */
r = verify_run_space_and_log("Refusing to reexecute");
if (r >= 0)
m->exit_code = MANAGER_REEXECUTE;
break;
}
_fallthrough_;
case SIGINT:
if (MANAGER_IS_SYSTEM(m))
manager_handle_ctrl_alt_del(m);
else
manager_start_target(m, SPECIAL_EXIT_TARGET,
JOB_REPLACE_IRREVERSIBLY);
break;
case SIGWINCH:
/* This is a nop on non-init */
if (MANAGER_IS_SYSTEM(m))
manager_start_target(m, SPECIAL_KBREQUEST_TARGET, JOB_REPLACE);
break;
case SIGPWR:
/* This is a nop on non-init */
if (MANAGER_IS_SYSTEM(m))
manager_start_target(m, SPECIAL_SIGPWR_TARGET, JOB_REPLACE);
break;
case SIGUSR1:
if (manager_dbus_is_running(m, false)) {
log_info("Trying to reconnect to bus...");
(void) bus_init_api(m);
if (MANAGER_IS_SYSTEM(m))
(void) bus_init_system(m);
} else {
log_info("Starting D-Bus service...");
manager_start_target(m, SPECIAL_DBUS_SERVICE, JOB_REPLACE);
}
break;
case SIGUSR2: {
_cleanup_free_ char *dump = NULL;
r = manager_get_dump_string(m, &dump);
if (r < 0) {
log_warning_errno(errno, "Failed to acquire manager dump: %m");
break;
}
log_dump(LOG_INFO, dump);
break;
}
case SIGHUP:
r = verify_run_space_and_log("Refusing to reload");
if (r >= 0)
m->exit_code = MANAGER_RELOAD;
break;
default: {
/* Starting SIGRTMIN+0 */
static const struct {
const char *target;
JobMode mode;
} target_table[] = {
[0] = { SPECIAL_DEFAULT_TARGET, JOB_ISOLATE },
[1] = { SPECIAL_RESCUE_TARGET, JOB_ISOLATE },
[2] = { SPECIAL_EMERGENCY_TARGET, JOB_ISOLATE },
[3] = { SPECIAL_HALT_TARGET, JOB_REPLACE_IRREVERSIBLY },
[4] = { SPECIAL_POWEROFF_TARGET, JOB_REPLACE_IRREVERSIBLY },
[5] = { SPECIAL_REBOOT_TARGET, JOB_REPLACE_IRREVERSIBLY },
[6] = { SPECIAL_KEXEC_TARGET, JOB_REPLACE_IRREVERSIBLY },
};
/* Starting SIGRTMIN+13, so that target halt and system halt are 10 apart */
static const ManagerExitCode code_table[] = {
[0] = MANAGER_HALT,
[1] = MANAGER_POWEROFF,
[2] = MANAGER_REBOOT,
[3] = MANAGER_KEXEC,
};
if ((int) sfsi.ssi_signo >= SIGRTMIN+0 &&
(int) sfsi.ssi_signo < SIGRTMIN+(int) ELEMENTSOF(target_table)) {
int idx = (int) sfsi.ssi_signo - SIGRTMIN;
manager_start_target(m, target_table[idx].target,
target_table[idx].mode);
break;
}
if ((int) sfsi.ssi_signo >= SIGRTMIN+13 &&
(int) sfsi.ssi_signo < SIGRTMIN+13+(int) ELEMENTSOF(code_table)) {
m->exit_code = code_table[sfsi.ssi_signo - SIGRTMIN - 13];
break;
}
switch (sfsi.ssi_signo - SIGRTMIN) {
case 20:
manager_set_show_status(m, SHOW_STATUS_YES);
break;
case 21:
manager_set_show_status(m, SHOW_STATUS_NO);
break;
case 22:
log_set_max_level(LOG_DEBUG);
log_info("Setting log level to debug.");
break;
case 23:
log_set_max_level(LOG_INFO);
log_info("Setting log level to info.");
break;
case 24:
if (MANAGER_IS_USER(m)) {
m->exit_code = MANAGER_EXIT;
return 0;
}
/* This is a nop on init */
break;
case 26:
case 29: /* compatibility: used to be mapped to LOG_TARGET_SYSLOG_OR_KMSG */
log_set_target(LOG_TARGET_JOURNAL_OR_KMSG);
log_notice("Setting log target to journal-or-kmsg.");
break;
case 27:
log_set_target(LOG_TARGET_CONSOLE);
log_notice("Setting log target to console.");
break;
case 28:
log_set_target(LOG_TARGET_KMSG);
log_notice("Setting log target to kmsg.");
break;
default:
log_warning("Got unhandled signal <%s>.", signal_to_string(sfsi.ssi_signo));
}
}}
return 0;
}
static int manager_dispatch_time_change_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata) {
Manager *m = userdata;
Iterator i;
Unit *u;
assert(m);
assert(m->time_change_fd == fd);
log_struct(LOG_DEBUG,
"MESSAGE_ID=" SD_MESSAGE_TIME_CHANGE_STR,
LOG_MESSAGE("Time has been changed"),
NULL);
/* Restart the watch */
m->time_change_event_source = sd_event_source_unref(m->time_change_event_source);
m->time_change_fd = safe_close(m->time_change_fd);
manager_setup_time_change(m);
HASHMAP_FOREACH(u, m->units, i)
if (UNIT_VTABLE(u)->time_change)
UNIT_VTABLE(u)->time_change(u);
return 0;
}
static int manager_dispatch_idle_pipe_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata) {
Manager *m = userdata;
assert(m);
assert(m->idle_pipe[2] == fd);
/* There's at least one Type=idle child that just gave up on us waiting for the boot process to complete. Let's
* now turn off any further console output if there's at least one service that needs console access, so that
* from now on our own output should not spill into that service's output anymore. After all, we support
* Type=idle only to beautify console output and it generally is set on services that want to own the console
* exclusively without our interference. */
m->no_console_output = m->n_on_console > 0;
/* Acknowledge the child's request, and let all all other children know too that they shouldn't wait any longer
* by closing the pipes towards them, which is what they are waiting for. */
manager_close_idle_pipe(m);
return 0;
}
static int manager_dispatch_jobs_in_progress(sd_event_source *source, usec_t usec, void *userdata) {
Manager *m = userdata;
int r;
uint64_t next;
assert(m);
assert(source);
manager_print_jobs_in_progress(m);
next = now(CLOCK_MONOTONIC) + JOBS_IN_PROGRESS_PERIOD_USEC;
r = sd_event_source_set_time(source, next);
if (r < 0)
return r;
return sd_event_source_set_enabled(source, SD_EVENT_ONESHOT);
}
int manager_loop(Manager *m) {
int r;
RATELIMIT_DEFINE(rl, 1*USEC_PER_SEC, 50000);
assert(m);
m->exit_code = MANAGER_OK;
/* Release the path cache */
m->unit_path_cache = set_free_free(m->unit_path_cache);
manager_check_finished(m);
/* There might still be some zombies hanging around from before we were exec()'ed. Let's reap them. */
r = sd_event_source_set_enabled(m->sigchld_event_source, SD_EVENT_ON);
if (r < 0)
return log_error_errno(r, "Failed to enable SIGCHLD event source: %m");
while (m->exit_code == MANAGER_OK) {
usec_t wait_usec;
if (m->runtime_watchdog > 0 && m->runtime_watchdog != USEC_INFINITY && MANAGER_IS_SYSTEM(m))
watchdog_ping();
if (!ratelimit_test(&rl)) {
/* Yay, something is going seriously wrong, pause a little */
log_warning("Looping too fast. Throttling execution a little.");
sleep(1);
}
if (manager_dispatch_load_queue(m) > 0)
continue;
if (manager_dispatch_gc_job_queue(m) > 0)
continue;
if (manager_dispatch_gc_unit_queue(m) > 0)
continue;
if (manager_dispatch_cleanup_queue(m) > 0)
continue;
if (manager_dispatch_cgroup_realize_queue(m) > 0)
continue;
if (manager_dispatch_dbus_queue(m) > 0)
continue;
/* Sleep for half the watchdog time */
if (m->runtime_watchdog > 0 && m->runtime_watchdog != USEC_INFINITY && MANAGER_IS_SYSTEM(m)) {
wait_usec = m->runtime_watchdog / 2;
if (wait_usec <= 0)
wait_usec = 1;
} else
wait_usec = USEC_INFINITY;
r = sd_event_run(m->event, wait_usec);
if (r < 0)
return log_error_errno(r, "Failed to run event loop: %m");
}
return m->exit_code;
}
int manager_load_unit_from_dbus_path(Manager *m, const char *s, sd_bus_error *e, Unit **_u) {
_cleanup_free_ char *n = NULL;
sd_id128_t invocation_id;
Unit *u;
int r;
assert(m);
assert(s);
assert(_u);
r = unit_name_from_dbus_path(s, &n);
if (r < 0)
return r;
/* Permit addressing units by invocation ID: if the passed bus path is suffixed by a 128bit ID then we use it
* as invocation ID. */
r = sd_id128_from_string(n, &invocation_id);
if (r >= 0) {
u = hashmap_get(m->units_by_invocation_id, &invocation_id);
if (u) {
*_u = u;
return 0;
}
return sd_bus_error_setf(e, BUS_ERROR_NO_UNIT_FOR_INVOCATION_ID, "No unit with the specified invocation ID " SD_ID128_FORMAT_STR " known.", SD_ID128_FORMAT_VAL(invocation_id));
}
/* If this didn't work, we check if this is a unit name */
if (!unit_name_is_valid(n, UNIT_NAME_PLAIN|UNIT_NAME_INSTANCE))
return sd_bus_error_setf(e, SD_BUS_ERROR_INVALID_ARGS, "Unit name %s is neither a valid invocation ID nor unit name.", n);
r = manager_load_unit(m, n, NULL, e, &u);
if (r < 0)
return r;
*_u = u;
return 0;
}
int manager_get_job_from_dbus_path(Manager *m, const char *s, Job **_j) {
const char *p;
unsigned id;
Job *j;
int r;
assert(m);
assert(s);
assert(_j);
p = startswith(s, "/org/freedesktop/systemd1/job/");
if (!p)
return -EINVAL;
r = safe_atou(p, &id);
if (r < 0)
return r;
j = manager_get_job(m, id);
if (!j)
return -ENOENT;
*_j = j;
return 0;
}
void manager_send_unit_audit(Manager *m, Unit *u, int type, bool success) {
#if HAVE_AUDIT
_cleanup_free_ char *p = NULL;
const char *msg;
int audit_fd, r;
if (!MANAGER_IS_SYSTEM(m))
return;
audit_fd = get_audit_fd();
if (audit_fd < 0)
return;
/* Don't generate audit events if the service was already
* started and we're just deserializing */
if (MANAGER_IS_RELOADING(m))
return;
if (u->type != UNIT_SERVICE)
return;
r = unit_name_to_prefix_and_instance(u->id, &p);
if (r < 0) {
log_error_errno(r, "Failed to extract prefix and instance of unit name: %m");
return;
}
msg = strjoina("unit=", p);
if (audit_log_user_comm_message(audit_fd, type, msg, "systemd", NULL, NULL, NULL, success) < 0) {
if (errno == EPERM)
/* We aren't allowed to send audit messages?
* Then let's not retry again. */
close_audit_fd();
else
log_warning_errno(errno, "Failed to send audit message: %m");
}
#endif
}
void manager_send_unit_plymouth(Manager *m, Unit *u) {
static const union sockaddr_union sa = PLYMOUTH_SOCKET;
_cleanup_free_ char *message = NULL;
_cleanup_close_ int fd = -1;
int n = 0;
/* Don't generate plymouth events if the service was already
* started and we're just deserializing */
if (MANAGER_IS_RELOADING(m))
return;
if (!MANAGER_IS_SYSTEM(m))
return;
if (detect_container() > 0)
return;
if (!IN_SET(u->type, UNIT_SERVICE, UNIT_MOUNT, UNIT_SWAP))
return;
/* We set SOCK_NONBLOCK here so that we rather drop the
* message then wait for plymouth */
fd = socket(AF_UNIX, SOCK_STREAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
if (fd < 0) {
log_error_errno(errno, "socket() failed: %m");
return;
}
if (connect(fd, &sa.sa, SOCKADDR_UN_LEN(sa.un)) < 0) {
if (!IN_SET(errno, EPIPE, EAGAIN, ENOENT, ECONNREFUSED, ECONNRESET, ECONNABORTED))
log_error_errno(errno, "connect() failed: %m");
return;
}
if (asprintf(&message, "U\002%c%s%n", (int) (strlen(u->id) + 1), u->id, &n) < 0) {
log_oom();
return;
}
errno = 0;
if (write(fd, message, n + 1) != n + 1)
if (!IN_SET(errno, EPIPE, EAGAIN, ENOENT, ECONNREFUSED, ECONNRESET, ECONNABORTED))
log_error_errno(errno, "Failed to write Plymouth message: %m");
}
int manager_open_serialization(Manager *m, FILE **_f) {
int fd;
FILE *f;
assert(_f);
fd = open_serialization_fd("systemd-state");
if (fd < 0)
return fd;
f = fdopen(fd, "w+");
if (!f) {
safe_close(fd);
return -errno;
}
*_f = f;
return 0;
}
int manager_serialize(Manager *m, FILE *f, FDSet *fds, bool switching_root) {
ManagerTimestamp q;
const char *t;
Iterator i;
Unit *u;
int r;
assert(m);
assert(f);
assert(fds);
m->n_reloading++;
fprintf(f, "current-job-id=%"PRIu32"\n", m->current_job_id);
fprintf(f, "n-installed-jobs=%u\n", m->n_installed_jobs);
fprintf(f, "n-failed-jobs=%u\n", m->n_failed_jobs);
fprintf(f, "taint-usr=%s\n", yes_no(m->taint_usr));
fprintf(f, "ready-sent=%s\n", yes_no(m->ready_sent));
fprintf(f, "taint-logged=%s\n", yes_no(m->taint_logged));
fprintf(f, "service-watchdogs=%s\n", yes_no(m->service_watchdogs));
for (q = 0; q < _MANAGER_TIMESTAMP_MAX; q++) {
/* The userspace and finish timestamps only apply to the host system, hence only serialize them there */
if (in_initrd() && IN_SET(q, MANAGER_TIMESTAMP_USERSPACE, MANAGER_TIMESTAMP_FINISH))
continue;
t = manager_timestamp_to_string(q);
{
char field[strlen(t) + STRLEN("-timestamp") + 1];
strcpy(stpcpy(field, t), "-timestamp");
dual_timestamp_serialize(f, field, m->timestamps + q);
}
}
if (!switching_root)
(void) serialize_environment(f, m->environment);
if (m->notify_fd >= 0) {
int copy;
copy = fdset_put_dup(fds, m->notify_fd);
if (copy < 0)
return copy;
fprintf(f, "notify-fd=%i\n", copy);
fprintf(f, "notify-socket=%s\n", m->notify_socket);
}
if (m->cgroups_agent_fd >= 0) {
int copy;
copy = fdset_put_dup(fds, m->cgroups_agent_fd);
if (copy < 0)
return copy;
fprintf(f, "cgroups-agent-fd=%i\n", copy);
}
if (m->user_lookup_fds[0] >= 0) {
int copy0, copy1;
copy0 = fdset_put_dup(fds, m->user_lookup_fds[0]);
if (copy0 < 0)
return copy0;
copy1 = fdset_put_dup(fds, m->user_lookup_fds[1]);
if (copy1 < 0)
return copy1;
fprintf(f, "user-lookup=%i %i\n", copy0, copy1);
}
bus_track_serialize(m->subscribed, f, "subscribed");
r = dynamic_user_serialize(m, f, fds);
if (r < 0)
return r;
manager_serialize_uid_refs(m, f);
manager_serialize_gid_refs(m, f);
r = exec_runtime_serialize(m, f, fds);
if (r < 0)
return r;
(void) fputc('\n', f);
HASHMAP_FOREACH_KEY(u, t, m->units, i) {
if (u->id != t)
continue;
/* Start marker */
fputs(u->id, f);
fputc('\n', f);
r = unit_serialize(u, f, fds, !switching_root);
if (r < 0) {
m->n_reloading--;
return r;
}
}
assert(m->n_reloading > 0);
m->n_reloading--;
if (ferror(f))
return -EIO;
r = bus_fdset_add_all(m, fds);
if (r < 0)
return r;
return 0;
}
int manager_deserialize(Manager *m, FILE *f, FDSet *fds) {
int r = 0;
assert(m);
assert(f);
log_debug("Deserializing state...");
m->n_reloading++;
for (;;) {
char line[LINE_MAX];
const char *val, *l;
if (!fgets(line, sizeof(line), f)) {
if (feof(f))
r = 0;
else
r = -errno;
goto finish;
}
char_array_0(line);
l = strstrip(line);
if (l[0] == 0)
break;
if ((val = startswith(l, "current-job-id="))) {
uint32_t id;
if (safe_atou32(val, &id) < 0)
log_notice("Failed to parse current job id value %s", val);
else
m->current_job_id = MAX(m->current_job_id, id);
} else if ((val = startswith(l, "n-installed-jobs="))) {
uint32_t n;
if (safe_atou32(val, &n) < 0)
log_notice("Failed to parse installed jobs counter %s", val);
else
m->n_installed_jobs += n;
} else if ((val = startswith(l, "n-failed-jobs="))) {
uint32_t n;
if (safe_atou32(val, &n) < 0)
log_notice("Failed to parse failed jobs counter %s", val);
else
m->n_failed_jobs += n;
} else if ((val = startswith(l, "taint-usr="))) {
int b;
b = parse_boolean(val);
if (b < 0)
log_notice("Failed to parse taint /usr flag %s", val);
else
m->taint_usr = m->taint_usr || b;
} else if ((val = startswith(l, "ready-sent="))) {
int b;
b = parse_boolean(val);
if (b < 0)
log_notice("Failed to parse ready-sent flag %s", val);
else
m->ready_sent = m->ready_sent || b;
} else if ((val = startswith(l, "taint-logged="))) {
int b;
b = parse_boolean(val);
if (b < 0)
log_notice("Failed to parse taint-logged flag %s", val);
else
m->taint_logged = m->taint_logged || b;
} else if ((val = startswith(l, "service-watchdogs="))) {
int b;
b = parse_boolean(val);
if (b < 0)
log_notice("Failed to parse service-watchdogs flag %s", val);
else
m->service_watchdogs = b;
} else if (startswith(l, "env=")) {
r = deserialize_environment(&m->environment, l);
if (r == -ENOMEM)
goto finish;
if (r < 0)
log_notice_errno(r, "Failed to parse environment entry: \"%s\": %m", l);
} else if ((val = startswith(l, "notify-fd="))) {
int fd;
if (safe_atoi(val, &fd) < 0 || fd < 0 || !fdset_contains(fds, fd))
log_notice("Failed to parse notify fd: \"%s\"", val);
else {
m->notify_event_source = sd_event_source_unref(m->notify_event_source);
safe_close(m->notify_fd);
m->notify_fd = fdset_remove(fds, fd);
}
} else if ((val = startswith(l, "notify-socket="))) {
char *n;
n = strdup(val);
if (!n) {
r = -ENOMEM;
goto finish;
}
free(m->notify_socket);
m->notify_socket = n;
} else if ((val = startswith(l, "cgroups-agent-fd="))) {
int fd;
if (safe_atoi(val, &fd) < 0 || fd < 0 || !fdset_contains(fds, fd))
log_notice("Failed to parse cgroups agent fd: %s", val);
else {
m->cgroups_agent_event_source = sd_event_source_unref(m->cgroups_agent_event_source);
safe_close(m->cgroups_agent_fd);
m->cgroups_agent_fd = fdset_remove(fds, fd);
}
} else if ((val = startswith(l, "user-lookup="))) {
int fd0, fd1;
if (sscanf(val, "%i %i", &fd0, &fd1) != 2 || fd0 < 0 || fd1 < 0 || fd0 == fd1 || !fdset_contains(fds, fd0) || !fdset_contains(fds, fd1))
log_notice("Failed to parse user lookup fd: %s", val);
else {
m->user_lookup_event_source = sd_event_source_unref(m->user_lookup_event_source);
safe_close_pair(m->user_lookup_fds);
m->user_lookup_fds[0] = fdset_remove(fds, fd0);
m->user_lookup_fds[1] = fdset_remove(fds, fd1);
}
} else if ((val = startswith(l, "dynamic-user=")))
dynamic_user_deserialize_one(m, val, fds);
else if ((val = startswith(l, "destroy-ipc-uid=")))
manager_deserialize_uid_refs_one(m, val);
else if ((val = startswith(l, "destroy-ipc-gid=")))
manager_deserialize_gid_refs_one(m, val);
else if ((val = startswith(l, "exec-runtime=")))
exec_runtime_deserialize_one(m, val, fds);
else if ((val = startswith(l, "subscribed="))) {
if (strv_extend(&m->deserialized_subscribed, val) < 0)
log_oom();
} else {
ManagerTimestamp q;
for (q = 0; q < _MANAGER_TIMESTAMP_MAX; q++) {
val = startswith(l, manager_timestamp_to_string(q));
if (!val)
continue;
val = startswith(val, "-timestamp=");
if (val)
break;
}
if (q < _MANAGER_TIMESTAMP_MAX) /* found it */
dual_timestamp_deserialize(val, m->timestamps + q);
else if (!startswith(l, "kdbus-fd=")) /* ignore kdbus */
log_notice("Unknown serialization item '%s'", l);
}
}
for (;;) {
Unit *u;
char name[UNIT_NAME_MAX+2];
const char* unit_name;
/* Start marker */
if (!fgets(name, sizeof(name), f)) {
if (feof(f))
r = 0;
else
r = -errno;
goto finish;
}
char_array_0(name);
unit_name = strstrip(name);
r = manager_load_unit(m, unit_name, NULL, NULL, &u);
if (r < 0) {
log_notice_errno(r, "Failed to load unit \"%s\", skipping deserialization: %m", unit_name);
if (r == -ENOMEM)
goto finish;
unit_deserialize_skip(f);
continue;
}
r = unit_deserialize(u, f, fds);
if (r < 0) {
log_notice_errno(r, "Failed to deserialize unit \"%s\": %m", unit_name);
if (r == -ENOMEM)
goto finish;
}
}
finish:
if (ferror(f))
r = -EIO;
assert(m->n_reloading > 0);
m->n_reloading--;
return r;
}
int manager_reload(Manager *m) {
int r, q;
_cleanup_fclose_ FILE *f = NULL;
_cleanup_fdset_free_ FDSet *fds = NULL;
assert(m);
r = manager_open_serialization(m, &f);
if (r < 0)
return r;
m->n_reloading++;
bus_manager_send_reloading(m, true);
fds = fdset_new();
if (!fds) {
m->n_reloading--;
return -ENOMEM;
}
r = manager_serialize(m, f, fds, false);
if (r < 0) {
m->n_reloading--;
return r;
}
if (fseeko(f, 0, SEEK_SET) < 0) {
m->n_reloading--;
return -errno;
}
/* From here on there is no way back. */
manager_clear_jobs_and_units(m);
lookup_paths_flush_generator(&m->lookup_paths);
lookup_paths_free(&m->lookup_paths);
exec_runtime_vacuum(m);
dynamic_user_vacuum(m, false);
m->uid_refs = hashmap_free(m->uid_refs);
m->gid_refs = hashmap_free(m->gid_refs);
q = lookup_paths_init(&m->lookup_paths, m->unit_file_scope, 0, NULL);
if (q < 0 && r >= 0)
r = q;
q = manager_run_environment_generators(m);
if (q < 0 && r >= 0)
r = q;
/* Find new unit paths */
q = manager_run_generators(m);
if (q < 0 && r >= 0)
r = q;
lookup_paths_reduce(&m->lookup_paths);
manager_build_unit_path_cache(m);
/* First, enumerate what we can from all config files */
manager_enumerate(m);
/* Second, deserialize our stored data */
q = manager_deserialize(m, f, fds);
if (q < 0) {
log_error_errno(q, "Deserialization failed: %m");
if (r >= 0)
r = q;
}
fclose(f);
f = NULL;
/* Re-register notify_fd as event source */
q = manager_setup_notify(m);
if (q < 0 && r >= 0)
r = q;
q = manager_setup_cgroups_agent(m);
if (q < 0 && r >= 0)
r = q;
q = manager_setup_user_lookup_fd(m);
if (q < 0 && r >= 0)
r = q;
/* Third, fire things up! */
manager_coldplug(m);
/* Release any dynamic users no longer referenced */
dynamic_user_vacuum(m, true);
/* Release any references to UIDs/GIDs no longer referenced, and destroy any IPC owned by them */
manager_vacuum_uid_refs(m);
manager_vacuum_gid_refs(m);
exec_runtime_vacuum(m);
assert(m->n_reloading > 0);
m->n_reloading--;
/* It might be safe to log to the journal now and connect to dbus */
manager_recheck_journal(m);
manager_recheck_dbus(m);
/* Sync current state of bus names with our set of listening units */
q = manager_enqueue_sync_bus_names(m);
if (q < 0 && r >= 0)
r = q;
m->send_reloading_done = true;
return r;
}
void manager_reset_failed(Manager *m) {
Unit *u;
Iterator i;
assert(m);
HASHMAP_FOREACH(u, m->units, i)
unit_reset_failed(u);
}
bool manager_unit_inactive_or_pending(Manager *m, const char *name) {
Unit *u;
assert(m);
assert(name);
/* Returns true if the unit is inactive or going down */
u = manager_get_unit(m, name);
if (!u)
return true;
return unit_inactive_or_pending(u);
}
static void log_taint_string(Manager *m) {
_cleanup_free_ char *taint = NULL;
assert(m);
if (MANAGER_IS_USER(m) || m->taint_logged)
return;
m->taint_logged = true; /* only check for taint once */
taint = manager_taint_string(m);
if (isempty(taint))
return;
log_struct(LOG_NOTICE,
LOG_MESSAGE("System is tainted: %s", taint),
"TAINT=%s", taint,
"MESSAGE_ID=" SD_MESSAGE_TAINTED_STR,
NULL);
}
static void manager_notify_finished(Manager *m) {
char userspace[FORMAT_TIMESPAN_MAX], initrd[FORMAT_TIMESPAN_MAX], kernel[FORMAT_TIMESPAN_MAX], sum[FORMAT_TIMESPAN_MAX];
usec_t firmware_usec, loader_usec, kernel_usec, initrd_usec, userspace_usec, total_usec;
if (m->test_run_flags)
return;
if (MANAGER_IS_SYSTEM(m) && detect_container() <= 0) {
char ts[FORMAT_TIMESPAN_MAX];
char buf[FORMAT_TIMESPAN_MAX + STRLEN(" (firmware) + ") + FORMAT_TIMESPAN_MAX + STRLEN(" (loader) + ")]
= {};
char *p = buf;
size_t size = sizeof buf;
/* Note that MANAGER_TIMESTAMP_KERNEL's monotonic value is always at 0, and
* MANAGER_TIMESTAMP_FIRMWARE's and MANAGER_TIMESTAMP_LOADER's monotonic value should be considered
* negative values. */
firmware_usec = m->timestamps[MANAGER_TIMESTAMP_FIRMWARE].monotonic - m->timestamps[MANAGER_TIMESTAMP_LOADER].monotonic;
loader_usec = m->timestamps[MANAGER_TIMESTAMP_LOADER].monotonic - m->timestamps[MANAGER_TIMESTAMP_KERNEL].monotonic;
userspace_usec = m->timestamps[MANAGER_TIMESTAMP_FINISH].monotonic - m->timestamps[MANAGER_TIMESTAMP_USERSPACE].monotonic;
total_usec = m->timestamps[MANAGER_TIMESTAMP_FIRMWARE].monotonic + m->timestamps[MANAGER_TIMESTAMP_FINISH].monotonic;
if (firmware_usec > 0)
size = strpcpyf(&p, size, "%s (firmware) + ", format_timespan(ts, sizeof(ts), firmware_usec, USEC_PER_MSEC));
if (loader_usec > 0)
size = strpcpyf(&p, size, "%s (loader) + ", format_timespan(ts, sizeof(ts), loader_usec, USEC_PER_MSEC));
if (dual_timestamp_is_set(&m->timestamps[MANAGER_TIMESTAMP_INITRD])) {
/* The initrd case on bare-metal*/
kernel_usec = m->timestamps[MANAGER_TIMESTAMP_INITRD].monotonic - m->timestamps[MANAGER_TIMESTAMP_KERNEL].monotonic;
initrd_usec = m->timestamps[MANAGER_TIMESTAMP_USERSPACE].monotonic - m->timestamps[MANAGER_TIMESTAMP_INITRD].monotonic;
log_struct(LOG_INFO,
"MESSAGE_ID=" SD_MESSAGE_STARTUP_FINISHED_STR,
"KERNEL_USEC="USEC_FMT, kernel_usec,
"INITRD_USEC="USEC_FMT, initrd_usec,
"USERSPACE_USEC="USEC_FMT, userspace_usec,
LOG_MESSAGE("Startup finished in %s%s (kernel) + %s (initrd) + %s (userspace) = %s.",
buf,
format_timespan(kernel, sizeof(kernel), kernel_usec, USEC_PER_MSEC),
format_timespan(initrd, sizeof(initrd), initrd_usec, USEC_PER_MSEC),
format_timespan(userspace, sizeof(userspace), userspace_usec, USEC_PER_MSEC),
format_timespan(sum, sizeof(sum), total_usec, USEC_PER_MSEC)),
NULL);
} else {
/* The initrd-less case on bare-metal*/
kernel_usec = m->timestamps[MANAGER_TIMESTAMP_USERSPACE].monotonic - m->timestamps[MANAGER_TIMESTAMP_KERNEL].monotonic;
initrd_usec = 0;
log_struct(LOG_INFO,
"MESSAGE_ID=" SD_MESSAGE_STARTUP_FINISHED_STR,
"KERNEL_USEC="USEC_FMT, kernel_usec,
"USERSPACE_USEC="USEC_FMT, userspace_usec,
LOG_MESSAGE("Startup finished in %s%s (kernel) + %s (userspace) = %s.",
buf,
format_timespan(kernel, sizeof(kernel), kernel_usec, USEC_PER_MSEC),
format_timespan(userspace, sizeof(userspace), userspace_usec, USEC_PER_MSEC),
format_timespan(sum, sizeof(sum), total_usec, USEC_PER_MSEC)),
NULL);
}
} else {
/* The container and --user case */
firmware_usec = loader_usec = initrd_usec = kernel_usec = 0;
total_usec = userspace_usec = m->timestamps[MANAGER_TIMESTAMP_FINISH].monotonic - m->timestamps[MANAGER_TIMESTAMP_USERSPACE].monotonic;
log_struct(LOG_INFO,
"MESSAGE_ID=" SD_MESSAGE_USER_STARTUP_FINISHED_STR,
"USERSPACE_USEC="USEC_FMT, userspace_usec,
LOG_MESSAGE("Startup finished in %s.",
format_timespan(sum, sizeof(sum), total_usec, USEC_PER_MSEC)),
NULL);
}
bus_manager_send_finished(m, firmware_usec, loader_usec, kernel_usec, initrd_usec, userspace_usec, total_usec);
sd_notifyf(false,
m->ready_sent ? "STATUS=Startup finished in %s."
: "READY=1\n"
"STATUS=Startup finished in %s.",
format_timespan(sum, sizeof(sum), total_usec, USEC_PER_MSEC));
m->ready_sent = true;
log_taint_string(m);
}
static void manager_send_ready(Manager *m) {
assert(m);
/* We send READY=1 on reaching basic.target only when running in --user mode. */
if (!MANAGER_IS_USER(m) || m->ready_sent)
return;
m->ready_sent = true;
sd_notifyf(false,
"READY=1\n"
"STATUS=Reached " SPECIAL_BASIC_TARGET ".");
}
static void manager_check_basic_target(Manager *m) {
Unit *u;
assert(m);
/* Small shortcut */
if (m->ready_sent && m->taint_logged)
return;
u = manager_get_unit(m, SPECIAL_BASIC_TARGET);
if (!u || !UNIT_IS_ACTIVE_OR_RELOADING(unit_active_state(u)))
return;
/* For user managers, send out READY=1 as soon as we reach basic.target */
manager_send_ready(m);
/* Log the taint string as soon as we reach basic.target */
log_taint_string(m);
}
void manager_check_finished(Manager *m) {
assert(m);
if (MANAGER_IS_RELOADING(m))
return;
/* Verify that we have entered the event loop already, and not left it again. */
if (!MANAGER_IS_RUNNING(m))
return;
manager_check_basic_target(m);
if (hashmap_size(m->jobs) > 0) {
if (m->jobs_in_progress_event_source)
/* Ignore any failure, this is only for feedback */
(void) sd_event_source_set_time(m->jobs_in_progress_event_source, now(CLOCK_MONOTONIC) + JOBS_IN_PROGRESS_WAIT_USEC);
return;
}
manager_flip_auto_status(m, false);
/* Notify Type=idle units that we are done now */
manager_close_idle_pipe(m);
/* Turn off confirm spawn now */
m->confirm_spawn = NULL;
/* No need to update ask password status when we're going non-interactive */
manager_close_ask_password(m);
/* This is no longer the first boot */
manager_set_first_boot(m, false);
if (MANAGER_IS_FINISHED(m))
return;
dual_timestamp_get(m->timestamps + MANAGER_TIMESTAMP_FINISH);
manager_notify_finished(m);
manager_invalidate_startup_units(m);
}
static bool generator_path_any(const char* const* paths) {
char **path;
bool found = false;
/* Optimize by skipping the whole process by not creating output directories
* if no generators are found. */
STRV_FOREACH(path, (char**) paths)
if (access(*path, F_OK) == 0)
found = true;
else if (errno != ENOENT)
log_warning_errno(errno, "Failed to open generator directory %s: %m", *path);
return found;
}
static const char* system_env_generator_binary_paths[] = {
"/run/systemd/system-environment-generators",
"/etc/systemd/system-environment-generators",
"/usr/local/lib/systemd/system-environment-generators",
SYSTEM_ENV_GENERATOR_PATH,
NULL
};
static const char* user_env_generator_binary_paths[] = {
"/run/systemd/user-environment-generators",
"/etc/systemd/user-environment-generators",
"/usr/local/lib/systemd/user-environment-generators",
USER_ENV_GENERATOR_PATH,
NULL
};
static int manager_run_environment_generators(Manager *m) {
char **tmp = NULL; /* this is only used in the forked process, no cleanup here */
const char **paths;
void* args[] = {&tmp, &tmp, &m->environment};
if (m->test_run_flags && !(m->test_run_flags & MANAGER_TEST_RUN_ENV_GENERATORS))
return 0;
paths = MANAGER_IS_SYSTEM(m) ? system_env_generator_binary_paths : user_env_generator_binary_paths;
if (!generator_path_any(paths))
return 0;
return execute_directories(paths, DEFAULT_TIMEOUT_USEC, gather_environment, args, NULL);
}
static int manager_run_generators(Manager *m) {
_cleanup_strv_free_ char **paths = NULL;
const char *argv[5];
int r;
assert(m);
if (m->test_run_flags && !(m->test_run_flags & MANAGER_TEST_RUN_GENERATORS))
return 0;
paths = generator_binary_paths(m->unit_file_scope);
if (!paths)
return log_oom();
if (!generator_path_any((const char* const*) paths))
return 0;
r = lookup_paths_mkdir_generator(&m->lookup_paths);
if (r < 0)
goto finish;
argv[0] = NULL; /* Leave this empty, execute_directory() will fill something in */
argv[1] = m->lookup_paths.generator;
argv[2] = m->lookup_paths.generator_early;
argv[3] = m->lookup_paths.generator_late;
argv[4] = NULL;
RUN_WITH_UMASK(0022)
execute_directories((const char* const*) paths, DEFAULT_TIMEOUT_USEC,
NULL, NULL, (char**) argv);
finish:
lookup_paths_trim_generator(&m->lookup_paths);
return r;
}
int manager_environment_add(Manager *m, char **minus, char **plus) {
char **a = NULL, **b = NULL, **l;
assert(m);
l = m->environment;
if (!strv_isempty(minus)) {
a = strv_env_delete(l, 1, minus);
if (!a)
return -ENOMEM;
l = a;
}
if (!strv_isempty(plus)) {
b = strv_env_merge(2, l, plus);
if (!b) {
strv_free(a);
return -ENOMEM;
}
l = b;
}
if (m->environment != l)
strv_free(m->environment);
if (a != l)
strv_free(a);
if (b != l)
strv_free(b);
m->environment = l;
manager_sanitize_environment(m);
return 0;
}
int manager_set_default_rlimits(Manager *m, struct rlimit **default_rlimit) {
int i;
assert(m);
for (i = 0; i < _RLIMIT_MAX; i++) {
m->rlimit[i] = mfree(m->rlimit[i]);
if (!default_rlimit[i])
continue;
m->rlimit[i] = newdup(struct rlimit, default_rlimit[i], 1);
if (!m->rlimit[i])
return log_oom();
}
return 0;
}
void manager_recheck_dbus(Manager *m) {
assert(m);
/* Connects to the bus if the dbus service and socket are running. If we are running in user mode this is all
* it does. In system mode we'll also connect to the system bus (which will most likely just reuse the
* connection of the API bus). That's because the system bus after all runs as service of the system instance,
* while in the user instance we can assume it's already there. */
if (MANAGER_IS_RELOADING(m))
return; /* don't check while we are reloading… */
if (manager_dbus_is_running(m, false)) {
(void) bus_init_api(m);
if (MANAGER_IS_SYSTEM(m))
(void) bus_init_system(m);
} else {
(void) bus_done_api(m);
if (MANAGER_IS_SYSTEM(m))
(void) bus_done_system(m);
}
}
static bool manager_journal_is_running(Manager *m) {
Unit *u;
assert(m);
if (m->test_run_flags != 0)
return false;
/* If we are the user manager we can safely assume that the journal is up */
if (!MANAGER_IS_SYSTEM(m))
return true;
/* Check that the socket is not only up, but in RUNNING state */
u = manager_get_unit(m, SPECIAL_JOURNALD_SOCKET);
if (!u)
return false;
if (SOCKET(u)->state != SOCKET_RUNNING)
return false;
/* Similar, check if the daemon itself is fully up, too */
u = manager_get_unit(m, SPECIAL_JOURNALD_SERVICE);
if (!u)
return false;
if (!IN_SET(SERVICE(u)->state, SERVICE_RELOAD, SERVICE_RUNNING))
return false;
return true;
}
void manager_recheck_journal(Manager *m) {
assert(m);
/* Don't bother with this unless we are in the special situation of being PID 1 */
if (getpid_cached() != 1)
return;
/* Don't check this while we are reloading, things might still change */
if (MANAGER_IS_RELOADING(m))
return;
/* The journal is fully and entirely up? If so, let's permit logging to it, if that's configured. If the
* journal is down, don't ever log to it, otherwise we might end up deadlocking ourselves as we might trigger
* an activation ourselves we can't fulfill. */
log_set_prohibit_ipc(!manager_journal_is_running(m));
log_open();
}
void manager_set_show_status(Manager *m, ShowStatus mode) {
assert(m);
assert(IN_SET(mode, SHOW_STATUS_AUTO, SHOW_STATUS_NO, SHOW_STATUS_YES, SHOW_STATUS_TEMPORARY));
if (!MANAGER_IS_SYSTEM(m))
return;
if (m->show_status != mode)
log_debug("%s showing of status.",
mode == SHOW_STATUS_NO ? "Disabling" : "Enabling");
m->show_status = mode;
if (mode > 0)
(void) touch("/run/systemd/show-status");
else
(void) unlink("/run/systemd/show-status");
}
static bool manager_get_show_status(Manager *m, StatusType type) {
assert(m);
if (!MANAGER_IS_SYSTEM(m))
return false;
if (m->no_console_output)
return false;
if (!IN_SET(manager_state(m), MANAGER_INITIALIZING, MANAGER_STARTING, MANAGER_STOPPING))
return false;
/* If we cannot find out the status properly, just proceed. */
if (type != STATUS_TYPE_EMERGENCY && manager_check_ask_password(m) > 0)
return false;
return m->show_status > 0;
}
const char *manager_get_confirm_spawn(Manager *m) {
static int last_errno = 0;
const char *vc = m->confirm_spawn;
struct stat st;
int r;
/* Here's the deal: we want to test the validity of the console but don't want
* PID1 to go through the whole console process which might block. But we also
* want to warn the user only once if something is wrong with the console so we
* cannot do the sanity checks after spawning our children. So here we simply do
* really basic tests to hopefully trap common errors.
*
* If the console suddenly disappear at the time our children will really it
* then they will simply fail to acquire it and a positive answer will be
* assumed. New children will fallback to /dev/console though.
*
* Note: TTYs are devices that can come and go any time, and frequently aren't
* available yet during early boot (consider a USB rs232 dongle...). If for any
* reason the configured console is not ready, we fallback to the default
* console. */
if (!vc || path_equal(vc, "/dev/console"))
return vc;
r = stat(vc, &st);
if (r < 0)
goto fail;
if (!S_ISCHR(st.st_mode)) {
errno = ENOTTY;
goto fail;
}
last_errno = 0;
return vc;
fail:
if (last_errno != errno) {
last_errno = errno;
log_warning_errno(errno, "Failed to open %s: %m, using default console", vc);
}
return "/dev/console";
}
void manager_set_first_boot(Manager *m, bool b) {
assert(m);
if (!MANAGER_IS_SYSTEM(m))
return;
if (m->first_boot != (int) b) {
if (b)
(void) touch("/run/systemd/first-boot");
else
(void) unlink("/run/systemd/first-boot");
}
m->first_boot = b;
}
void manager_disable_confirm_spawn(void) {
(void) touch("/run/systemd/confirm_spawn_disabled");
}
bool manager_is_confirm_spawn_disabled(Manager *m) {
if (!m->confirm_spawn)
return true;
return access("/run/systemd/confirm_spawn_disabled", F_OK) >= 0;
}
void manager_status_printf(Manager *m, StatusType type, const char *status, const char *format, ...) {
va_list ap;
/* If m is NULL, assume we're after shutdown and let the messages through. */
if (m && !manager_get_show_status(m, type))
return;
/* XXX We should totally drop the check for ephemeral here
* and thus effectively make 'Type=idle' pointless. */
if (type == STATUS_TYPE_EPHEMERAL && m && m->n_on_console > 0)
return;
va_start(ap, format);
status_vprintf(status, true, type == STATUS_TYPE_EPHEMERAL, format, ap);
va_end(ap);
}
Set *manager_get_units_requiring_mounts_for(Manager *m, const char *path) {
char p[strlen(path)+1];
assert(m);
assert(path);
strcpy(p, path);
path_kill_slashes(p);
return hashmap_get(m->units_requiring_mounts_for, streq(p, "/") ? "" : p);
}
int manager_update_failed_units(Manager *m, Unit *u, bool failed) {
unsigned size;
int r;
assert(m);
assert(u->manager == m);
size = set_size(m->failed_units);
if (failed) {
r = set_ensure_allocated(&m->failed_units, NULL);
if (r < 0)
return log_oom();
if (set_put(m->failed_units, u) < 0)
return log_oom();
} else
(void) set_remove(m->failed_units, u);
if (set_size(m->failed_units) != size)
bus_manager_send_change_signal(m);
return 0;
}
ManagerState manager_state(Manager *m) {
Unit *u;
assert(m);
/* Did we ever finish booting? If not then we are still starting up */
if (!MANAGER_IS_FINISHED(m)) {
u = manager_get_unit(m, SPECIAL_BASIC_TARGET);
if (!u || !UNIT_IS_ACTIVE_OR_RELOADING(unit_active_state(u)))
return MANAGER_INITIALIZING;
return MANAGER_STARTING;
}
/* Is the special shutdown target active or queued? If so, we are in shutdown state */
u = manager_get_unit(m, SPECIAL_SHUTDOWN_TARGET);
if (u && unit_active_or_pending(u))
return MANAGER_STOPPING;
if (MANAGER_IS_SYSTEM(m)) {
/* Are the rescue or emergency targets active or queued? If so we are in maintenance state */
u = manager_get_unit(m, SPECIAL_RESCUE_TARGET);
if (u && unit_active_or_pending(u))
return MANAGER_MAINTENANCE;
u = manager_get_unit(m, SPECIAL_EMERGENCY_TARGET);
if (u && unit_active_or_pending(u))
return MANAGER_MAINTENANCE;
}
/* Are there any failed units? If so, we are in degraded mode */
if (set_size(m->failed_units) > 0)
return MANAGER_DEGRADED;
return MANAGER_RUNNING;
}
#define DESTROY_IPC_FLAG (UINT32_C(1) << 31)
static void manager_unref_uid_internal(
Manager *m,
Hashmap **uid_refs,
uid_t uid,
bool destroy_now,
int (*_clean_ipc)(uid_t uid)) {
uint32_t c, n;
assert(m);
assert(uid_refs);
assert(uid_is_valid(uid));
assert(_clean_ipc);
/* A generic implementation, covering both manager_unref_uid() and manager_unref_gid(), under the assumption
* that uid_t and gid_t are actually defined the same way, with the same validity rules.
*
* We store a hashmap where the UID/GID is they key and the value is a 32bit reference counter, whose highest
* bit is used as flag for marking UIDs/GIDs whose IPC objects to remove when the last reference to the UID/GID
* is dropped. The flag is set to on, once at least one reference from a unit where RemoveIPC= is set is added
* on a UID/GID. It is reset when the UID's/GID's reference counter drops to 0 again. */
assert_cc(sizeof(uid_t) == sizeof(gid_t));
assert_cc(UID_INVALID == (uid_t) GID_INVALID);
if (uid == 0) /* We don't keep track of root, and will never destroy it */
return;
c = PTR_TO_UINT32(hashmap_get(*uid_refs, UID_TO_PTR(uid)));
n = c & ~DESTROY_IPC_FLAG;
assert(n > 0);
n--;
if (destroy_now && n == 0) {
hashmap_remove(*uid_refs, UID_TO_PTR(uid));
if (c & DESTROY_IPC_FLAG) {
log_debug("%s " UID_FMT " is no longer referenced, cleaning up its IPC.",
_clean_ipc == clean_ipc_by_uid ? "UID" : "GID",
uid);
(void) _clean_ipc(uid);
}
} else {
c = n | (c & DESTROY_IPC_FLAG);
assert_se(hashmap_update(*uid_refs, UID_TO_PTR(uid), UINT32_TO_PTR(c)) >= 0);
}
}
void manager_unref_uid(Manager *m, uid_t uid, bool destroy_now) {
manager_unref_uid_internal(m, &m->uid_refs, uid, destroy_now, clean_ipc_by_uid);
}
void manager_unref_gid(Manager *m, gid_t gid, bool destroy_now) {
manager_unref_uid_internal(m, &m->gid_refs, (uid_t) gid, destroy_now, clean_ipc_by_gid);
}
static int manager_ref_uid_internal(
Manager *m,
Hashmap **uid_refs,
uid_t uid,
bool clean_ipc) {
uint32_t c, n;
int r;
assert(m);
assert(uid_refs);
assert(uid_is_valid(uid));
/* A generic implementation, covering both manager_ref_uid() and manager_ref_gid(), under the assumption
* that uid_t and gid_t are actually defined the same way, with the same validity rules. */
assert_cc(sizeof(uid_t) == sizeof(gid_t));
assert_cc(UID_INVALID == (uid_t) GID_INVALID);
if (uid == 0) /* We don't keep track of root, and will never destroy it */
return 0;
r = hashmap_ensure_allocated(uid_refs, &trivial_hash_ops);
if (r < 0)
return r;
c = PTR_TO_UINT32(hashmap_get(*uid_refs, UID_TO_PTR(uid)));
n = c & ~DESTROY_IPC_FLAG;
n++;
if (n & DESTROY_IPC_FLAG) /* check for overflow */
return -EOVERFLOW;
c = n | (c & DESTROY_IPC_FLAG) | (clean_ipc ? DESTROY_IPC_FLAG : 0);
return hashmap_replace(*uid_refs, UID_TO_PTR(uid), UINT32_TO_PTR(c));
}
int manager_ref_uid(Manager *m, uid_t uid, bool clean_ipc) {
return manager_ref_uid_internal(m, &m->uid_refs, uid, clean_ipc);
}
int manager_ref_gid(Manager *m, gid_t gid, bool clean_ipc) {
return manager_ref_uid_internal(m, &m->gid_refs, (uid_t) gid, clean_ipc);
}
static void manager_vacuum_uid_refs_internal(
Manager *m,
Hashmap **uid_refs,
int (*_clean_ipc)(uid_t uid)) {
Iterator i;
void *p, *k;
assert(m);
assert(uid_refs);
assert(_clean_ipc);
HASHMAP_FOREACH_KEY(p, k, *uid_refs, i) {
uint32_t c, n;
uid_t uid;
uid = PTR_TO_UID(k);
c = PTR_TO_UINT32(p);
n = c & ~DESTROY_IPC_FLAG;
if (n > 0)
continue;
if (c & DESTROY_IPC_FLAG) {
log_debug("Found unreferenced %s " UID_FMT " after reload/reexec. Cleaning up.",
_clean_ipc == clean_ipc_by_uid ? "UID" : "GID",
uid);
(void) _clean_ipc(uid);
}
assert_se(hashmap_remove(*uid_refs, k) == p);
}
}
void manager_vacuum_uid_refs(Manager *m) {
manager_vacuum_uid_refs_internal(m, &m->uid_refs, clean_ipc_by_uid);
}
void manager_vacuum_gid_refs(Manager *m) {
manager_vacuum_uid_refs_internal(m, &m->gid_refs, clean_ipc_by_gid);
}
static void manager_serialize_uid_refs_internal(
Manager *m,
FILE *f,
Hashmap **uid_refs,
const char *field_name) {
Iterator i;
void *p, *k;
assert(m);
assert(f);
assert(uid_refs);
assert(field_name);
/* Serialize the UID reference table. Or actually, just the IPC destruction flag of it, as the actual counter
* of it is better rebuild after a reload/reexec. */
HASHMAP_FOREACH_KEY(p, k, *uid_refs, i) {
uint32_t c;
uid_t uid;
uid = PTR_TO_UID(k);
c = PTR_TO_UINT32(p);
if (!(c & DESTROY_IPC_FLAG))
continue;
fprintf(f, "%s=" UID_FMT "\n", field_name, uid);
}
}
void manager_serialize_uid_refs(Manager *m, FILE *f) {
manager_serialize_uid_refs_internal(m, f, &m->uid_refs, "destroy-ipc-uid");
}
void manager_serialize_gid_refs(Manager *m, FILE *f) {
manager_serialize_uid_refs_internal(m, f, &m->gid_refs, "destroy-ipc-gid");
}
static void manager_deserialize_uid_refs_one_internal(
Manager *m,
Hashmap** uid_refs,
const char *value) {
uid_t uid;
uint32_t c;
int r;
assert(m);
assert(uid_refs);
assert(value);
r = parse_uid(value, &uid);
if (r < 0 || uid == 0) {
log_debug("Unable to parse UID reference serialization");
return;
}
r = hashmap_ensure_allocated(uid_refs, &trivial_hash_ops);
if (r < 0) {
log_oom();
return;
}
c = PTR_TO_UINT32(hashmap_get(*uid_refs, UID_TO_PTR(uid)));
if (c & DESTROY_IPC_FLAG)
return;
c |= DESTROY_IPC_FLAG;
r = hashmap_replace(*uid_refs, UID_TO_PTR(uid), UINT32_TO_PTR(c));
if (r < 0) {
log_debug("Failed to add UID reference entry");
return;
}
}
void manager_deserialize_uid_refs_one(Manager *m, const char *value) {
manager_deserialize_uid_refs_one_internal(m, &m->uid_refs, value);
}
void manager_deserialize_gid_refs_one(Manager *m, const char *value) {
manager_deserialize_uid_refs_one_internal(m, &m->gid_refs, value);
}
int manager_dispatch_user_lookup_fd(sd_event_source *source, int fd, uint32_t revents, void *userdata) {
struct buffer {
uid_t uid;
gid_t gid;
char unit_name[UNIT_NAME_MAX+1];
} _packed_ buffer;
Manager *m = userdata;
ssize_t l;
size_t n;
Unit *u;
assert_se(source);
assert_se(m);
/* Invoked whenever a child process succeeded resolving its user/group to use and sent us the resulting UID/GID
* in a datagram. We parse the datagram here and pass it off to the unit, so that it can add a reference to the
* UID/GID so that it can destroy the UID/GID's IPC objects when the reference counter drops to 0. */
l = recv(fd, &buffer, sizeof(buffer), MSG_DONTWAIT);
if (l < 0) {
if (IN_SET(errno, EINTR, EAGAIN))
return 0;
return log_error_errno(errno, "Failed to read from user lookup fd: %m");
}
if ((size_t) l <= offsetof(struct buffer, unit_name)) {
log_warning("Received too short user lookup message, ignoring.");
return 0;
}
if ((size_t) l > offsetof(struct buffer, unit_name) + UNIT_NAME_MAX) {
log_warning("Received too long user lookup message, ignoring.");
return 0;
}
if (!uid_is_valid(buffer.uid) && !gid_is_valid(buffer.gid)) {
log_warning("Got user lookup message with invalid UID/GID pair, ignoring.");
return 0;
}
n = (size_t) l - offsetof(struct buffer, unit_name);
if (memchr(buffer.unit_name, 0, n)) {
log_warning("Received lookup message with embedded NUL character, ignoring.");
return 0;
}
buffer.unit_name[n] = 0;
u = manager_get_unit(m, buffer.unit_name);
if (!u) {
log_debug("Got user lookup message but unit doesn't exist, ignoring.");
return 0;
}
log_unit_debug(u, "User lookup succeeded: uid=" UID_FMT " gid=" GID_FMT, buffer.uid, buffer.gid);
unit_notify_user_lookup(u, buffer.uid, buffer.gid);
return 0;
}
char *manager_taint_string(Manager *m) {
_cleanup_free_ char *destination = NULL, *overflowuid = NULL, *overflowgid = NULL;
char *buf, *e;
int r;
/* Returns a "taint string", e.g. "local-hwclock:var-run-bad".
* Only things that are detected at runtime should be tagged
* here. For stuff that is set during compilation, emit a warning
* in the configuration phase. */
assert(m);
buf = new(char, sizeof("split-usr:"
"cgroups-missing:"
"local-hwclock:"
"var-run-bad:"
"overflowuid-not-65534:"
"overflowgid-not-65534:"));
if (!buf)
return NULL;
e = buf;
buf[0] = 0;
if (m->taint_usr)
e = stpcpy(e, "split-usr:");
if (access("/proc/cgroups", F_OK) < 0)
e = stpcpy(e, "cgroups-missing:");
if (clock_is_localtime(NULL) > 0)
e = stpcpy(e, "local-hwclock:");
r = readlink_malloc("/var/run", &destination);
if (r < 0 || !PATH_IN_SET(destination, "../run", "/run"))
e = stpcpy(e, "var-run-bad:");
r = read_one_line_file("/proc/sys/kernel/overflowuid", &overflowuid);
if (r >= 0 && !streq(overflowuid, "65534"))
e = stpcpy(e, "overflowuid-not-65534:");
r = read_one_line_file("/proc/sys/kernel/overflowgid", &overflowgid);
if (r >= 0 && !streq(overflowgid, "65534"))
e = stpcpy(e, "overflowgid-not-65534:");
/* remove the last ':' */
if (e != buf)
e[-1] = 0;
return buf;
}
void manager_ref_console(Manager *m) {
assert(m);
m->n_on_console++;
}
void manager_unref_console(Manager *m) {
assert(m->n_on_console > 0);
m->n_on_console--;
if (m->n_on_console == 0)
m->no_console_output = false; /* unset no_console_output flag, since the console is definitely free now */
}
static const char *const manager_state_table[_MANAGER_STATE_MAX] = {
[MANAGER_INITIALIZING] = "initializing",
[MANAGER_STARTING] = "starting",
[MANAGER_RUNNING] = "running",
[MANAGER_DEGRADED] = "degraded",
[MANAGER_MAINTENANCE] = "maintenance",
[MANAGER_STOPPING] = "stopping",
};
DEFINE_STRING_TABLE_LOOKUP(manager_state, ManagerState);
static const char *const manager_timestamp_table[_MANAGER_TIMESTAMP_MAX] = {
[MANAGER_TIMESTAMP_FIRMWARE] = "firmware",
[MANAGER_TIMESTAMP_LOADER] = "loader",
[MANAGER_TIMESTAMP_KERNEL] = "kernel",
[MANAGER_TIMESTAMP_INITRD] = "initrd",
[MANAGER_TIMESTAMP_USERSPACE] = "userspace",
[MANAGER_TIMESTAMP_FINISH] = "finish",
[MANAGER_TIMESTAMP_SECURITY_START] = "security-start",
[MANAGER_TIMESTAMP_SECURITY_FINISH] = "security-finish",
[MANAGER_TIMESTAMP_GENERATORS_START] = "generators-start",
[MANAGER_TIMESTAMP_GENERATORS_FINISH] = "generators-finish",
[MANAGER_TIMESTAMP_UNITS_LOAD_START] = "units-load-start",
[MANAGER_TIMESTAMP_UNITS_LOAD_FINISH] = "units-load-finish",
};
DEFINE_STRING_TABLE_LOOKUP(manager_timestamp, ManagerTimestamp);
|