summaryrefslogtreecommitdiff
path: root/drivers/char/ipmi/ipmi_si_intf.c
blob: c7cc8538b84aec726b3483197636e4e38e3ff4ae (plain)
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
// SPDX-License-Identifier: GPL-2.0+
/*
 * ipmi_si.c
 *
 * The interface to the IPMI driver for the system interfaces (KCS, SMIC,
 * BT).
 *
 * Author: MontaVista Software, Inc.
 *         Corey Minyard <minyard@mvista.com>
 *         source@mvista.com
 *
 * Copyright 2002 MontaVista Software Inc.
 * Copyright 2006 IBM Corp., Christian Krafft <krafft@de.ibm.com>
 */

/*
 * This file holds the "policy" for the interface to the SMI state
 * machine.  It does the configuration, handles timers and interrupts,
 * and drives the real SMI state machine.
 */

#define pr_fmt(fmt) "ipmi_si: " fmt

#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/sched.h>
#include <linux/seq_file.h>
#include <linux/timer.h>
#include <linux/errno.h>
#include <linux/spinlock.h>
#include <linux/slab.h>
#include <linux/delay.h>
#include <linux/list.h>
#include <linux/notifier.h>
#include <linux/mutex.h>
#include <linux/kthread.h>
#include <asm/irq.h>
#include <linux/interrupt.h>
#include <linux/rcupdate.h>
#include <linux/ipmi.h>
#include <linux/ipmi_smi.h>
#include "ipmi_si.h"
#include "ipmi_si_sm.h"
#include <linux/string.h>
#include <linux/ctype.h>

/* Measure times between events in the driver. */
#undef DEBUG_TIMING

/* Call every 10 ms. */
#define SI_TIMEOUT_TIME_USEC	10000
#define SI_USEC_PER_JIFFY	(1000000/HZ)
#define SI_TIMEOUT_JIFFIES	(SI_TIMEOUT_TIME_USEC/SI_USEC_PER_JIFFY)
#define SI_SHORT_TIMEOUT_USEC  250 /* .25ms when the SM request a
				      short timeout */

enum si_intf_state {
	SI_NORMAL,
	SI_GETTING_FLAGS,
	SI_GETTING_EVENTS,
	SI_CLEARING_FLAGS,
	SI_GETTING_MESSAGES,
	SI_CHECKING_ENABLES,
	SI_SETTING_ENABLES
	/* FIXME - add watchdog stuff. */
};

/* Some BT-specific defines we need here. */
#define IPMI_BT_INTMASK_REG		2
#define IPMI_BT_INTMASK_CLEAR_IRQ_BIT	2
#define IPMI_BT_INTMASK_ENABLE_IRQ_BIT	1

static const char * const si_to_str[] = { "invalid", "kcs", "smic", "bt" };

static bool initialized;

/*
 * Indexes into stats[] in smi_info below.
 */
enum si_stat_indexes {
	/*
	 * Number of times the driver requested a timer while an operation
	 * was in progress.
	 */
	SI_STAT_short_timeouts = 0,

	/*
	 * Number of times the driver requested a timer while nothing was in
	 * progress.
	 */
	SI_STAT_long_timeouts,

	/* Number of times the interface was idle while being polled. */
	SI_STAT_idles,

	/* Number of interrupts the driver handled. */
	SI_STAT_interrupts,

	/* Number of time the driver got an ATTN from the hardware. */
	SI_STAT_attentions,

	/* Number of times the driver requested flags from the hardware. */
	SI_STAT_flag_fetches,

	/* Number of times the hardware didn't follow the state machine. */
	SI_STAT_hosed_count,

	/* Number of completed messages. */
	SI_STAT_complete_transactions,

	/* Number of IPMI events received from the hardware. */
	SI_STAT_events,

	/* Number of watchdog pretimeouts. */
	SI_STAT_watchdog_pretimeouts,

	/* Number of asynchronous messages received. */
	SI_STAT_incoming_messages,


	/* This *must* remain last, add new values above this. */
	SI_NUM_STATS
};

struct smi_info {
	int                    si_num;
	struct ipmi_smi        *intf;
	struct si_sm_data      *si_sm;
	const struct si_sm_handlers *handlers;
	spinlock_t             si_lock;
	struct ipmi_smi_msg    *waiting_msg;
	struct ipmi_smi_msg    *curr_msg;
	enum si_intf_state     si_state;

	/*
	 * Used to handle the various types of I/O that can occur with
	 * IPMI
	 */
	struct si_sm_io io;

	/*
	 * Per-OEM handler, called from handle_flags().  Returns 1
	 * when handle_flags() needs to be re-run or 0 indicating it
	 * set si_state itself.
	 */
	int (*oem_data_avail_handler)(struct smi_info *smi_info);

	/*
	 * Flags from the last GET_MSG_FLAGS command, used when an ATTN
	 * is set to hold the flags until we are done handling everything
	 * from the flags.
	 */
#define RECEIVE_MSG_AVAIL	0x01
#define EVENT_MSG_BUFFER_FULL	0x02
#define WDT_PRE_TIMEOUT_INT	0x08
#define OEM0_DATA_AVAIL     0x20
#define OEM1_DATA_AVAIL     0x40
#define OEM2_DATA_AVAIL     0x80
#define OEM_DATA_AVAIL      (OEM0_DATA_AVAIL | \
			     OEM1_DATA_AVAIL | \
			     OEM2_DATA_AVAIL)
	unsigned char       msg_flags;

	/* Does the BMC have an event buffer? */
	bool		    has_event_buffer;

	/*
	 * If set to true, this will request events the next time the
	 * state machine is idle.
	 */
	atomic_t            req_events;

	/*
	 * If true, run the state machine to completion on every send
	 * call.  Generally used after a panic to make sure stuff goes
	 * out.
	 */
	bool                run_to_completion;

	/* The timer for this si. */
	struct timer_list   si_timer;

	/* This flag is set, if the timer can be set */
	bool		    timer_can_start;

	/* This flag is set, if the timer is running (timer_pending() isn't enough) */
	bool		    timer_running;

	/* The time (in jiffies) the last timeout occurred at. */
	unsigned long       last_timeout_jiffies;

	/* Are we waiting for the events, pretimeouts, received msgs? */
	atomic_t            need_watch;

	/*
	 * The driver will disable interrupts when it gets into a
	 * situation where it cannot handle messages due to lack of
	 * memory.  Once that situation clears up, it will re-enable
	 * interrupts.
	 */
	bool interrupt_disabled;

	/*
	 * Does the BMC support events?
	 */
	bool supports_event_msg_buff;

	/*
	 * Can we disable interrupts the global enables receive irq
	 * bit?  There are currently two forms of brokenness, some
	 * systems cannot disable the bit (which is technically within
	 * the spec but a bad idea) and some systems have the bit
	 * forced to zero even though interrupts work (which is
	 * clearly outside the spec).  The next bool tells which form
	 * of brokenness is present.
	 */
	bool cannot_disable_irq;

	/*
	 * Some systems are broken and cannot set the irq enable
	 * bit, even if they support interrupts.
	 */
	bool irq_enable_broken;

	/* Is the driver in maintenance mode? */
	bool in_maintenance_mode;

	/*
	 * Did we get an attention that we did not handle?
	 */
	bool got_attn;

	/* From the get device id response... */
	struct ipmi_device_id device_id;

	/* Have we added the device group to the device? */
	bool dev_group_added;

	/* Counters and things for the proc filesystem. */
	atomic_t stats[SI_NUM_STATS];

	struct task_struct *thread;

	struct list_head link;
};

#define smi_inc_stat(smi, stat) \
	atomic_inc(&(smi)->stats[SI_STAT_ ## stat])
#define smi_get_stat(smi, stat) \
	((unsigned int) atomic_read(&(smi)->stats[SI_STAT_ ## stat]))

#define IPMI_MAX_INTFS 4
static int force_kipmid[IPMI_MAX_INTFS];
static int num_force_kipmid;

static unsigned int kipmid_max_busy_us[IPMI_MAX_INTFS];
static int num_max_busy_us;

static bool unload_when_empty = true;

static int try_smi_init(struct smi_info *smi);
static void cleanup_one_si(struct smi_info *smi_info);
static void cleanup_ipmi_si(void);

#ifdef DEBUG_TIMING
void debug_timestamp(char *msg)
{
	struct timespec64 t;

	ktime_get_ts64(&t);
	pr_debug("**%s: %lld.%9.9ld\n", msg, t.tv_sec, t.tv_nsec);
}
#else
#define debug_timestamp(x)
#endif

static ATOMIC_NOTIFIER_HEAD(xaction_notifier_list);
static int register_xaction_notifier(struct notifier_block *nb)
{
	return atomic_notifier_chain_register(&xaction_notifier_list, nb);
}

static void deliver_recv_msg(struct smi_info *smi_info,
			     struct ipmi_smi_msg *msg)
{
	/* Deliver the message to the upper layer. */
	ipmi_smi_msg_received(smi_info->intf, msg);
}

static void return_hosed_msg(struct smi_info *smi_info, int cCode)
{
	struct ipmi_smi_msg *msg = smi_info->curr_msg;

	if (cCode < 0 || cCode > IPMI_ERR_UNSPECIFIED)
		cCode = IPMI_ERR_UNSPECIFIED;
	/* else use it as is */

	/* Make it a response */
	msg->rsp[0] = msg->data[0] | 4;
	msg->rsp[1] = msg->data[1];
	msg->rsp[2] = cCode;
	msg->rsp_size = 3;

	smi_info->curr_msg = NULL;
	deliver_recv_msg(smi_info, msg);
}

static enum si_sm_result start_next_msg(struct smi_info *smi_info)
{
	int              rv;

	if (!smi_info->waiting_msg) {
		smi_info->curr_msg = NULL;
		rv = SI_SM_IDLE;
	} else {
		int err;

		smi_info->curr_msg = smi_info->waiting_msg;
		smi_info->waiting_msg = NULL;
		debug_timestamp("Start2");
		err = atomic_notifier_call_chain(&xaction_notifier_list,
				0, smi_info);
		if (err & NOTIFY_STOP_MASK) {
			rv = SI_SM_CALL_WITHOUT_DELAY;
			goto out;
		}
		err = smi_info->handlers->start_transaction(
			smi_info->si_sm,
			smi_info->curr_msg->data,
			smi_info->curr_msg->data_size);
		if (err)
			return_hosed_msg(smi_info, err);

		rv = SI_SM_CALL_WITHOUT_DELAY;
	}
out:
	return rv;
}

static void smi_mod_timer(struct smi_info *smi_info, unsigned long new_val)
{
	if (!smi_info->timer_can_start)
		return;
	smi_info->last_timeout_jiffies = jiffies;
	mod_timer(&smi_info->si_timer, new_val);
	smi_info->timer_running = true;
}

/*
 * Start a new message and (re)start the timer and thread.
 */
static void start_new_msg(struct smi_info *smi_info, unsigned char *msg,
			  unsigned int size)
{
	smi_mod_timer(smi_info, jiffies + SI_TIMEOUT_JIFFIES);

	if (smi_info->thread)
		wake_up_process(smi_info->thread);

	smi_info->handlers->start_transaction(smi_info->si_sm, msg, size);
}

static void start_check_enables(struct smi_info *smi_info)
{
	unsigned char msg[2];

	msg[0] = (IPMI_NETFN_APP_REQUEST << 2);
	msg[1] = IPMI_GET_BMC_GLOBAL_ENABLES_CMD;

	start_new_msg(smi_info, msg, 2);
	smi_info->si_state = SI_CHECKING_ENABLES;
}

static void start_clear_flags(struct smi_info *smi_info)
{
	unsigned char msg[3];

	/* Make sure the watchdog pre-timeout flag is not set at startup. */
	msg[0] = (IPMI_NETFN_APP_REQUEST << 2);
	msg[1] = IPMI_CLEAR_MSG_FLAGS_CMD;
	msg[2] = WDT_PRE_TIMEOUT_INT;

	start_new_msg(smi_info, msg, 3);
	smi_info->si_state = SI_CLEARING_FLAGS;
}

static void start_getting_msg_queue(struct smi_info *smi_info)
{
	smi_info->curr_msg->data[0] = (IPMI_NETFN_APP_REQUEST << 2);
	smi_info->curr_msg->data[1] = IPMI_GET_MSG_CMD;
	smi_info->curr_msg->data_size = 2;

	start_new_msg(smi_info, smi_info->curr_msg->data,
		      smi_info->curr_msg->data_size);
	smi_info->si_state = SI_GETTING_MESSAGES;
}

static void start_getting_events(struct smi_info *smi_info)
{
	smi_info->curr_msg->data[0] = (IPMI_NETFN_APP_REQUEST << 2);
	smi_info->curr_msg->data[1] = IPMI_READ_EVENT_MSG_BUFFER_CMD;
	smi_info->curr_msg->data_size = 2;

	start_new_msg(smi_info, smi_info->curr_msg->data,
		      smi_info->curr_msg->data_size);
	smi_info->si_state = SI_GETTING_EVENTS;
}

/*
 * When we have a situtaion where we run out of memory and cannot
 * allocate messages, we just leave them in the BMC and run the system
 * polled until we can allocate some memory.  Once we have some
 * memory, we will re-enable the interrupt.
 *
 * Note that we cannot just use disable_irq(), since the interrupt may
 * be shared.
 */
static inline bool disable_si_irq(struct smi_info *smi_info)
{
	if ((smi_info->io.irq) && (!smi_info->interrupt_disabled)) {
		smi_info->interrupt_disabled = true;
		start_check_enables(smi_info);
		return true;
	}
	return false;
}

static inline bool enable_si_irq(struct smi_info *smi_info)
{
	if ((smi_info->io.irq) && (smi_info->interrupt_disabled)) {
		smi_info->interrupt_disabled = false;
		start_check_enables(smi_info);
		return true;
	}
	return false;
}

/*
 * Allocate a message.  If unable to allocate, start the interrupt
 * disable process and return NULL.  If able to allocate but
 * interrupts are disabled, free the message and return NULL after
 * starting the interrupt enable process.
 */
static struct ipmi_smi_msg *alloc_msg_handle_irq(struct smi_info *smi_info)
{
	struct ipmi_smi_msg *msg;

	msg = ipmi_alloc_smi_msg();
	if (!msg) {
		if (!disable_si_irq(smi_info))
			smi_info->si_state = SI_NORMAL;
	} else if (enable_si_irq(smi_info)) {
		ipmi_free_smi_msg(msg);
		msg = NULL;
	}
	return msg;
}

static void handle_flags(struct smi_info *smi_info)
{
retry:
	if (smi_info->msg_flags & WDT_PRE_TIMEOUT_INT) {
		/* Watchdog pre-timeout */
		smi_inc_stat(smi_info, watchdog_pretimeouts);

		start_clear_flags(smi_info);
		smi_info->msg_flags &= ~WDT_PRE_TIMEOUT_INT;
		ipmi_smi_watchdog_pretimeout(smi_info->intf);
	} else if (smi_info->msg_flags & RECEIVE_MSG_AVAIL) {
		/* Messages available. */
		smi_info->curr_msg = alloc_msg_handle_irq(smi_info);
		if (!smi_info->curr_msg)
			return;

		start_getting_msg_queue(smi_info);
	} else if (smi_info->msg_flags & EVENT_MSG_BUFFER_FULL) {
		/* Events available. */
		smi_info->curr_msg = alloc_msg_handle_irq(smi_info);
		if (!smi_info->curr_msg)
			return;

		start_getting_events(smi_info);
	} else if (smi_info->msg_flags & OEM_DATA_AVAIL &&
		   smi_info->oem_data_avail_handler) {
		if (smi_info->oem_data_avail_handler(smi_info))
			goto retry;
	} else
		smi_info->si_state = SI_NORMAL;
}

/*
 * Global enables we care about.
 */
#define GLOBAL_ENABLES_MASK (IPMI_BMC_EVT_MSG_BUFF | IPMI_BMC_RCV_MSG_INTR | \
			     IPMI_BMC_EVT_MSG_INTR)

static u8 current_global_enables(struct smi_info *smi_info, u8 base,
				 bool *irq_on)
{
	u8 enables = 0;

	if (smi_info->supports_event_msg_buff)
		enables |= IPMI_BMC_EVT_MSG_BUFF;

	if (((smi_info->io.irq && !smi_info->interrupt_disabled) ||
	     smi_info->cannot_disable_irq) &&
	    !smi_info->irq_enable_broken)
		enables |= IPMI_BMC_RCV_MSG_INTR;

	if (smi_info->supports_event_msg_buff &&
	    smi_info->io.irq && !smi_info->interrupt_disabled &&
	    !smi_info->irq_enable_broken)
		enables |= IPMI_BMC_EVT_MSG_INTR;

	*irq_on = enables & (IPMI_BMC_EVT_MSG_INTR | IPMI_BMC_RCV_MSG_INTR);

	return enables;
}

static void check_bt_irq(struct smi_info *smi_info, bool irq_on)
{
	u8 irqstate = smi_info->io.inputb(&smi_info->io, IPMI_BT_INTMASK_REG);

	irqstate &= IPMI_BT_INTMASK_ENABLE_IRQ_BIT;

	if ((bool)irqstate == irq_on)
		return;

	if (irq_on)
		smi_info->io.outputb(&smi_info->io, IPMI_BT_INTMASK_REG,
				     IPMI_BT_INTMASK_ENABLE_IRQ_BIT);
	else
		smi_info->io.outputb(&smi_info->io, IPMI_BT_INTMASK_REG, 0);
}

static void handle_transaction_done(struct smi_info *smi_info)
{
	struct ipmi_smi_msg *msg;

	debug_timestamp("Done");
	switch (smi_info->si_state) {
	case SI_NORMAL:
		if (!smi_info->curr_msg)
			break;

		smi_info->curr_msg->rsp_size
			= smi_info->handlers->get_result(
				smi_info->si_sm,
				smi_info->curr_msg->rsp,
				IPMI_MAX_MSG_LENGTH);

		/*
		 * Do this here becase deliver_recv_msg() releases the
		 * lock, and a new message can be put in during the
		 * time the lock is released.
		 */
		msg = smi_info->curr_msg;
		smi_info->curr_msg = NULL;
		deliver_recv_msg(smi_info, msg);
		break;

	case SI_GETTING_FLAGS:
	{
		unsigned char msg[4];
		unsigned int  len;

		/* We got the flags from the SMI, now handle them. */
		len = smi_info->handlers->get_result(smi_info->si_sm, msg, 4);
		if (msg[2] != 0) {
			/* Error fetching flags, just give up for now. */
			smi_info->si_state = SI_NORMAL;
		} else if (len < 4) {
			/*
			 * Hmm, no flags.  That's technically illegal, but
			 * don't use uninitialized data.
			 */
			smi_info->si_state = SI_NORMAL;
		} else {
			smi_info->msg_flags = msg[3];
			handle_flags(smi_info);
		}
		break;
	}

	case SI_CLEARING_FLAGS:
	{
		unsigned char msg[3];

		/* We cleared the flags. */
		smi_info->handlers->get_result(smi_info->si_sm, msg, 3);
		if (msg[2] != 0) {
			/* Error clearing flags */
			dev_warn(smi_info->io.dev,
				 "Error clearing flags: %2.2x\n", msg[2]);
		}
		smi_info->si_state = SI_NORMAL;
		break;
	}

	case SI_GETTING_EVENTS:
	{
		smi_info->curr_msg->rsp_size
			= smi_info->handlers->get_result(
				smi_info->si_sm,
				smi_info->curr_msg->rsp,
				IPMI_MAX_MSG_LENGTH);

		/*
		 * Do this here becase deliver_recv_msg() releases the
		 * lock, and a new message can be put in during the
		 * time the lock is released.
		 */
		msg = smi_info->curr_msg;
		smi_info->curr_msg = NULL;
		if (msg->rsp[2] != 0) {
			/* Error getting event, probably done. */
			msg->done(msg);

			/* Take off the event flag. */
			smi_info->msg_flags &= ~EVENT_MSG_BUFFER_FULL;
			handle_flags(smi_info);
		} else {
			smi_inc_stat(smi_info, events);

			/*
			 * Do this before we deliver the message
			 * because delivering the message releases the
			 * lock and something else can mess with the
			 * state.
			 */
			handle_flags(smi_info);

			deliver_recv_msg(smi_info, msg);
		}
		break;
	}

	case SI_GETTING_MESSAGES:
	{
		smi_info->curr_msg->rsp_size
			= smi_info->handlers->get_result(
				smi_info->si_sm,
				smi_info->curr_msg->rsp,
				IPMI_MAX_MSG_LENGTH);

		/*
		 * Do this here becase deliver_recv_msg() releases the
		 * lock, and a new message can be put in during the
		 * time the lock is released.
		 */
		msg = smi_info->curr_msg;
		smi_info->curr_msg = NULL;
		if (msg->rsp[2] != 0) {
			/* Error getting event, probably done. */
			msg->done(msg);

			/* Take off the msg flag. */
			smi_info->msg_flags &= ~RECEIVE_MSG_AVAIL;
			handle_flags(smi_info);
		} else {
			smi_inc_stat(smi_info, incoming_messages);

			/*
			 * Do this before we deliver the message
			 * because delivering the message releases the
			 * lock and something else can mess with the
			 * state.
			 */
			handle_flags(smi_info);

			deliver_recv_msg(smi_info, msg);
		}
		break;
	}

	case SI_CHECKING_ENABLES:
	{
		unsigned char msg[4];
		u8 enables;
		bool irq_on;

		/* We got the flags from the SMI, now handle them. */
		smi_info->handlers->get_result(smi_info->si_sm, msg, 4);
		if (msg[2] != 0) {
			dev_warn(smi_info->io.dev,
				 "Couldn't get irq info: %x.\n", msg[2]);
			dev_warn(smi_info->io.dev,
				 "Maybe ok, but ipmi might run very slowly.\n");
			smi_info->si_state = SI_NORMAL;
			break;
		}
		enables = current_global_enables(smi_info, 0, &irq_on);
		if (smi_info->io.si_type == SI_BT)
			/* BT has its own interrupt enable bit. */
			check_bt_irq(smi_info, irq_on);
		if (enables != (msg[3] & GLOBAL_ENABLES_MASK)) {
			/* Enables are not correct, fix them. */
			msg[0] = (IPMI_NETFN_APP_REQUEST << 2);
			msg[1] = IPMI_SET_BMC_GLOBAL_ENABLES_CMD;
			msg[2] = enables | (msg[3] & ~GLOBAL_ENABLES_MASK);
			smi_info->handlers->start_transaction(
				smi_info->si_sm, msg, 3);
			smi_info->si_state = SI_SETTING_ENABLES;
		} else if (smi_info->supports_event_msg_buff) {
			smi_info->curr_msg = ipmi_alloc_smi_msg();
			if (!smi_info->curr_msg) {
				smi_info->si_state = SI_NORMAL;
				break;
			}
			start_getting_events(smi_info);
		} else {
			smi_info->si_state = SI_NORMAL;
		}
		break;
	}

	case SI_SETTING_ENABLES:
	{
		unsigned char msg[4];

		smi_info->handlers->get_result(smi_info->si_sm, msg, 4);
		if (msg[2] != 0)
			dev_warn(smi_info->io.dev,
				 "Could not set the global enables: 0x%x.\n",
				 msg[2]);

		if (smi_info->supports_event_msg_buff) {
			smi_info->curr_msg = ipmi_alloc_smi_msg();
			if (!smi_info->curr_msg) {
				smi_info->si_state = SI_NORMAL;
				break;
			}
			start_getting_events(smi_info);
		} else {
			smi_info->si_state = SI_NORMAL;
		}
		break;
	}
	}
}

/*
 * Called on timeouts and events.  Timeouts should pass the elapsed
 * time, interrupts should pass in zero.  Must be called with
 * si_lock held and interrupts disabled.
 */
static enum si_sm_result smi_event_handler(struct smi_info *smi_info,
					   int time)
{
	enum si_sm_result si_sm_result;

restart:
	/*
	 * There used to be a loop here that waited a little while
	 * (around 25us) before giving up.  That turned out to be
	 * pointless, the minimum delays I was seeing were in the 300us
	 * range, which is far too long to wait in an interrupt.  So
	 * we just run until the state machine tells us something
	 * happened or it needs a delay.
	 */
	si_sm_result = smi_info->handlers->event(smi_info->si_sm, time);
	time = 0;
	while (si_sm_result == SI_SM_CALL_WITHOUT_DELAY)
		si_sm_result = smi_info->handlers->event(smi_info->si_sm, 0);

	if (si_sm_result == SI_SM_TRANSACTION_COMPLETE) {
		smi_inc_stat(smi_info, complete_transactions);

		handle_transaction_done(smi_info);
		goto restart;
	} else if (si_sm_result == SI_SM_HOSED) {
		smi_inc_stat(smi_info, hosed_count);

		/*
		 * Do the before return_hosed_msg, because that
		 * releases the lock.
		 */
		smi_info->si_state = SI_NORMAL;
		if (smi_info->curr_msg != NULL) {
			/*
			 * If we were handling a user message, format
			 * a response to send to the upper layer to
			 * tell it about the error.
			 */
			return_hosed_msg(smi_info, IPMI_ERR_UNSPECIFIED);
		}
		goto restart;
	}

	/*
	 * We prefer handling attn over new messages.  But don't do
	 * this if there is not yet an upper layer to handle anything.
	 */
	if (si_sm_result == SI_SM_ATTN || smi_info->got_attn) {
		unsigned char msg[2];

		if (smi_info->si_state != SI_NORMAL) {
			/*
			 * We got an ATTN, but we are doing something else.
			 * Handle the ATTN later.
			 */
			smi_info->got_attn = true;
		} else {
			smi_info->got_attn = false;
			smi_inc_stat(smi_info, attentions);

			/*
			 * Got a attn, send down a get message flags to see
			 * what's causing it.  It would be better to handle
			 * this in the upper layer, but due to the way
			 * interrupts work with the SMI, that's not really
			 * possible.
			 */
			msg[0] = (IPMI_NETFN_APP_REQUEST << 2);
			msg[1] = IPMI_GET_MSG_FLAGS_CMD;

			start_new_msg(smi_info, msg, 2);
			smi_info->si_state = SI_GETTING_FLAGS;
			goto restart;
		}
	}

	/* If we are currently idle, try to start the next message. */
	if (si_sm_result == SI_SM_IDLE) {
		smi_inc_stat(smi_info, idles);

		si_sm_result = start_next_msg(smi_info);
		if (si_sm_result != SI_SM_IDLE)
			goto restart;
	}

	if ((si_sm_result == SI_SM_IDLE)
	    && (atomic_read(&smi_info->req_events))) {
		/*
		 * We are idle and the upper layer requested that I fetch
		 * events, so do so.
		 */
		atomic_set(&smi_info->req_events, 0);

		/*
		 * Take this opportunity to check the interrupt and
		 * message enable state for the BMC.  The BMC can be
		 * asynchronously reset, and may thus get interrupts
		 * disable and messages disabled.
		 */
		if (smi_info->supports_event_msg_buff || smi_info->io.irq) {
			start_check_enables(smi_info);
		} else {
			smi_info->curr_msg = alloc_msg_handle_irq(smi_info);
			if (!smi_info->curr_msg)
				goto out;

			start_getting_events(smi_info);
		}
		goto restart;
	}

	if (si_sm_result == SI_SM_IDLE && smi_info->timer_running) {
		/* Ok it if fails, the timer will just go off. */
		if (del_timer(&smi_info->si_timer))
			smi_info->timer_running = false;
	}

out:
	return si_sm_result;
}

static void check_start_timer_thread(struct smi_info *smi_info)
{
	if (smi_info->si_state == SI_NORMAL && smi_info->curr_msg == NULL) {
		smi_mod_timer(smi_info, jiffies + SI_TIMEOUT_JIFFIES);

		if (smi_info->thread)
			wake_up_process(smi_info->thread);

		start_next_msg(smi_info);
		smi_event_handler(smi_info, 0);
	}
}

static void flush_messages(void *send_info)
{
	struct smi_info *smi_info = send_info;
	enum si_sm_result result;

	/*
	 * Currently, this function is called only in run-to-completion
	 * mode.  This means we are single-threaded, no need for locks.
	 */
	result = smi_event_handler(smi_info, 0);
	while (result != SI_SM_IDLE) {
		udelay(SI_SHORT_TIMEOUT_USEC);
		result = smi_event_handler(smi_info, SI_SHORT_TIMEOUT_USEC);
	}
}

static void sender(void                *send_info,
		   struct ipmi_smi_msg *msg)
{
	struct smi_info   *smi_info = send_info;
	unsigned long     flags;

	debug_timestamp("Enqueue");

	if (smi_info->run_to_completion) {
		/*
		 * If we are running to completion, start it.  Upper
		 * layer will call flush_messages to clear it out.
		 */
		smi_info->waiting_msg = msg;
		return;
	}

	spin_lock_irqsave(&smi_info->si_lock, flags);
	/*
	 * The following two lines don't need to be under the lock for
	 * the lock's sake, but they do need SMP memory barriers to
	 * avoid getting things out of order.  We are already claiming
	 * the lock, anyway, so just do it under the lock to avoid the
	 * ordering problem.
	 */
	BUG_ON(smi_info->waiting_msg);
	smi_info->waiting_msg = msg;
	check_start_timer_thread(smi_info);
	spin_unlock_irqrestore(&smi_info->si_lock, flags);
}

static void set_run_to_completion(void *send_info, bool i_run_to_completion)
{
	struct smi_info   *smi_info = send_info;

	smi_info->run_to_completion = i_run_to_completion;
	if (i_run_to_completion)
		flush_messages(smi_info);
}

/*
 * Use -1 as a special constant to tell that we are spinning in kipmid
 * looking for something and not delaying between checks
 */
#define IPMI_TIME_NOT_BUSY ns_to_ktime(-1ull)
static inline bool ipmi_thread_busy_wait(enum si_sm_result smi_result,
					 const struct smi_info *smi_info,
					 ktime_t *busy_until)
{
	unsigned int max_busy_us = 0;

	if (smi_info->si_num < num_max_busy_us)
		max_busy_us = kipmid_max_busy_us[smi_info->si_num];
	if (max_busy_us == 0 || smi_result != SI_SM_CALL_WITH_DELAY)
		*busy_until = IPMI_TIME_NOT_BUSY;
	else if (*busy_until == IPMI_TIME_NOT_BUSY) {
		*busy_until = ktime_get() + max_busy_us * NSEC_PER_USEC;
	} else {
		if (unlikely(ktime_get() > *busy_until)) {
			*busy_until = IPMI_TIME_NOT_BUSY;
			return false;
		}
	}
	return true;
}


/*
 * A busy-waiting loop for speeding up IPMI operation.
 *
 * Lousy hardware makes this hard.  This is only enabled for systems
 * that are not BT and do not have interrupts.  It starts spinning
 * when an operation is complete or until max_busy tells it to stop
 * (if that is enabled).  See the paragraph on kimid_max_busy_us in
 * Documentation/IPMI.txt for details.
 */
static int ipmi_thread(void *data)
{
	struct smi_info *smi_info = data;
	unsigned long flags;
	enum si_sm_result smi_result;
	ktime_t busy_until = IPMI_TIME_NOT_BUSY;

	set_user_nice(current, MAX_NICE);
	while (!kthread_should_stop()) {
		int busy_wait;

		spin_lock_irqsave(&(smi_info->si_lock), flags);
		smi_result = smi_event_handler(smi_info, 0);

		/*
		 * If the driver is doing something, there is a possible
		 * race with the timer.  If the timer handler see idle,
		 * and the thread here sees something else, the timer
		 * handler won't restart the timer even though it is
		 * required.  So start it here if necessary.
		 */
		if (smi_result != SI_SM_IDLE && !smi_info->timer_running)
			smi_mod_timer(smi_info, jiffies + SI_TIMEOUT_JIFFIES);

		spin_unlock_irqrestore(&(smi_info->si_lock), flags);
		busy_wait = ipmi_thread_busy_wait(smi_result, smi_info,
						  &busy_until);
		if (smi_result == SI_SM_CALL_WITHOUT_DELAY) {
			; /* do nothing */
		} else if (smi_result == SI_SM_CALL_WITH_DELAY && busy_wait) {
			/*
			 * In maintenance mode we run as fast as
			 * possible to allow firmware updates to
			 * complete as fast as possible, but normally
			 * don't bang on the scheduler.
			 */
			if (smi_info->in_maintenance_mode)
				schedule();
			else
				usleep_range(100, 200);
		} else if (smi_result == SI_SM_IDLE) {
			if (atomic_read(&smi_info->need_watch)) {
				schedule_timeout_interruptible(100);
			} else {
				/* Wait to be woken up when we are needed. */
				__set_current_state(TASK_INTERRUPTIBLE);
				schedule();
			}
		} else {
			schedule_timeout_interruptible(1);
		}
	}
	return 0;
}


static void poll(void *send_info)
{
	struct smi_info *smi_info = send_info;
	unsigned long flags = 0;
	bool run_to_completion = smi_info->run_to_completion;

	/*
	 * Make sure there is some delay in the poll loop so we can
	 * drive time forward and timeout things.
	 */
	udelay(10);
	if (!run_to_completion)
		spin_lock_irqsave(&smi_info->si_lock, flags);
	smi_event_handler(smi_info, 10);
	if (!run_to_completion)
		spin_unlock_irqrestore(&smi_info->si_lock, flags);
}

static void request_events(void *send_info)
{
	struct smi_info *smi_info = send_info;

	if (!smi_info->has_event_buffer)
		return;

	atomic_set(&smi_info->req_events, 1);
}

static void set_need_watch(void *send_info, unsigned int watch_mask)
{
	struct smi_info *smi_info = send_info;
	unsigned long flags;
	int enable;

	enable = !!watch_mask;

	atomic_set(&smi_info->need_watch, enable);
	spin_lock_irqsave(&smi_info->si_lock, flags);
	check_start_timer_thread(smi_info);
	spin_unlock_irqrestore(&smi_info->si_lock, flags);
}

static void smi_timeout(struct timer_list *t)
{
	struct smi_info   *smi_info = from_timer(smi_info, t, si_timer);
	enum si_sm_result smi_result;
	unsigned long     flags;
	unsigned long     jiffies_now;
	long              time_diff;
	long		  timeout;

	spin_lock_irqsave(&(smi_info->si_lock), flags);
	debug_timestamp("Timer");

	jiffies_now = jiffies;
	time_diff = (((long)jiffies_now - (long)smi_info->last_timeout_jiffies)
		     * SI_USEC_PER_JIFFY);
	smi_result = smi_event_handler(smi_info, time_diff);

	if ((smi_info->io.irq) && (!smi_info->interrupt_disabled)) {
		/* Running with interrupts, only do long timeouts. */
		timeout = jiffies + SI_TIMEOUT_JIFFIES;
		smi_inc_stat(smi_info, long_timeouts);
		goto do_mod_timer;
	}

	/*
	 * If the state machine asks for a short delay, then shorten
	 * the timer timeout.
	 */
	if (smi_result == SI_SM_CALL_WITH_DELAY) {
		smi_inc_stat(smi_info, short_timeouts);
		timeout = jiffies + 1;
	} else {
		smi_inc_stat(smi_info, long_timeouts);
		timeout = jiffies + SI_TIMEOUT_JIFFIES;
	}

do_mod_timer:
	if (smi_result != SI_SM_IDLE)
		smi_mod_timer(smi_info, timeout);
	else
		smi_info->timer_running = false;
	spin_unlock_irqrestore(&(smi_info->si_lock), flags);
}

irqreturn_t ipmi_si_irq_handler(int irq, void *data)
{
	struct smi_info *smi_info = data;
	unsigned long   flags;

	if (smi_info->io.si_type == SI_BT)
		/* We need to clear the IRQ flag for the BT interface. */
		smi_info->io.outputb(&smi_info->io, IPMI_BT_INTMASK_REG,
				     IPMI_BT_INTMASK_CLEAR_IRQ_BIT
				     | IPMI_BT_INTMASK_ENABLE_IRQ_BIT);

	spin_lock_irqsave(&(smi_info->si_lock), flags);

	smi_inc_stat(smi_info, interrupts);

	debug_timestamp("Interrupt");

	smi_event_handler(smi_info, 0);
	spin_unlock_irqrestore(&(smi_info->si_lock), flags);
	return IRQ_HANDLED;
}

static int smi_start_processing(void            *send_info,
				struct ipmi_smi *intf)
{
	struct smi_info *new_smi = send_info;
	int             enable = 0;

	new_smi->intf = intf;

	/* Set up the timer that drives the interface. */
	timer_setup(&new_smi->si_timer, smi_timeout, 0);
	new_smi->timer_can_start = true;
	smi_mod_timer(new_smi, jiffies + SI_TIMEOUT_JIFFIES);

	/* Try to claim any interrupts. */
	if (new_smi->io.irq_setup) {
		new_smi->io.irq_handler_data = new_smi;
		new_smi->io.irq_setup(&new_smi->io);
	}

	/*
	 * Check if the user forcefully enabled the daemon.
	 */
	if (new_smi->si_num < num_force_kipmid)
		enable = force_kipmid[new_smi->si_num];
	/*
	 * The BT interface is efficient enough to not need a thread,
	 * and there is no need for a thread if we have interrupts.
	 */
	else if ((new_smi->io.si_type != SI_BT) && (!new_smi->io.irq))
		enable = 1;

	if (enable) {
		new_smi->thread = kthread_run(ipmi_thread, new_smi,
					      "kipmi%d", new_smi->si_num);
		if (IS_ERR(new_smi->thread)) {
			dev_notice(new_smi->io.dev, "Could not start"
				   " kernel thread due to error %ld, only using"
				   " timers to drive the interface\n",
				   PTR_ERR(new_smi->thread));
			new_smi->thread = NULL;
		}
	}

	return 0;
}

static int get_smi_info(void *send_info, struct ipmi_smi_info *data)
{
	struct smi_info *smi = send_info;

	data->addr_src = smi->io.addr_source;
	data->dev = smi->io.dev;
	data->addr_info = smi->io.addr_info;
	get_device(smi->io.dev);

	return 0;
}

static void set_maintenance_mode(void *send_info, bool enable)
{
	struct smi_info   *smi_info = send_info;

	if (!enable)
		atomic_set(&smi_info->req_events, 0);
	smi_info->in_maintenance_mode = enable;
}

static void shutdown_smi(void *send_info);
static const struct ipmi_smi_handlers handlers = {
	.owner                  = THIS_MODULE,
	.start_processing       = smi_start_processing,
	.shutdown               = shutdown_smi,
	.get_smi_info		= get_smi_info,
	.sender			= sender,
	.request_events		= request_events,
	.set_need_watch		= set_need_watch,
	.set_maintenance_mode   = set_maintenance_mode,
	.set_run_to_completion  = set_run_to_completion,
	.flush_messages		= flush_messages,
	.poll			= poll,
};

static LIST_HEAD(smi_infos);
static DEFINE_MUTEX(smi_infos_lock);
static int smi_num; /* Used to sequence the SMIs */

static const char * const addr_space_to_str[] = { "i/o", "mem" };

module_param_array(force_kipmid, int, &num_force_kipmid, 0);
MODULE_PARM_DESC(force_kipmid, "Force the kipmi daemon to be enabled (1) or"
		 " disabled(0).  Normally the IPMI driver auto-detects"
		 " this, but the value may be overridden by this parm.");
module_param(unload_when_empty, bool, 0);
MODULE_PARM_DESC(unload_when_empty, "Unload the module if no interfaces are"
		 " specified or found, default is 1.  Setting to 0"
		 " is useful for hot add of devices using hotmod.");
module_param_array(kipmid_max_busy_us, uint, &num_max_busy_us, 0644);
MODULE_PARM_DESC(kipmid_max_busy_us,
		 "Max time (in microseconds) to busy-wait for IPMI data before"
		 " sleeping. 0 (default) means to wait forever. Set to 100-500"
		 " if kipmid is using up a lot of CPU time.");

void ipmi_irq_finish_setup(struct si_sm_io *io)
{
	if (io->si_type == SI_BT)
		/* Enable the interrupt in the BT interface. */
		io->outputb(io, IPMI_BT_INTMASK_REG,
			    IPMI_BT_INTMASK_ENABLE_IRQ_BIT);
}

void ipmi_irq_start_cleanup(struct si_sm_io *io)
{
	if (io->si_type == SI_BT)
		/* Disable the interrupt in the BT interface. */
		io->outputb(io, IPMI_BT_INTMASK_REG, 0);
}

static void std_irq_cleanup(struct si_sm_io *io)
{
	ipmi_irq_start_cleanup(io);
	free_irq(io->irq, io->irq_handler_data);
}

int ipmi_std_irq_setup(struct si_sm_io *io)
{
	int rv;

	if (!io->irq)
		return 0;

	rv = request_irq(io->irq,
			 ipmi_si_irq_handler,
			 IRQF_SHARED,
			 SI_DEVICE_NAME,
			 io->irq_handler_data);
	if (rv) {
		dev_warn(io->dev, "%s unable to claim interrupt %d,"
			 " running polled\n",
			 SI_DEVICE_NAME, io->irq);
		io->irq = 0;
	} else {
		io->irq_cleanup = std_irq_cleanup;
		ipmi_irq_finish_setup(io);
		dev_info(io->dev, "Using irq %d\n", io->irq);
	}

	return rv;
}

static int wait_for_msg_done(struct smi_info *smi_info)
{
	enum si_sm_result     smi_result;

	smi_result = smi_info->handlers->event(smi_info->si_sm, 0);
	for (;;) {
		if (smi_result == SI_SM_CALL_WITH_DELAY ||
		    smi_result == SI_SM_CALL_WITH_TICK_DELAY) {
			schedule_timeout_uninterruptible(1);
			smi_result = smi_info->handlers->event(
				smi_info->si_sm, jiffies_to_usecs(1));
		} else if (smi_result == SI_SM_CALL_WITHOUT_DELAY) {
			smi_result = smi_info->handlers->event(
				smi_info->si_sm, 0);
		} else
			break;
	}
	if (smi_result == SI_SM_HOSED)
		/*
		 * We couldn't get the state machine to run, so whatever's at
		 * the port is probably not an IPMI SMI interface.
		 */
		return -ENODEV;

	return 0;
}

static int try_get_dev_id(struct smi_info *smi_info)
{
	unsigned char         msg[2];
	unsigned char         *resp;
	unsigned long         resp_len;
	int                   rv = 0;

	resp = kmalloc(IPMI_MAX_MSG_LENGTH, GFP_KERNEL);
	if (!resp)
		return -ENOMEM;

	/*
	 * Do a Get Device ID command, since it comes back with some
	 * useful info.
	 */
	msg[0] = IPMI_NETFN_APP_REQUEST << 2;
	msg[1] = IPMI_GET_DEVICE_ID_CMD;
	smi_info->handlers->start_transaction(smi_info->si_sm, msg, 2);

	rv = wait_for_msg_done(smi_info);
	if (rv)
		goto out;

	resp_len = smi_info->handlers->get_result(smi_info->si_sm,
						  resp, IPMI_MAX_MSG_LENGTH);

	/* Check and record info from the get device id, in case we need it. */
	rv = ipmi_demangle_device_id(resp[0] >> 2, resp[1],
			resp + 2, resp_len - 2, &smi_info->device_id);

out:
	kfree(resp);
	return rv;
}

static int get_global_enables(struct smi_info *smi_info, u8 *enables)
{
	unsigned char         msg[3];
	unsigned char         *resp;
	unsigned long         resp_len;
	int                   rv;

	resp = kmalloc(IPMI_MAX_MSG_LENGTH, GFP_KERNEL);
	if (!resp)
		return -ENOMEM;

	msg[0] = IPMI_NETFN_APP_REQUEST << 2;
	msg[1] = IPMI_GET_BMC_GLOBAL_ENABLES_CMD;
	smi_info->handlers->start_transaction(smi_info->si_sm, msg, 2);

	rv = wait_for_msg_done(smi_info);
	if (rv) {
		dev_warn(smi_info->io.dev,
			 "Error getting response from get global enables command: %d\n",
			 rv);
		goto out;
	}

	resp_len = smi_info->handlers->get_result(smi_info->si_sm,
						  resp, IPMI_MAX_MSG_LENGTH);

	if (resp_len < 4 ||
			resp[0] != (IPMI_NETFN_APP_REQUEST | 1) << 2 ||
			resp[1] != IPMI_GET_BMC_GLOBAL_ENABLES_CMD   ||
			resp[2] != 0) {
		dev_warn(smi_info->io.dev,
			 "Invalid return from get global enables command: %ld %x %x %x\n",
			 resp_len, resp[0], resp[1], resp[2]);
		rv = -EINVAL;
		goto out;
	} else {
		*enables = resp[3];
	}

out:
	kfree(resp);
	return rv;
}

/*
 * Returns 1 if it gets an error from the command.
 */
static int set_global_enables(struct smi_info *smi_info, u8 enables)
{
	unsigned char         msg[3];
	unsigned char         *resp;
	unsigned long         resp_len;
	int                   rv;

	resp = kmalloc(IPMI_MAX_MSG_LENGTH, GFP_KERNEL);
	if (!resp)
		return -ENOMEM;

	msg[0] = IPMI_NETFN_APP_REQUEST << 2;
	msg[1] = IPMI_SET_BMC_GLOBAL_ENABLES_CMD;
	msg[2] = enables;
	smi_info->handlers->start_transaction(smi_info->si_sm, msg, 3);

	rv = wait_for_msg_done(smi_info);
	if (rv) {
		dev_warn(smi_info->io.dev,
			 "Error getting response from set global enables command: %d\n",
			 rv);
		goto out;
	}

	resp_len = smi_info->handlers->get_result(smi_info->si_sm,
						  resp, IPMI_MAX_MSG_LENGTH);

	if (resp_len < 3 ||
			resp[0] != (IPMI_NETFN_APP_REQUEST | 1) << 2 ||
			resp[1] != IPMI_SET_BMC_GLOBAL_ENABLES_CMD) {
		dev_warn(smi_info->io.dev,
			 "Invalid return from set global enables command: %ld %x %x\n",
			 resp_len, resp[0], resp[1]);
		rv = -EINVAL;
		goto out;
	}

	if (resp[2] != 0)
		rv = 1;

out:
	kfree(resp);
	return rv;
}

/*
 * Some BMCs do not support clearing the receive irq bit in the global
 * enables (even if they don't support interrupts on the BMC).  Check
 * for this and handle it properly.
 */
static void check_clr_rcv_irq(struct smi_info *smi_info)
{
	u8 enables = 0;
	int rv;

	rv = get_global_enables(smi_info, &enables);
	if (!rv) {
		if ((enables & IPMI_BMC_RCV_MSG_INTR) == 0)
			/* Already clear, should work ok. */
			return;

		enables &= ~IPMI_BMC_RCV_MSG_INTR;
		rv = set_global_enables(smi_info, enables);
	}

	if (rv < 0) {
		dev_err(smi_info->io.dev,
			"Cannot check clearing the rcv irq: %d\n", rv);
		return;
	}

	if (rv) {
		/*
		 * An error when setting the event buffer bit means
		 * clearing the bit is not supported.
		 */
		dev_warn(smi_info->io.dev,
			 "The BMC does not support clearing the recv irq bit, compensating, but the BMC needs to be fixed.\n");
		smi_info->cannot_disable_irq = true;
	}
}

/*
 * Some BMCs do not support setting the interrupt bits in the global
 * enables even if they support interrupts.  Clearly bad, but we can
 * compensate.
 */
static void check_set_rcv_irq(struct smi_info *smi_info)
{
	u8 enables = 0;
	int rv;

	if (!smi_info->io.irq)
		return;

	rv = get_global_enables(smi_info, &enables);
	if (!rv) {
		enables |= IPMI_BMC_RCV_MSG_INTR;
		rv = set_global_enables(smi_info, enables);
	}

	if (rv < 0) {
		dev_err(smi_info->io.dev,
			"Cannot check setting the rcv irq: %d\n", rv);
		return;
	}

	if (rv) {
		/*
		 * An error when setting the event buffer bit means
		 * setting the bit is not supported.
		 */
		dev_warn(smi_info->io.dev,
			 "The BMC does not support setting the recv irq bit, compensating, but the BMC needs to be fixed.\n");
		smi_info->cannot_disable_irq = true;
		smi_info->irq_enable_broken = true;
	}
}

static int try_enable_event_buffer(struct smi_info *smi_info)
{
	unsigned char         msg[3];
	unsigned char         *resp;
	unsigned long         resp_len;
	int                   rv = 0;

	resp = kmalloc(IPMI_MAX_MSG_LENGTH, GFP_KERNEL);
	if (!resp)
		return -ENOMEM;

	msg[0] = IPMI_NETFN_APP_REQUEST << 2;
	msg[1] = IPMI_GET_BMC_GLOBAL_ENABLES_CMD;
	smi_info->handlers->start_transaction(smi_info->si_sm, msg, 2);

	rv = wait_for_msg_done(smi_info);
	if (rv) {
		pr_warn("Error getting response from get global enables command, the event buffer is not enabled\n");
		goto out;
	}

	resp_len = smi_info->handlers->get_result(smi_info->si_sm,
						  resp, IPMI_MAX_MSG_LENGTH);

	if (resp_len < 4 ||
			resp[0] != (IPMI_NETFN_APP_REQUEST | 1) << 2 ||
			resp[1] != IPMI_GET_BMC_GLOBAL_ENABLES_CMD   ||
			resp[2] != 0) {
		pr_warn("Invalid return from get global enables command, cannot enable the event buffer\n");
		rv = -EINVAL;
		goto out;
	}

	if (resp[3] & IPMI_BMC_EVT_MSG_BUFF) {
		/* buffer is already enabled, nothing to do. */
		smi_info->supports_event_msg_buff = true;
		goto out;
	}

	msg[0] = IPMI_NETFN_APP_REQUEST << 2;
	msg[1] = IPMI_SET_BMC_GLOBAL_ENABLES_CMD;
	msg[2] = resp[3] | IPMI_BMC_EVT_MSG_BUFF;
	smi_info->handlers->start_transaction(smi_info->si_sm, msg, 3);

	rv = wait_for_msg_done(smi_info);
	if (rv) {
		pr_warn("Error getting response from set global, enables command, the event buffer is not enabled\n");
		goto out;
	}

	resp_len = smi_info->handlers->get_result(smi_info->si_sm,
						  resp, IPMI_MAX_MSG_LENGTH);

	if (resp_len < 3 ||
			resp[0] != (IPMI_NETFN_APP_REQUEST | 1) << 2 ||
			resp[1] != IPMI_SET_BMC_GLOBAL_ENABLES_CMD) {
		pr_warn("Invalid return from get global, enables command, not enable the event buffer\n");
		rv = -EINVAL;
		goto out;
	}

	if (resp[2] != 0)
		/*
		 * An error when setting the event buffer bit means
		 * that the event buffer is not supported.
		 */
		rv = -ENOENT;
	else
		smi_info->supports_event_msg_buff = true;

out:
	kfree(resp);
	return rv;
}

#define IPMI_SI_ATTR(name) \
static ssize_t name##_show(struct device *dev,			\
			   struct device_attribute *attr,		\
			   char *buf)					\
{									\
	struct smi_info *smi_info = dev_get_drvdata(dev);		\
									\
	return snprintf(buf, 10, "%u\n", smi_get_stat(smi_info, name));	\
}									\
static DEVICE_ATTR(name, 0444, name##_show, NULL)

static ssize_t type_show(struct device *dev,
			 struct device_attribute *attr,
			 char *buf)
{
	struct smi_info *smi_info = dev_get_drvdata(dev);

	return snprintf(buf, 10, "%s\n", si_to_str[smi_info->io.si_type]);
}
static DEVICE_ATTR(type, 0444, type_show, NULL);

static ssize_t interrupts_enabled_show(struct device *dev,
				       struct device_attribute *attr,
				       char *buf)
{
	struct smi_info *smi_info = dev_get_drvdata(dev);
	int enabled = smi_info->io.irq && !smi_info->interrupt_disabled;

	return snprintf(buf, 10, "%d\n", enabled);
}
static DEVICE_ATTR(interrupts_enabled, 0444,
		   interrupts_enabled_show, NULL);

IPMI_SI_ATTR(short_timeouts);
IPMI_SI_ATTR(long_timeouts);
IPMI_SI_ATTR(idles);
IPMI_SI_ATTR(interrupts);
IPMI_SI_ATTR(attentions);
IPMI_SI_ATTR(flag_fetches);
IPMI_SI_ATTR(hosed_count);
IPMI_SI_ATTR(complete_transactions);
IPMI_SI_ATTR(events);
IPMI_SI_ATTR(watchdog_pretimeouts);
IPMI_SI_ATTR(incoming_messages);

static ssize_t params_show(struct device *dev,
			   struct device_attribute *attr,
			   char *buf)
{
	struct smi_info *smi_info = dev_get_drvdata(dev);

	return snprintf(buf, 200,
			"%s,%s,0x%lx,rsp=%d,rsi=%d,rsh=%d,irq=%d,ipmb=%d\n",
			si_to_str[smi_info->io.si_type],
			addr_space_to_str[smi_info->io.addr_space],
			smi_info->io.addr_data,
			smi_info->io.regspacing,
			smi_info->io.regsize,
			smi_info->io.regshift,
			smi_info->io.irq,
			smi_info->io.slave_addr);
}
static DEVICE_ATTR(params, 0444, params_show, NULL);

static struct attribute *ipmi_si_dev_attrs[] = {
	&dev_attr_type.attr,
	&dev_attr_interrupts_enabled.attr,
	&dev_attr_short_timeouts.attr,
	&dev_attr_long_timeouts.attr,
	&dev_attr_idles.attr,
	&dev_attr_interrupts.attr,
	&dev_attr_attentions.attr,
	&dev_attr_flag_fetches.attr,
	&dev_attr_hosed_count.attr,
	&dev_attr_complete_transactions.attr,
	&dev_attr_events.attr,
	&dev_attr_watchdog_pretimeouts.attr,
	&dev_attr_incoming_messages.attr,
	&dev_attr_params.attr,
	NULL
};

static const struct attribute_group ipmi_si_dev_attr_group = {
	.attrs		= ipmi_si_dev_attrs,
};

/*
 * oem_data_avail_to_receive_msg_avail
 * @info - smi_info structure with msg_flags set
 *
 * Converts flags from OEM_DATA_AVAIL to RECEIVE_MSG_AVAIL
 * Returns 1 indicating need to re-run handle_flags().
 */
static int oem_data_avail_to_receive_msg_avail(struct smi_info *smi_info)
{
	smi_info->msg_flags = ((smi_info->msg_flags & ~OEM_DATA_AVAIL) |
			       RECEIVE_MSG_AVAIL);
	return 1;
}

/*
 * setup_dell_poweredge_oem_data_handler
 * @info - smi_info.device_id must be populated
 *
 * Systems that match, but have firmware version < 1.40 may assert
 * OEM0_DATA_AVAIL on their own, without being told via Set Flags that
 * it's safe to do so.  Such systems will de-assert OEM1_DATA_AVAIL
 * upon receipt of IPMI_GET_MSG_CMD, so we should treat these flags
 * as RECEIVE_MSG_AVAIL instead.
 *
 * As Dell has no plans to release IPMI 1.5 firmware that *ever*
 * assert the OEM[012] bits, and if it did, the driver would have to
 * change to handle that properly, we don't actually check for the
 * firmware version.
 * Device ID = 0x20                BMC on PowerEdge 8G servers
 * Device Revision = 0x80
 * Firmware Revision1 = 0x01       BMC version 1.40
 * Firmware Revision2 = 0x40       BCD encoded
 * IPMI Version = 0x51             IPMI 1.5
 * Manufacturer ID = A2 02 00      Dell IANA
 *
 * Additionally, PowerEdge systems with IPMI < 1.5 may also assert
 * OEM0_DATA_AVAIL and needs to be treated as RECEIVE_MSG_AVAIL.
 *
 */
#define DELL_POWEREDGE_8G_BMC_DEVICE_ID  0x20
#define DELL_POWEREDGE_8G_BMC_DEVICE_REV 0x80
#define DELL_POWEREDGE_8G_BMC_IPMI_VERSION 0x51
#define DELL_IANA_MFR_ID 0x0002a2
static void setup_dell_poweredge_oem_data_handler(struct smi_info *smi_info)
{
	struct ipmi_device_id *id = &smi_info->device_id;
	if (id->manufacturer_id == DELL_IANA_MFR_ID) {
		if (id->device_id       == DELL_POWEREDGE_8G_BMC_DEVICE_ID  &&
		    id->device_revision == DELL_POWEREDGE_8G_BMC_DEVICE_REV &&
		    id->ipmi_version   == DELL_POWEREDGE_8G_BMC_IPMI_VERSION) {
			smi_info->oem_data_avail_handler =
				oem_data_avail_to_receive_msg_avail;
		} else if (ipmi_version_major(id) < 1 ||
			   (ipmi_version_major(id) == 1 &&
			    ipmi_version_minor(id) < 5)) {
			smi_info->oem_data_avail_handler =
				oem_data_avail_to_receive_msg_avail;
		}
	}
}

#define CANNOT_RETURN_REQUESTED_LENGTH 0xCA
static void return_hosed_msg_badsize(struct smi_info *smi_info)
{
	struct ipmi_smi_msg *msg = smi_info->curr_msg;

	/* Make it a response */
	msg->rsp[0] = msg->data[0] | 4;
	msg->rsp[1] = msg->data[1];
	msg->rsp[2] = CANNOT_RETURN_REQUESTED_LENGTH;
	msg->rsp_size = 3;
	smi_info->curr_msg = NULL;
	deliver_recv_msg(smi_info, msg);
}

/*
 * dell_poweredge_bt_xaction_handler
 * @info - smi_info.device_id must be populated
 *
 * Dell PowerEdge servers with the BT interface (x6xx and 1750) will
 * not respond to a Get SDR command if the length of the data
 * requested is exactly 0x3A, which leads to command timeouts and no
 * data returned.  This intercepts such commands, and causes userspace
 * callers to try again with a different-sized buffer, which succeeds.
 */

#define STORAGE_NETFN 0x0A
#define STORAGE_CMD_GET_SDR 0x23
static int dell_poweredge_bt_xaction_handler(struct notifier_block *self,
					     unsigned long unused,
					     void *in)
{
	struct smi_info *smi_info = in;
	unsigned char *data = smi_info->curr_msg->data;
	unsigned int size   = smi_info->curr_msg->data_size;
	if (size >= 8 &&
	    (data[0]>>2) == STORAGE_NETFN &&
	    data[1] == STORAGE_CMD_GET_SDR &&
	    data[7] == 0x3A) {
		return_hosed_msg_badsize(smi_info);
		return NOTIFY_STOP;
	}
	return NOTIFY_DONE;
}

static struct notifier_block dell_poweredge_bt_xaction_notifier = {
	.notifier_call	= dell_poweredge_bt_xaction_handler,
};

/*
 * setup_dell_poweredge_bt_xaction_handler
 * @info - smi_info.device_id must be filled in already
 *
 * Fills in smi_info.device_id.start_transaction_pre_hook
 * when we know what function to use there.
 */
static void
setup_dell_poweredge_bt_xaction_handler(struct smi_info *smi_info)
{
	struct ipmi_device_id *id = &smi_info->device_id;
	if (id->manufacturer_id == DELL_IANA_MFR_ID &&
	    smi_info->io.si_type == SI_BT)
		register_xaction_notifier(&dell_poweredge_bt_xaction_notifier);
}

/*
 * setup_oem_data_handler
 * @info - smi_info.device_id must be filled in already
 *
 * Fills in smi_info.device_id.oem_data_available_handler
 * when we know what function to use there.
 */

static void setup_oem_data_handler(struct smi_info *smi_info)
{
	setup_dell_poweredge_oem_data_handler(smi_info);
}

static void setup_xaction_handlers(struct smi_info *smi_info)
{
	setup_dell_poweredge_bt_xaction_handler(smi_info);
}

static void check_for_broken_irqs(struct smi_info *smi_info)
{
	check_clr_rcv_irq(smi_info);
	check_set_rcv_irq(smi_info);
}

static inline void stop_timer_and_thread(struct smi_info *smi_info)
{
	if (smi_info->thread != NULL) {
		kthread_stop(smi_info->thread);
		smi_info->thread = NULL;
	}

	smi_info->timer_can_start = false;
	del_timer_sync(&smi_info->si_timer);
}

static struct smi_info *find_dup_si(struct smi_info *info)
{
	struct smi_info *e;

	list_for_each_entry(e, &smi_infos, link) {
		if (e->io.addr_space != info->io.addr_space)
			continue;
		if (e->io.addr_data == info->io.addr_data) {
			/*
			 * This is a cheap hack, ACPI doesn't have a defined
			 * slave address but SMBIOS does.  Pick it up from
			 * any source that has it available.
			 */
			if (info->io.slave_addr && !e->io.slave_addr)
				e->io.slave_addr = info->io.slave_addr;
			return e;
		}
	}

	return NULL;
}

int ipmi_si_add_smi(struct si_sm_io *io)
{
	int rv = 0;
	struct smi_info *new_smi, *dup;

	/*
	 * If the user gave us a hard-coded device at the same
	 * address, they presumably want us to use it and not what is
	 * in the firmware.
	 */
	if (io->addr_source != SI_HARDCODED && io->addr_source != SI_HOTMOD &&
	    ipmi_si_hardcode_match(io->addr_space, io->addr_data)) {
		dev_info(io->dev,
			 "Hard-coded device at this address already exists");
		return -ENODEV;
	}

	if (!io->io_setup) {
		if (io->addr_space == IPMI_IO_ADDR_SPACE) {
			io->io_setup = ipmi_si_port_setup;
		} else if (io->addr_space == IPMI_MEM_ADDR_SPACE) {
			io->io_setup = ipmi_si_mem_setup;
		} else {
			return -EINVAL;
		}
	}

	new_smi = kzalloc(sizeof(*new_smi), GFP_KERNEL);
	if (!new_smi)
		return -ENOMEM;
	spin_lock_init(&new_smi->si_lock);

	new_smi->io = *io;

	mutex_lock(&smi_infos_lock);
	dup = find_dup_si(new_smi);
	if (dup) {
		if (new_smi->io.addr_source == SI_ACPI &&
		    dup->io.addr_source == SI_SMBIOS) {
			/* We prefer ACPI over SMBIOS. */
			dev_info(dup->io.dev,
				 "Removing SMBIOS-specified %s state machine in favor of ACPI\n",
				 si_to_str[new_smi->io.si_type]);
			cleanup_one_si(dup);
		} else {
			dev_info(new_smi->io.dev,
				 "%s-specified %s state machine: duplicate\n",
				 ipmi_addr_src_to_str(new_smi->io.addr_source),
				 si_to_str[new_smi->io.si_type]);
			rv = -EBUSY;
			kfree(new_smi);
			goto out_err;
		}
	}

	pr_info("Adding %s-specified %s state machine\n",
		ipmi_addr_src_to_str(new_smi->io.addr_source),
		si_to_str[new_smi->io.si_type]);

	list_add_tail(&new_smi->link, &smi_infos);

	if (initialized)
		rv = try_smi_init(new_smi);
out_err:
	mutex_unlock(&smi_infos_lock);
	return rv;
}

/*
 * Try to start up an interface.  Must be called with smi_infos_lock
 * held, primarily to keep smi_num consistent, we only one to do these
 * one at a time.
 */
static int try_smi_init(struct smi_info *new_smi)
{
	int rv = 0;
	int i;

	pr_info("Trying %s-specified %s state machine at %s address 0x%lx, slave address 0x%x, irq %d\n",
		ipmi_addr_src_to_str(new_smi->io.addr_source),
		si_to_str[new_smi->io.si_type],
		addr_space_to_str[new_smi->io.addr_space],
		new_smi->io.addr_data,
		new_smi->io.slave_addr, new_smi->io.irq);

	switch (new_smi->io.si_type) {
	case SI_KCS:
		new_smi->handlers = &kcs_smi_handlers;
		break;

	case SI_SMIC:
		new_smi->handlers = &smic_smi_handlers;
		break;

	case SI_BT:
		new_smi->handlers = &bt_smi_handlers;
		break;

	default:
		/* No support for anything else yet. */
		rv = -EIO;
		goto out_err;
	}

	new_smi->si_num = smi_num;

	/* Do this early so it's available for logs. */
	if (!new_smi->io.dev) {
		pr_err("IPMI interface added with no device\n");
		rv = EIO;
		goto out_err;
	}

	/* Allocate the state machine's data and initialize it. */
	new_smi->si_sm = kmalloc(new_smi->handlers->size(), GFP_KERNEL);
	if (!new_smi->si_sm) {
		rv = -ENOMEM;
		goto out_err;
	}
	new_smi->io.io_size = new_smi->handlers->init_data(new_smi->si_sm,
							   &new_smi->io);

	/* Now that we know the I/O size, we can set up the I/O. */
	rv = new_smi->io.io_setup(&new_smi->io);
	if (rv) {
		dev_err(new_smi->io.dev, "Could not set up I/O space\n");
		goto out_err;
	}

	/* Do low-level detection first. */
	if (new_smi->handlers->detect(new_smi->si_sm)) {
		if (new_smi->io.addr_source)
			dev_err(new_smi->io.dev,
				"Interface detection failed\n");
		rv = -ENODEV;
		goto out_err;
	}

	/*
	 * Attempt a get device id command.  If it fails, we probably
	 * don't have a BMC here.
	 */
	rv = try_get_dev_id(new_smi);
	if (rv) {
		if (new_smi->io.addr_source)
			dev_err(new_smi->io.dev,
			       "There appears to be no BMC at this location\n");
		goto out_err;
	}

	setup_oem_data_handler(new_smi);
	setup_xaction_handlers(new_smi);
	check_for_broken_irqs(new_smi);

	new_smi->waiting_msg = NULL;
	new_smi->curr_msg = NULL;
	atomic_set(&new_smi->req_events, 0);
	new_smi->run_to_completion = false;
	for (i = 0; i < SI_NUM_STATS; i++)
		atomic_set(&new_smi->stats[i], 0);

	new_smi->interrupt_disabled = true;
	atomic_set(&new_smi->need_watch, 0);

	rv = try_enable_event_buffer(new_smi);
	if (rv == 0)
		new_smi->has_event_buffer = true;

	/*
	 * Start clearing the flags before we enable interrupts or the
	 * timer to avoid racing with the timer.
	 */
	start_clear_flags(new_smi);

	/*
	 * IRQ is defined to be set when non-zero.  req_events will
	 * cause a global flags check that will enable interrupts.
	 */
	if (new_smi->io.irq) {
		new_smi->interrupt_disabled = false;
		atomic_set(&new_smi->req_events, 1);
	}

	dev_set_drvdata(new_smi->io.dev, new_smi);
	rv = device_add_group(new_smi->io.dev, &ipmi_si_dev_attr_group);
	if (rv) {
		dev_err(new_smi->io.dev,
			"Unable to add device attributes: error %d\n",
			rv);
		goto out_err;
	}
	new_smi->dev_group_added = true;

	rv = ipmi_register_smi(&handlers,
			       new_smi,
			       new_smi->io.dev,
			       new_smi->io.slave_addr);
	if (rv) {
		dev_err(new_smi->io.dev,
			"Unable to register device: error %d\n",
			rv);
		goto out_err;
	}

	/* Don't increment till we know we have succeeded. */
	smi_num++;

	dev_info(new_smi->io.dev, "IPMI %s interface initialized\n",
		 si_to_str[new_smi->io.si_type]);

	WARN_ON(new_smi->io.dev->init_name != NULL);

 out_err:
	if (rv && new_smi->io.io_cleanup) {
		new_smi->io.io_cleanup(&new_smi->io);
		new_smi->io.io_cleanup = NULL;
	}

	return rv;
}

static int __init init_ipmi_si(void)
{
	struct smi_info *e;
	enum ipmi_addr_src type = SI_INVALID;

	if (initialized)
		return 0;

	ipmi_hardcode_init();

	pr_info("IPMI System Interface driver\n");

	ipmi_si_platform_init();

	ipmi_si_pci_init();

	ipmi_si_parisc_init();

	/* We prefer devices with interrupts, but in the case of a machine
	   with multiple BMCs we assume that there will be several instances
	   of a given type so if we succeed in registering a type then also
	   try to register everything else of the same type */
	mutex_lock(&smi_infos_lock);
	list_for_each_entry(e, &smi_infos, link) {
		/* Try to register a device if it has an IRQ and we either
		   haven't successfully registered a device yet or this
		   device has the same type as one we successfully registered */
		if (e->io.irq && (!type || e->io.addr_source == type)) {
			if (!try_smi_init(e)) {
				type = e->io.addr_source;
			}
		}
	}

	/* type will only have been set if we successfully registered an si */
	if (type)
		goto skip_fallback_noirq;

	/* Fall back to the preferred device */

	list_for_each_entry(e, &smi_infos, link) {
		if (!e->io.irq && (!type || e->io.addr_source == type)) {
			if (!try_smi_init(e)) {
				type = e->io.addr_source;
			}
		}
	}

skip_fallback_noirq:
	initialized = true;
	mutex_unlock(&smi_infos_lock);

	if (type)
		return 0;

	mutex_lock(&smi_infos_lock);
	if (unload_when_empty && list_empty(&smi_infos)) {
		mutex_unlock(&smi_infos_lock);
		cleanup_ipmi_si();
		pr_warn("Unable to find any System Interface(s)\n");
		return -ENODEV;
	} else {
		mutex_unlock(&smi_infos_lock);
		return 0;
	}
}
module_init(init_ipmi_si);

static void shutdown_smi(void *send_info)
{
	struct smi_info *smi_info = send_info;

	if (smi_info->dev_group_added) {
		device_remove_group(smi_info->io.dev, &ipmi_si_dev_attr_group);
		smi_info->dev_group_added = false;
	}
	if (smi_info->io.dev)
		dev_set_drvdata(smi_info->io.dev, NULL);

	/*
	 * Make sure that interrupts, the timer and the thread are
	 * stopped and will not run again.
	 */
	smi_info->interrupt_disabled = true;
	if (smi_info->io.irq_cleanup) {
		smi_info->io.irq_cleanup(&smi_info->io);
		smi_info->io.irq_cleanup = NULL;
	}
	stop_timer_and_thread(smi_info);

	/*
	 * Wait until we know that we are out of any interrupt
	 * handlers might have been running before we freed the
	 * interrupt.
	 */
	synchronize_rcu();

	/*
	 * Timeouts are stopped, now make sure the interrupts are off
	 * in the BMC.  Note that timers and CPU interrupts are off,
	 * so no need for locks.
	 */
	while (smi_info->curr_msg || (smi_info->si_state != SI_NORMAL)) {
		poll(smi_info);
		schedule_timeout_uninterruptible(1);
	}
	if (smi_info->handlers)
		disable_si_irq(smi_info);
	while (smi_info->curr_msg || (smi_info->si_state != SI_NORMAL)) {
		poll(smi_info);
		schedule_timeout_uninterruptible(1);
	}
	if (smi_info->handlers)
		smi_info->handlers->cleanup(smi_info->si_sm);

	if (smi_info->io.addr_source_cleanup) {
		smi_info->io.addr_source_cleanup(&smi_info->io);
		smi_info->io.addr_source_cleanup = NULL;
	}
	if (smi_info->io.io_cleanup) {
		smi_info->io.io_cleanup(&smi_info->io);
		smi_info->io.io_cleanup = NULL;
	}

	kfree(smi_info->si_sm);
	smi_info->si_sm = NULL;

	smi_info->intf = NULL;
}

/*
 * Must be called with smi_infos_lock held, to serialize the
 * smi_info->intf check.
 */
static void cleanup_one_si(struct smi_info *smi_info)
{
	if (!smi_info)
		return;

	list_del(&smi_info->link);

	if (smi_info->intf)
		ipmi_unregister_smi(smi_info->intf);

	kfree(smi_info);
}

int ipmi_si_remove_by_dev(struct device *dev)
{
	struct smi_info *e;
	int rv = -ENOENT;

	mutex_lock(&smi_infos_lock);
	list_for_each_entry(e, &smi_infos, link) {
		if (e->io.dev == dev) {
			cleanup_one_si(e);
			rv = 0;
			break;
		}
	}
	mutex_unlock(&smi_infos_lock);

	return rv;
}

struct device *ipmi_si_remove_by_data(int addr_space, enum si_type si_type,
				      unsigned long addr)
{
	/* remove */
	struct smi_info *e, *tmp_e;
	struct device *dev = NULL;

	mutex_lock(&smi_infos_lock);
	list_for_each_entry_safe(e, tmp_e, &smi_infos, link) {
		if (e->io.addr_space != addr_space)
			continue;
		if (e->io.si_type != si_type)
			continue;
		if (e->io.addr_data == addr) {
			dev = get_device(e->io.dev);
			cleanup_one_si(e);
		}
	}
	mutex_unlock(&smi_infos_lock);

	return dev;
}

static void cleanup_ipmi_si(void)
{
	struct smi_info *e, *tmp_e;

	if (!initialized)
		return;

	ipmi_si_pci_shutdown();

	ipmi_si_parisc_shutdown();

	ipmi_si_platform_shutdown();

	mutex_lock(&smi_infos_lock);
	list_for_each_entry_safe(e, tmp_e, &smi_infos, link)
		cleanup_one_si(e);
	mutex_unlock(&smi_infos_lock);

	ipmi_si_hardcode_exit();
	ipmi_si_hotmod_exit();
}
module_exit(cleanup_ipmi_si);

MODULE_ALIAS("platform:dmi-ipmi-si");
MODULE_LICENSE("GPL");
MODULE_AUTHOR("Corey Minyard <minyard@mvista.com>");
MODULE_DESCRIPTION("Interface to the IPMI driver for the KCS, SMIC, and BT"
		   " system interfaces.");