aboutsummaryrefslogtreecommitdiff
path: root/core/hmi.c
blob: 7f4211c5d19f37586cfc12ded38e83b16e652b26 (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
// SPDX-License-Identifier: Apache-2.0 OR GPL-2.0-or-later
/*
 * Deal with Hypervisor Maintenance Interrupts
 *
 * Copyright 2013-2019 IBM Corp.
 */

#define pr_fmt(fmt)	"HMI: " fmt

#include <skiboot.h>
#include <opal.h>
#include <opal-msg.h>
#include <processor.h>
#include <chiptod.h>
#include <xscom.h>
#include <xscom-p8-regs.h>
#include <xscom-p9-regs.h>
#include <xscom-p10-regs.h>
#include <pci.h>
#include <cpu.h>
#include <chip.h>
#include <pau-regs.h>
#include <npu-regs.h>
#include <npu2-regs.h>
#include <pau.h>
#include <npu2.h>
#include <npu.h>
#include <capp.h>
#include <nvram.h>
#include <cpu.h>

/*
 * P9 HMER register layout:
 * +===+==========+============================+========+===================+
 * |Bit|Name      |Description                 |PowerKVM|Action             |
 * |   |          |                            |HMI     |                   |
 * |   |          |                            |enabled |                   |
 * |   |          |                            |for this|                   |
 * |   |          |                            |bit ?   |                   |
 * +===+==========+============================+========+===================+
 * |0  |malfunctio|A processor core in the     |Yes     |Raise attn from    |
 * |   |n_allert  |system has checkstopped     |        |sapphire resulting |
 * |   |          |(failed recovery) and has   |        |xstop              |
 * |   |          |requested a CP Sparing      |        |                   |
 * |   |          |to occur. This is           |        |                   |
 * |   |          |broadcasted to every        |        |                   |
 * |   |          |processor in the system     |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |1  |Reserved  |reserved                    |n/a     |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |2  |proc_recv_|Processor recovery occurred |Yes     |Log message and    |
 * |   |done      |error-bit in fir not masked |        |continue working.  |
 * |   |          |(see bit 11)                |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |3  |proc_recv_|Processor went through      |Yes     |Log message and    |
 * |   |error_mask|recovery for an error which |        |continue working.  |
 * |   |ed        |is actually masked for      |        |                   |
 * |   |          |reporting                   |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |4  |          |Timer facility experienced  |Yes     |Raise attn from    |
 * |   |tfac_error|an error.                   |        |sapphire resulting |
 * |   |          |TB, DEC, HDEC, PURR or SPURR|        |xstop              |
 * |   |          |may be corrupted (details in|        |                   |
 * |   |          |TFMR)                       |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |5  |          |TFMR SPR itself is          |Yes     |Raise attn from    |
 * |   |tfmr_parit|corrupted.                  |        |sapphire resulting |
 * |   |y_error   |Entire timing facility may  |        |xstop              |
 * |   |          |be compromised.             |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |6  |ha_overflo| UPS (Uniterrupted Power    |No      |N/A                |
 * |   |w_warning |System) Overflow indication |        |                   |
 * |   |          |indicating that the UPS     |        |                   |
 * |   |          |DirtyAddrTable has          |        |                   |
 * |   |          |reached a limit where it    |        |                   |
 * |   |          |requires PHYP unload support|        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |7  |reserved  |reserved                    |n/a     |n/a                |
 * |---+----------+----------------------------+--------+-------------------|
 * |8  |xscom_fail|An XSCOM operation caused by|No      |We handle it by    |
 * |   |          |a cache inhibited load/store|        |manually reading   |
 * |   |          |from this thread failed. A  |        |HMER register.     |
 * |   |          |trap register is            |        |                   |
 * |   |          |available.                  |        |                   |
 * |   |          |                            |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |9  |xscom_done|An XSCOM operation caused by|No      |We handle it by    |
 * |   |          |a cache inhibited load/store|        |manually reading   |
 * |   |          |from this thread completed. |        |HMER register.     |
 * |   |          |If hypervisor               |        |                   |
 * |   |          |intends to use this bit, it |        |                   |
 * |   |          |is responsible for clearing |        |                   |
 * |   |          |it before performing the    |        |                   |
 * |   |          |xscom operation.            |        |                   |
 * |   |          |NOTE: this bit should always|        |                   |
 * |   |          |be masked in HMEER          |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |10 |reserved  |reserved                    |n/a     |n/a                |
 * |---+----------+----------------------------+--------+-------------------|
 * |11 |proc_recv_|Processor recovery occurred |y       |Log message and    |
 * |   |again     |again before bit2 or bit3   |        |continue working.  |
 * |   |          |was cleared                 |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |12-|reserved  |was temperature sensor      |n/a     |n/a                |
 * |15 |          |passed the critical point on|        |                   |
 * |   |          |the way up                  |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |16 |          |SCOM has set a reserved FIR |No      |n/a                |
 * |   |scom_fir_h|bit to cause recovery       |        |                   |
 * |   |m         |                            |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |17 |trig_fir_h|Debug trigger has set a     |No      |n/a                |
 * |   |mi        |reserved FIR bit to cause   |        |                   |
 * |   |          |recovery                    |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |18 |reserved  |reserved                    |n/a     |n/a                |
 * |---+----------+----------------------------+--------+-------------------|
 * |19 |reserved  |reserved                    |n/a     |n/a                |
 * |---+----------+----------------------------+--------+-------------------|
 * |20 |hyp_resour|A hypervisor resource error |y       |Raise attn from    |
 * |   |ce_err    |occurred: data parity error |        |sapphire resulting |
 * |   |          |on, SPRC0:3; SPR_Modereg or |        |xstop.             |
 * |   |          |HMEER.                      |        |                   |
 * |   |          |Note: this bit will cause an|        |                   |
 * |   |          |check_stop when (HV=1, PR=0 |        |                   |
 * |   |          |and EE=0)                   |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |21-|          |if bit 8 is active, the     |No      |We handle it by    |
 * |23 |xscom_stat|reason will be detailed in  |        |Manually reading   |
 * |   |us        |these bits. see chapter 11.1|        |HMER register.     |
 * |   |          |This bits are information   |        |                   |
 * |   |          |only and always masked      |        |                   |
 * |   |          |(mask = '0')                |        |                   |
 * |   |          |If hypervisor intends to use|        |                   |
 * |   |          |this bit, it is responsible |        |                   |
 * |   |          |for clearing it before      |        |                   |
 * |   |          |performing the xscom        |        |                   |
 * |   |          |operation.                  |        |                   |
 * |---+----------+----------------------------+--------+-------------------|
 * |24-|Not       |Not implemented             |n/a     |n/a                |
 * |63 |implemente|                            |        |                   |
 * |   |d         |                            |        |                   |
 * +-- +----------+----------------------------+--------+-------------------+
 *
 * Above HMER bits can be enabled/disabled by modifying
 * SPR_HMEER_HMI_ENABLE_MASK #define in include/processor.h
 * If you modify support for any of the bits listed above, please make sure
 * you change the above table to refelct that.
 *
 * NOTE: Per Dave Larson, never enable 8,9,21-23
 */

/*
 * P10 HMER register layout:
 * Bit   Name                Description
 * 0     malfunction_alert   A processor core in the system has checkstopped
 *                           (failed recovery). This is broadcasted to every
 *                           processor in the system
 *
 * 1     reserved            reserved
 *
 * 2     proc_rcvy_done      Processor recovery occurred error-bit in fir not
 *                           masked (see bit 11)
 *
 * 3     reserved            reserved
 *
 * 4     tfac_error          Timer facility experienced an error. TB, DEC,
 *                           HDEC, PURR or SPURR may be corrupted (details in
 *                           TFMR)
 *
 * 5     tfx_error           Error occurred on transfer from tfac shadow to
 *                           core
 *
 * 6     spurr_scale_limit   Nominal frequency exceeded 399 percent
 *
 * 7     reserved            reserved
 *
 * 8     xscom_fail          An XSCOM operation caused by a cache inhibited
 *                           load/store from this thread failed. A trap
 *                           register is available.
 *
 * 9     xscom_done          An XSCOM operation caused by a cache inhibited
 *                           load/store from this thread completed. If
 *                           hypervisor intends to use this bit, it is
 *                           responsible for clearing it before performing the
 *                           xscom operation. NOTE: this bit should always be
 *                           masked in HMEER
 *
 * 10    reserved            reserved
 *
 * 11    proc_rcvy_again     Processor recovery occurred again before bit 2
 *                           was cleared
 *
 * 12-15 reserved            reserved
 *
 * 16    scom_fir_hmi        An error inject to PC FIR has occurred to set HMI.
 *                           This error inject can also set FIR(61) to cause
 *                           recovery.
 *
 * 17    reserved            reserved
 *
 * 18    trig_fir_hmi        Debug trigger has occurred to set HMI. This
 *                           trigger can also set FIR(60) to cause recovery
 *
 * 19-20 reserved            reserved
 *
 * 21-23 xscom_status        If bit 8 is active, the reason will be detailed in
 *                           these bits. These bits are information only and
 *                           always masked (mask = ‘0’) If hypervisor intends
 *                           to use this field, it is responsible for clearing
 *                           it before performing the xscom operation.
 *
 * 24:63 Not implemented     Not implemented.
 *
 * P10 HMEER enabled bits:
 * Name                      Action
 * malfunction_alert         Decode and log FIR bits.
 * proc_rcvy_done            Log and continue.
 * tfac_error                Log and attempt to recover time facilities.
 * tfx_error                 Log and attempt to recover time facilities.
 * spurr_scale_limit         Log and continue. XXX?
 * proc_rcvy_again           Log and continue.
 */

/* Used for tracking cpu threads inside hmi handling. */
#define HMI_STATE_CLEANUP_DONE	0x100
#define CORE_THREAD_MASK	0x0ff
#define SUBCORE_THREAD_MASK(s_id, t_count) \
		((((1UL) << (t_count)) - 1) << ((s_id) * (t_count)))
#define SINGLE_THREAD_MASK(t_id)	((1UL) << (t_id))

/*
 * Number of iterations for the various timeouts. We can't use the timebase
 * as it might be broken. We measured experimentally that 40 millions loops
 * of cpu_relax() gives us more than 1s. The margin is comfortable enough.
 */
#define TIMEOUT_LOOPS		40000000

/* TFMR other errors. (other than bit 26 and 45) */
#define SPR_TFMR_OTHER_ERRORS	\
	(SPR_TFMR_TBST_CORRUPT | SPR_TFMR_TB_MISSING_SYNC |	\
	 SPR_TFMR_TB_MISSING_STEP | SPR_TFMR_FW_CONTROL_ERR |	\
	 SPR_TFMR_PURR_PARITY_ERR | SPR_TFMR_SPURR_PARITY_ERR |	\
	 SPR_TFMR_DEC_PARITY_ERR | SPR_TFMR_TFMR_CORRUPT |	\
	 SPR_TFMR_CHIP_TOD_INTERRUPT)

/* TFMR "all core" errors (sent to all threads) */
#define SPR_TFMR_CORE_ERRORS	\
	(SPR_TFMR_TBST_CORRUPT | SPR_TFMR_TB_MISSING_SYNC |	\
	 SPR_TFMR_TB_MISSING_STEP | SPR_TFMR_FW_CONTROL_ERR |	\
	 SPR_TFMR_TFMR_CORRUPT | SPR_TFMR_TB_RESIDUE_ERR |	\
	 SPR_TFMR_HDEC_PARITY_ERROR | SPR_TFMR_TFAC_XFER_ERROR)

/* TFMR "thread" errors  */
#define SPR_TFMR_THREAD_ERRORS \
	(SPR_TFMR_PURR_PARITY_ERR | SPR_TFMR_SPURR_PARITY_ERR |	\
	 SPR_TFMR_DEC_PARITY_ERR)

/*
 * Starting from p9, core inits are setup to escalate all core
 * local checkstop to system checkstop. Review this list when that changes.
 */
static const struct core_xstop_bit_info {
	uint8_t bit;		/* CORE FIR bit number */
	enum OpalHMI_CoreXstopReason reason;
} xstop_bits[] = {
	{ 3, CORE_CHECKSTOP_IFU_REGFILE },
	{ 5, CORE_CHECKSTOP_IFU_LOGIC },
	{ 8, CORE_CHECKSTOP_PC_DURING_RECOV },
	{ 10, CORE_CHECKSTOP_ISU_REGFILE },
	{ 12, CORE_CHECKSTOP_ISU_LOGIC },
	{ 21, CORE_CHECKSTOP_FXU_LOGIC },
	{ 25, CORE_CHECKSTOP_VSU_LOGIC },
	{ 26, CORE_CHECKSTOP_PC_RECOV_IN_MAINT_MODE },
	{ 32, CORE_CHECKSTOP_LSU_REGFILE },
	{ 36, CORE_CHECKSTOP_PC_FWD_PROGRESS },
	{ 38, CORE_CHECKSTOP_LSU_LOGIC },
	{ 45, CORE_CHECKSTOP_PC_LOGIC },
	{ 48, CORE_CHECKSTOP_PC_HYP_RESOURCE },
	{ 52, CORE_CHECKSTOP_PC_HANG_RECOV_FAILED },
	{ 54, CORE_CHECKSTOP_PC_AMBI_HANG_DETECTED },
	{ 63, CORE_CHECKSTOP_PC_SPRD_HYP_ERR_INJ },
};

struct core_fir_bit_info {
	uint8_t bit;		/* CORE FIR bit number */
	const char *reason;
};

static const struct core_fir_bit_info p9_recoverable_bits[] = {
	{ 0, "IFU - SRAM (ICACHE parity, etc)" },
	{ 2, "IFU - RegFile" },
	{ 4, "IFU - Logic" },
	{ 9, "ISU - RegFile" },
	{ 11, "ISU - Logic" },
	{ 13, "ISU - Recoverable due to not in MT window" },
	{ 24, "VSU - Logic" },
	{ 27, "VSU - DFU logic" },
	{ 29, "LSU - SRAM (DCACHE parity, etc)" },
	{ 31, "LSU - RegFile" },
	/* The following 3 bits may be set by SRAM errors. */
	{ 33, "LSU - TLB multi hit" },
	{ 34, "LSU - SLB multi hit" },
	{ 35, "LSU - ERAT multi hit" },
	{ 37, "LSU - Logic" },
	{ 39, "LSU - Recoverable due to not in MT window" },
	{ 43, "PC - Thread hang recovery" },
};

static const struct core_fir_bit_info p10_core_fir_bits[] = {
	{ 0,  "IFU - SRAM recoverable error (ICACHE parity error, etc.)" },
	{ 1,  "PC - TC checkstop" },
	{ 2,  "IFU - RegFile recoverable error" },
	{ 3,  "IFU - RegFile core checkstop" },
	{ 4,  "IFU - Logic recoverable error" },
	{ 5,  "IFU - Logic core checkstop" },
	{ 7,  "VSU - Inference accumulator recoverable error" },
	{ 8,  "PC - Recovery core checkstop" },
	{ 9,  "VSU - Slice Target File (STF) recoverable error" },
	{ 11, "ISU - Logic recoverable error" },
	{ 12, "ISU - Logic core checkstop" },
	{ 14, "ISU - Machine check received while ME=0 checkstop" },
	{ 15, "ISU - UE from L2" },
	{ 16, "ISU - Number of UEs from L2 above threshold" },
	{ 17, "ISU - UE on CI load" },
	{ 18, "MMU - TLB recoverable error" },
	{ 19, "MMU - SLB error" },
	{ 21, "MMU - CXT recoverable error" },
	{ 22, "MMU - Logic core checkstop" },
	{ 23, "MMU - MMU system checkstop" },
	{ 24, "VSU - Logic recoverable error" },
	{ 25, "VSU - Logic core checkstop" },
	{ 26, "PC - In maint mode and recovery in progress" },
	{ 28, "PC - PC system checkstop" },
	{ 29, "LSU - SRAM recoverable error (DCACHE parity error, etc.)" },
	{ 30, "LSU - Set deleted" },
	{ 31, "LSU - RegFile recoverable error" },
	{ 32, "LSU - RegFile core checkstop" },
	{ 33, "MMU - TLB multi hit error occurred" },
	{ 34, "MMU - SLB multi hit error occurred" },
	{ 35, "LSU - ERAT multi hit error occurred" },
	{ 36, "PC - Forward progress error" },
	{ 37, "LSU - Logic recoverable error" },
	{ 38, "LSU - Logic core checkstop" },
	{ 41, "LSU - System checkstop" },
	{ 43, "PC - Thread hang recoverable error" },
	{ 45, "PC - Logic core checkstop" },
	{ 47, "PC - TimeBase facility checkstop" },
	{ 52, "PC - Hang recovery failed core checkstop" },
	{ 53, "PC - Core internal hang detected" },
	{ 55, "PC - Nest hang detected" },
	{ 56, "PC - Other core chiplet recoverable error" },
	{ 57, "PC - Other core chiplet core checkstop" },
	{ 58, "PC - Other core chiplet system checkstop" },
	{ 59, "PC - SCOM satellite error detected" },
	{ 60, "PC - Debug trigger error inject" },
	{ 61, "PC - SCOM or firmware recoverable error inject" },
	{ 62, "PC - Firmware checkstop error inject" },
	{ 63, "PC - Firmware SPRC / SPRD checkstop" },
};

static const struct nx_xstop_bit_info {
	uint8_t bit;		/* NX FIR bit number */
	enum OpalHMI_NestAccelXstopReason reason;
} nx_dma_xstop_bits[] = {
	{ 1, NX_CHECKSTOP_SHM_INVAL_STATE_ERR },
	{ 15, NX_CHECKSTOP_DMA_INVAL_STATE_ERR_1 },
	{ 16, NX_CHECKSTOP_DMA_INVAL_STATE_ERR_2 },
	{ 20, NX_CHECKSTOP_DMA_CH0_INVAL_STATE_ERR },
	{ 21, NX_CHECKSTOP_DMA_CH1_INVAL_STATE_ERR },
	{ 22, NX_CHECKSTOP_DMA_CH2_INVAL_STATE_ERR },
	{ 23, NX_CHECKSTOP_DMA_CH3_INVAL_STATE_ERR },
	{ 24, NX_CHECKSTOP_DMA_CH4_INVAL_STATE_ERR },
	{ 25, NX_CHECKSTOP_DMA_CH5_INVAL_STATE_ERR },
	{ 26, NX_CHECKSTOP_DMA_CH6_INVAL_STATE_ERR },
	{ 27, NX_CHECKSTOP_DMA_CH7_INVAL_STATE_ERR },
	{ 31, NX_CHECKSTOP_DMA_CRB_UE },
	{ 32, NX_CHECKSTOP_DMA_CRB_SUE },
};

static const struct nx_xstop_bit_info nx_pbi_xstop_bits[] = {
	{ 12, NX_CHECKSTOP_PBI_ISN_UE },
};

static struct lock hmi_lock = LOCK_UNLOCKED;
static uint32_t malf_alert_scom;
static uint32_t nx_status_reg;
static uint32_t nx_dma_engine_fir;
static uint32_t nx_pbi_fir;

static int setup_scom_addresses(void)
{
	switch (proc_gen) {
	case proc_gen_p8:
		malf_alert_scom = P8_MALFUNC_ALERT;
		nx_status_reg = P8_NX_STATUS_REG;
		nx_dma_engine_fir = P8_NX_DMA_ENGINE_FIR;
		nx_pbi_fir = P8_NX_PBI_FIR;
		return 1;
	case proc_gen_p9:
		malf_alert_scom = P9_MALFUNC_ALERT;
		nx_status_reg = P9_NX_STATUS_REG;
		nx_dma_engine_fir = P9_NX_DMA_ENGINE_FIR;
		nx_pbi_fir = P9_NX_PBI_FIR;
		return 1;
	case proc_gen_p10:
		malf_alert_scom = P10_MALFUNC_ALERT;
		nx_status_reg = P10_NX_STATUS_REG;
		nx_dma_engine_fir = P10_NX_DMA_ENGINE_FIR;
		nx_pbi_fir = P10_NX_PBI_FIR;
		return 1;
	default:
		prerror("%s: Unknown CPU type\n", __func__);
		break;
	}
	return 0;
}

static int queue_hmi_event(struct OpalHMIEvent *hmi_evt, int recover, uint64_t *out_flags)
{
	size_t size;

	/* Don't queue up event if recover == -1 */
	if (recover == -1)
		return 0;

	/* set disposition */
	if (recover == 1)
		hmi_evt->disposition = OpalHMI_DISPOSITION_RECOVERED;
	else if (recover == 0)
		hmi_evt->disposition = OpalHMI_DISPOSITION_NOT_RECOVERED;

	/*
	 * V2 of struct OpalHMIEvent is of (5 * 64 bits) size and well packed
	 * structure. Hence use uint64_t pointer to pass entire structure
	 * using 5 params in generic message format. Instead of hard coding
	 * num_params divide the struct size by 8 bytes to get exact
	 * num_params value.
	 */
	size = ALIGN_UP(sizeof(*hmi_evt), sizeof(u64));

	*out_flags |= OPAL_HMI_FLAGS_NEW_EVENT;

	/* queue up for delivery to host. */
	return _opal_queue_msg(OPAL_MSG_HMI_EVT, NULL, NULL,
				size, hmi_evt);
}

static int read_core_fir(uint32_t chip_id, uint32_t core_id, uint64_t *core_fir)
{
	int rc;

	switch (proc_gen) {
	case proc_gen_p8:
		rc = xscom_read(chip_id,
			XSCOM_ADDR_P8_EX(core_id, P8_CORE_FIR), core_fir);
		break;
	case proc_gen_p9:
		rc = xscom_read(chip_id,
			XSCOM_ADDR_P9_EC(core_id, P9_CORE_FIR), core_fir);
		break;
	case proc_gen_p10:
		rc = xscom_read(chip_id,
			XSCOM_ADDR_P10_EC(core_id, P10_CORE_FIR), core_fir);
		break;
	default:
		rc = OPAL_HARDWARE;
	}
	return rc;
}

static int read_core_wof(uint32_t chip_id, uint32_t core_id, uint64_t *core_wof)
{
	int rc;

	switch (proc_gen) {
	case proc_gen_p9:
		rc = xscom_read(chip_id,
			XSCOM_ADDR_P9_EC(core_id, P9_CORE_WOF), core_wof);
		break;
	case proc_gen_p10:
		rc = xscom_read(chip_id,
			XSCOM_ADDR_P10_EC(core_id, P10_CORE_WOF), core_wof);
		break;
	default:
		rc = OPAL_HARDWARE;
	}
	return rc;
}

static bool decode_core_fir(struct cpu_thread *cpu,
				struct OpalHMIEvent *hmi_evt)
{
	uint64_t core_fir;
	uint32_t core_id;
	int i, swkup_rc;
	bool found = false;
	int64_t ret;
	const char *loc;

	/* Sanity check */
	if (!cpu || !hmi_evt)
		return false;

	core_id = pir_to_core_id(cpu->pir);

	/* Force the core to wakeup, otherwise reading core_fir is unrealiable
	 * if stop-state 5 is enabled.
	 */
	swkup_rc = dctl_set_special_wakeup(cpu);

	/* Get CORE FIR register value. */
	ret = read_core_fir(cpu->chip_id, core_id, &core_fir);

	if (!swkup_rc)
		dctl_clear_special_wakeup(cpu);


	if (ret == OPAL_WRONG_STATE) {
		/*
		 * CPU is asleep, so it probably didn't cause the checkstop.
		 * If no other HMI cause is found a "catchall" checkstop
		 * will be raised, so if this CPU should've been awake the
		 * error will be handled appropriately.
		 */
		prlog(PR_DEBUG,
		      "FIR read failed, chip %d core %d asleep\n",
		      cpu->chip_id, core_id);
		return false;
	} else if (ret != OPAL_SUCCESS) {
		prerror("XSCOM error reading CORE FIR\n");
		/* If the FIR can't be read, we should checkstop. */
		return true;
	}

	if (!core_fir)
		return false;

	loc = chip_loc_code(cpu->chip_id);
	prlog(PR_INFO, "[Loc: %s]: CHIP ID: %x, CORE ID: %x, FIR: %016llx\n",
			loc ? loc : "Not Available",
			cpu->chip_id, core_id, core_fir);

	if (proc_gen == proc_gen_p10) {
		for (i = 0; i < ARRAY_SIZE(p10_core_fir_bits); i++) {
			if (core_fir & PPC_BIT(p10_core_fir_bits[i].bit))
				prlog(PR_INFO, "    %s\n", p10_core_fir_bits[i].reason);
		}
	}

	/* Check CORE FIR bits and populate HMI event with error info. */
	for (i = 0; i < ARRAY_SIZE(xstop_bits); i++) {
		if (core_fir & PPC_BIT(xstop_bits[i].bit)) {
			found = true;
			hmi_evt->u.xstop_error.xstop_reason
					|= cpu_to_be32(xstop_bits[i].reason);
		}
	}
	return found;
}

static void find_core_checkstop_reason(struct OpalHMIEvent *hmi_evt,
				       uint64_t *out_flags)
{
	struct cpu_thread *cpu;

	/* Initialize HMI event */
	hmi_evt->severity = OpalHMI_SEV_FATAL;
	hmi_evt->type = OpalHMI_ERROR_MALFUNC_ALERT;
	hmi_evt->u.xstop_error.xstop_type = CHECKSTOP_TYPE_CORE;

	/*
	 * Check CORE FIRs and find the reason for core checkstop.
	 * Send a separate HMI event for each core that has checkstopped.
	 */
	for_each_cpu(cpu) {
		/* GARDed CPUs are marked unavailable. Skip them.  */
		if (cpu->state == cpu_state_unavailable)
			continue;

		/* Only check on primaries (ie. core), not threads */
		if (cpu->is_secondary)
			continue;

		/* Initialize xstop_error fields. */
		hmi_evt->u.xstop_error.xstop_reason = 0;
		hmi_evt->u.xstop_error.u.pir = cpu_to_be32(cpu->pir);

		if (decode_core_fir(cpu, hmi_evt))
			queue_hmi_event(hmi_evt, 0, out_flags);
	}
}

static void find_capp_checkstop_reason(int flat_chip_id,
				       struct OpalHMIEvent *hmi_evt,
				       uint64_t *out_flags)
{
	struct capp_info info;
	struct phb *phb;
	uint64_t capp_fir;
	uint64_t capp_fir_mask;
	uint64_t capp_fir_action0;
	uint64_t capp_fir_action1;
	uint64_t reg;
	int64_t rc;

	/* CAPP exists on P8 and P9 only */
	if (proc_gen != proc_gen_p8 && proc_gen != proc_gen_p9)
		return;

	/* Find the CAPP on the chip associated with the HMI. */
	for_each_phb(phb) {
		/* get the CAPP info */
		rc = capp_get_info(flat_chip_id, phb, &info);
		if (rc == OPAL_PARAMETER)
			continue;

		if (xscom_read(flat_chip_id, info.capp_fir_reg, &capp_fir) ||
		    xscom_read(flat_chip_id, info.capp_fir_mask_reg,
			       &capp_fir_mask) ||
		    xscom_read(flat_chip_id, info.capp_fir_action0_reg,
			       &capp_fir_action0) ||
		    xscom_read(flat_chip_id, info.capp_fir_action1_reg,
			       &capp_fir_action1)) {
			prerror("CAPP: Couldn't read CAPP#%d (PHB:#%x) FIR registers by XSCOM!\n",
				info.capp_index, info.phb_index);
			continue;
		}

		if (!(capp_fir & ~capp_fir_mask))
			continue;

		prlog(PR_DEBUG, "CAPP#%d (PHB:#%x): FIR 0x%016llx mask 0x%016llx\n",
		      info.capp_index, info.phb_index, capp_fir,
		      capp_fir_mask);
		prlog(PR_DEBUG, "CAPP#%d (PHB:#%x): ACTION0 0x%016llx, ACTION1 0x%016llx\n",
		      info.capp_index, info.phb_index, capp_fir_action0,
		      capp_fir_action1);

		/*
		 * If this bit is set (=1) a Recoverable Error has been
		 * detected
		 */
		xscom_read(flat_chip_id, info.capp_err_status_ctrl_reg, &reg);
		if ((reg & PPC_BIT(0)) != 0) {
			phb_lock(phb);
			phb->ops->set_capp_recovery(phb);
			phb_unlock(phb);

			hmi_evt->severity = OpalHMI_SEV_NO_ERROR;
			hmi_evt->type = OpalHMI_ERROR_CAPP_RECOVERY;
			queue_hmi_event(hmi_evt, 1, out_flags);

			return;
		}
	}
}

static void find_nx_checkstop_reason(int flat_chip_id,
				     struct OpalHMIEvent *hmi_evt,
				     uint64_t *out_flags)
{
	uint64_t nx_status;
	uint64_t nx_dma_fir;
	uint64_t nx_pbi_fir_val;
	int i;

	/* Get NX status register value. */
	if (xscom_read(flat_chip_id, nx_status_reg, &nx_status) != 0) {
		prerror("XSCOM error reading NX_STATUS_REG\n");
		return;
	}

	/* Check if NX has driven an HMI interrupt. */
	if (!(nx_status & NX_HMI_ACTIVE))
		return;

	/* Initialize HMI event */
	hmi_evt->severity = OpalHMI_SEV_FATAL;
	hmi_evt->type = OpalHMI_ERROR_MALFUNC_ALERT;
	hmi_evt->u.xstop_error.xstop_type = CHECKSTOP_TYPE_NX;
	hmi_evt->u.xstop_error.u.chip_id = cpu_to_be32(flat_chip_id);

	/* Get DMA & Engine FIR data register value. */
	if (xscom_read(flat_chip_id, nx_dma_engine_fir, &nx_dma_fir) != 0) {
		prerror("XSCOM error reading NX_DMA_ENGINE_FIR\n");
		return;
	}

	/* Get PowerBus Interface FIR data register value. */
	if (xscom_read(flat_chip_id, nx_pbi_fir, &nx_pbi_fir_val) != 0) {
		prerror("XSCOM error reading NX_PBI_FIR\n");
		return;
	}

	/* Find NX checkstop reason and populate HMI event with error info. */
	for (i = 0; i < ARRAY_SIZE(nx_dma_xstop_bits); i++)
		if (nx_dma_fir & PPC_BIT(nx_dma_xstop_bits[i].bit))
			hmi_evt->u.xstop_error.xstop_reason
				|= cpu_to_be32(nx_dma_xstop_bits[i].reason);

	for (i = 0; i < ARRAY_SIZE(nx_pbi_xstop_bits); i++)
		if (nx_pbi_fir_val & PPC_BIT(nx_pbi_xstop_bits[i].bit))
			hmi_evt->u.xstop_error.xstop_reason
				|= cpu_to_be32(nx_pbi_xstop_bits[i].reason);

	/*
	 * Set NXDMAENGFIR[38] to signal PRD that service action is required.
	 * Without this inject, PRD will not be able to do NX unit checkstop
	 * error analysis. NXDMAENGFIR[38] is a spare bit and used to report
	 * a software initiated attention.
	 *
	 * The behavior of this bit and all FIR bits are documented in
	 * RAS spreadsheet.
	 */
	xscom_write(flat_chip_id, nx_dma_engine_fir, PPC_BIT(38));

	/* Send an HMI event. */
	queue_hmi_event(hmi_evt, 0, out_flags);
}

static void add_npu_xstop_reason(uint32_t *xstop_reason, uint8_t reason)
{
	int i, reason_count;
	uint8_t *ptr;

	reason_count = sizeof(*xstop_reason) / sizeof(reason);
	ptr = (uint8_t *) xstop_reason;
	for (i = 0; i < reason_count; i++) {
		if (*ptr == 0) {
			*ptr = reason;
			break;
		}
		ptr++;
	}
}

static void encode_npu_xstop_reason(uint32_t *xstop_reason,
				    uint64_t fir, int fir_number)
{
	int bit;
	uint8_t reason;

	/*
	 * There are three 64-bit FIRs but the xstop reason field of
	 * the hmi event is only 32-bit. Encode which FIR bit is set as:
	 * - 2 bits for the FIR number
	 * - 6 bits for the bit number (0 -> 63)
	 *
	 * So we could even encode up to 4 reasons for the HMI, if
	 * that can ever happen
	 */
	while (fir) {
		bit = ilog2(fir);
		reason = fir_number << 6;
		reason |= (63 - bit); // IBM numbering
		add_npu_xstop_reason(xstop_reason, reason);
		fir ^= 1ULL << bit;
	}
}

static bool npu_fir_errors(struct phb *phb, int flat_chip_id,
			   uint32_t *xstop_reason)
{
	uint64_t fir, fir_mask;
	uint64_t fir_action0, fir_action1;
	uint64_t fir_reg, fir_mask_reg;
	uint64_t fir_action0_reg, fir_action1_reg;
	uint64_t fatal_errors;
	uint64_t xscom_base;
	bool fir_errors = false;
	int fir_regs;
	const char *loc;
	struct npu *npu;
	struct npu2 *npu2 = NULL;
	struct npu2_dev *dev;
	struct pau *pau = NULL;

	fir_regs = (phb->phb_type == phb_type_pcie_v3) ? 1 : 3;

	for (uint32_t i = 0; i < fir_regs; i++) {
		switch (phb->phb_type) {
		case phb_type_pcie_v3:
			fir_reg = NX_FIR;
			fir_mask_reg = NX_FIR_MASK;
			fir_action0_reg = NX_FIR_ACTION0;
			fir_action1_reg = NX_FIR_ACTION1;

			npu = phb_to_npu(phb);
			if (npu != NULL)
				xscom_base = npu->at_xscom;
			else
				continue;
		break;
		case phb_type_npu_v2:
			fir_reg = NPU2_FIR(i);
			fir_mask_reg = NPU2_FIR_MASK(i);
			fir_action0_reg = NPU2_FIR_ACTION0(i);
			fir_action1_reg = NPU2_FIR_ACTION1(i);
			npu2 = phb_to_npu2_nvlink(phb);
			xscom_base = npu2->xscom_base;
		break;
		case phb_type_npu_v2_opencapi:
			fir_reg = NPU2_FIR(i);
			fir_mask_reg = NPU2_FIR_MASK(i);
			fir_action0_reg = NPU2_FIR_ACTION0(i);
			fir_action1_reg = NPU2_FIR_ACTION1(i);
			dev = phb_to_npu2_dev_ocapi(phb);
			npu2 = dev->npu;
			xscom_base = npu2->xscom_base;
		break;
		case phb_type_pau_opencapi:
			fir_reg = PAU_FIR(i);
			fir_mask_reg = PAU_FIR_MASK(i);
			fir_action0_reg = PAU_FIR_ACTION0(i);
			fir_action1_reg = PAU_FIR_ACTION1(i);
			pau = ((struct pau_dev *)(pau_phb_to_opencapi_dev(phb)))->pau;
			xscom_base = pau->xscom_base;
		break;
		default:
			continue;
		}

		if (xscom_read(flat_chip_id, xscom_base + fir_reg, &fir) ||
		    xscom_read(flat_chip_id, xscom_base + fir_mask_reg, &fir_mask) ||
		    xscom_read(flat_chip_id, xscom_base + fir_action0_reg, &fir_action0) ||
		    xscom_read(flat_chip_id, xscom_base + fir_action1_reg, &fir_action1)) {
			prerror("HMI: Couldn't read NPU/PAU FIR register%d with XSCOM\n", i);
			continue;
		}

		fatal_errors = fir & ~fir_mask & fir_action0 & fir_action1;

		if (fatal_errors) {
			loc = chip_loc_code(flat_chip_id);
			if (!loc)
				loc = "Not Available";
			prlog(PR_ERR, "NPU/PAU: [Loc: %s] P:%d FIR#%d "
				      "FIR 0x%016llx mask 0x%016llx\n",
				      loc, flat_chip_id, i, fir, fir_mask);
			prlog(PR_ERR, "NPU/PAU: [Loc: %s] P:%d ACTION0 "
				      "0x%016llx, ACTION1 0x%016llx\n",
				      loc, flat_chip_id, fir_action0, fir_action1);
			if (phb->phb_type != phb_type_pcie_v3)
				encode_npu_xstop_reason(xstop_reason,
							fatal_errors,
							i);
			fir_errors = true;
		}
	}

	/* dump registers */
	if (fir_errors) {
		switch (phb->phb_type) {
		case phb_type_npu_v2:
		case phb_type_npu_v2_opencapi:
			npu2_dump_scoms(npu2, flat_chip_id);
		break;
		case phb_type_pau_opencapi:
			pau_opencapi_dump_scoms(pau);
		break;
		default:
		break;
		}

		prlog(PR_ERR, " _________________________ \n");
		prlog(PR_ERR, "< It's Debug time!        >\n");
		prlog(PR_ERR, " ------------------------- \n");
		prlog(PR_ERR, "       \\   ,__,           \n");
		prlog(PR_ERR, "        \\  (oo)____       \n");
		prlog(PR_ERR, "           (__)    )\\     \n");
		prlog(PR_ERR, "              ||--|| *     \n");
	}

	return fir_errors;
}

static void find_npu_checkstop_reason(int flat_chip_id,
				      struct OpalHMIEvent *hmi_evt,
				      uint64_t *out_flags)
{
	struct phb *phb;
	struct dt_node *dn;
	uint32_t xstop_reason = 0;

	/* Only check for NPU errors if the chip has a NPU/PAU */
	if ((PVR_TYPE(mfspr(SPR_PVR)) != PVR_TYPE_P8NVL) &&
	    (PVR_TYPE(mfspr(SPR_PVR)) != PVR_TYPE_P9) &&
	    (PVR_TYPE(mfspr(SPR_PVR)) != PVR_TYPE_P10))
		return;

	/* Find the NPU/PAU on the chip associated with the HMI. */
	for_each_phb(phb) {
		dn = phb->dt_node;

		if (!(dt_node_is_compatible(dn, "ibm,power8-npu-pciex") ||
		      dt_node_is_compatible(dn, "ibm,power9-npu-pciex") ||
		      dt_node_is_compatible(dn, "ibm,power9-npu-opencapi-pciex") ||
		      dt_node_is_compatible(dn, "ibm,power10-pau-opencapi-pciex")))
			continue;

		if (dt_get_chip_id(dn) != flat_chip_id)
			continue;

		/* Read all the registers necessary to find a checkstop condition. */
		if (!npu_fir_errors(phb, flat_chip_id, &xstop_reason))
			continue;

		if (phb->phb_type == phb_type_pcie_v3) {
			/* Set the NPU to fenced since it can't recover. */
			npu_set_fence_state(phb_to_npu(phb), true);
		}

		/* Set up the HMI event */
		hmi_evt->severity = OpalHMI_SEV_WARNING;
		hmi_evt->type = OpalHMI_ERROR_MALFUNC_ALERT;
		hmi_evt->u.xstop_error.xstop_type = CHECKSTOP_TYPE_NPU;
		hmi_evt->u.xstop_error.xstop_reason = xstop_reason;
		hmi_evt->u.xstop_error.u.chip_id = cpu_to_be32(flat_chip_id);

		/* Marking the event as recoverable so that we don't crash */
		queue_hmi_event(hmi_evt, 1, out_flags);
	}
}

static void decode_malfunction(struct OpalHMIEvent *hmi_evt, uint64_t *out_flags)
{
	int i;
	uint64_t malf_alert, flags;

	flags = 0;

	if (!setup_scom_addresses()) {
		prerror("Failed to setup scom addresses\n");
		/* Send an unknown HMI event. */
		hmi_evt->u.xstop_error.xstop_type = CHECKSTOP_TYPE_UNKNOWN;
		hmi_evt->u.xstop_error.xstop_reason = 0;
		queue_hmi_event(hmi_evt, false, out_flags);
		return;
	}

	xscom_read(this_cpu()->chip_id, malf_alert_scom, &malf_alert);

	if (!malf_alert)
		return;

	for (i = 0; i < 64; i++) {
		if (malf_alert & PPC_BIT(i)) {
			xscom_write(this_cpu()->chip_id, malf_alert_scom,
								~PPC_BIT(i));
			find_capp_checkstop_reason(i, hmi_evt, &flags);
			if (proc_gen != proc_gen_p10)
				find_nx_checkstop_reason(i, hmi_evt, &flags);
			find_npu_checkstop_reason(i, hmi_evt, &flags);
		}
	}

	find_core_checkstop_reason(hmi_evt, &flags);

	/*
	 * If we fail to find checkstop reason, send an unknown HMI event.
	 */
	if (!(flags & OPAL_HMI_FLAGS_NEW_EVENT)) {
		hmi_evt->u.xstop_error.xstop_type = CHECKSTOP_TYPE_UNKNOWN;
		hmi_evt->u.xstop_error.xstop_reason = 0;
		queue_hmi_event(hmi_evt, false, &flags);
	}
	*out_flags |= flags;
}

/*
 * This will "rendez-vous" all threads on the core to the rendez-vous
 * id "sig". You need to make sure that "sig" is different from the
 * previous rendez vous. The sig value must be between 0 and 7 with
 * boot time being set to 0.
 *
 * Note: in theory, we could just use a flip flop "sig" in the thread
 * structure (binary rendez-vous with no argument). This is a bit more
 * debuggable and better at handling timeouts (arguably).
 *
 * This should be called with the no lock held
 */
static void hmi_rendez_vous(uint32_t sig)
{
	struct cpu_thread *t = this_cpu();
	uint32_t my_id = cpu_get_thread_index(t);
	uint32_t my_shift = my_id << 2;
	uint32_t *sptr = t->core_hmi_state_ptr;
	uint32_t val, prev, shift, i;
	uint64_t timeout;

	assert(sig <= 0x7);

	/*
	 * Mark ourselves as having reached the rendez vous point with
	 * the exit bit cleared
	 */
	do {
		val = prev = *sptr;
		val &= ~(0xfu << my_shift);
		val |= sig << my_shift;
	} while (cmpxchg32(sptr, prev, val) != prev);

	/*
	 * Wait for everybody else to reach that point, ignore the
	 * exit bit as another thread could have already set it.
	 */
	for (i = 0; i < cpu_thread_count; i++) {
		shift = i << 2;

		timeout = TIMEOUT_LOOPS;
		while (((*sptr >> shift) & 0x7) != sig && --timeout)
			cpu_relax();
		if (!timeout)
			prlog(PR_ERR, "Rendez-vous stage 1 timeout, CPU 0x%x"
			      " waiting for thread %d (sptr=%08x)\n",
						      t->pir, i, *sptr);
	}

	/* Set the exit bit */
	do {
		val = prev = *sptr;
		val &= ~(0xfu << my_shift);
		val |= (sig | 8) << my_shift;
	} while (cmpxchg32(sptr, prev, val) != prev);

	/* At this point, we need to wait for everybody else to have a value
	 * that is *not* sig. IE. they either have set the exit bit *or* they
	 * have changed the rendez-vous (meaning they have moved on to another
	 * rendez vous point).
	 */
	for (i = 0; i < cpu_thread_count; i++) {
		shift = i << 2;

		timeout = TIMEOUT_LOOPS;
		while (((*sptr >> shift) & 0xf) == sig && --timeout)
			cpu_relax();
		if (!timeout)
			prlog(PR_ERR, "Rendez-vous stage 2 timeout, CPU 0x%x"
			      " waiting for thread %d (sptr=%08x)\n",
						      t->pir, i, *sptr);
	}
}

static void hmi_print_debug(const uint8_t *msg, uint64_t hmer)
{
	const char *loc;
	uint32_t core_id, thread_index;

	core_id = pir_to_core_id(this_cpu()->pir);
	thread_index = cpu_get_thread_index(this_cpu());

	loc = chip_loc_code(this_cpu()->chip_id);
	if (!loc)
		loc = "Not Available";

	/* Also covers P10 SPR_HMER_TFAC_SHADOW_XFER_ERROR */
	if (hmer & (SPR_HMER_TFAC_ERROR | SPR_HMER_TFMR_PARITY_ERROR)) {
		prlog(PR_DEBUG, "[Loc: %s]: P:%d C:%d T:%d: TFMR(%016lx) %s\n",
			loc, this_cpu()->chip_id, core_id, thread_index,
			mfspr(SPR_TFMR), msg);
	} else {
		prlog(PR_DEBUG, "[Loc: %s]: P:%d C:%d T:%d: %s\n",
			loc, this_cpu()->chip_id, core_id, thread_index,
			msg);
	}
}

static int handle_thread_tfac_error(uint64_t tfmr, uint64_t *out_flags)
{
	int recover = 1;

	if (tfmr & SPR_TFMR_DEC_PARITY_ERR)
		*out_flags |= OPAL_HMI_FLAGS_DEC_LOST;
	if (!tfmr_recover_local_errors(tfmr))
		recover = 0;
	tfmr &= ~(SPR_TFMR_PURR_PARITY_ERR |
		  SPR_TFMR_SPURR_PARITY_ERR |
		  SPR_TFMR_DEC_PARITY_ERR);
	return recover;
}

static int64_t opal_handle_hmi(void);

static void opal_handle_hmi_job(void *data __unused)
{
	opal_handle_hmi();
}

/*
 * Queue hmi handling job If secondaries are still in OPAL
 * This function is called by thread 0.
 */
static struct cpu_job **hmi_kick_secondaries(void)
{
	struct cpu_thread *ts = this_cpu();
	struct cpu_job **hmi_jobs = NULL;
	int job_sz = sizeof(struct cpu_job *) * cpu_thread_count;
	int i;

	for (i = 1; i < cpu_thread_count; i++) {
		ts = next_cpu(ts);

		/* Is this thread still in OPAL ? */
		if (ts->state == cpu_state_active) {
			if (!hmi_jobs) {
				hmi_jobs = zalloc(job_sz);
				assert(hmi_jobs);
			}

			prlog(PR_DEBUG, "Sending hmi job to thread %d\n", i);
			hmi_jobs[i] = cpu_queue_job(ts, "handle_hmi_job",
					opal_handle_hmi_job, NULL);
		}
	}
	return hmi_jobs;
}

static int handle_all_core_tfac_error(uint64_t tfmr, uint64_t *out_flags)
{
	struct cpu_thread *t, *t0;
	int recover = -1;
	struct cpu_job **hmi_jobs = NULL;
	bool hmi_with_no_error = false;

	t = this_cpu();
	t0 = find_cpu_by_pir(cpu_get_thread0(t));

	if (t == t0 && t0->state == cpu_state_os)
		hmi_jobs = hmi_kick_secondaries();

	/*
	 * Handle special case: If TB is in invalid state and no TB error
	 * reported in TFMR for this HMI, then treat this as TFMR corrupt error
	 * to force the recovery procedure recover_corrupt_tfmr(). This will
	 * also reset the core level TB erorrs including Missing step. Do this
	 * only on thread 0, otherwise every thread will repeat the same
	 * procedure unnecessarily.
	 */
	if (t == t0 && !(tfmr & SPR_TFMR_CORE_ERRORS) && this_cpu()->tb_invalid) {
		tfmr |= SPR_TFMR_TFMR_CORRUPT;
		hmi_with_no_error = true;
	}

	/* Rendez vous all threads */
	hmi_rendez_vous(1);

	/* We use a lock here as some of the TFMR bits are shared and I
	 * prefer avoiding doing the cleanup simultaneously.
	 */
	lock(&hmi_lock);

	/* First handle corrupt TFMR otherwise we can't trust anything.
	 * We'll use a lock here so that the threads don't try to do it at
	 * the same time
	 */
	if (tfmr & SPR_TFMR_TFMR_CORRUPT) {
		/* Check if it's still in error state */
		if (hmi_with_no_error || mfspr(SPR_TFMR) & SPR_TFMR_TFMR_CORRUPT)
			if (!recover_corrupt_tfmr()) {
				unlock(&hmi_lock);
				recover = 0;
				goto error_out;
			}

		tfmr = mfspr(SPR_TFMR);

		/* We could have got new thread errors in the meantime */
		if (tfmr & SPR_TFMR_THREAD_ERRORS) {
			recover = handle_thread_tfac_error(tfmr, out_flags);
			tfmr &= ~SPR_TFMR_THREAD_ERRORS;
		}
		if (!recover) {
			unlock(&hmi_lock);
			goto error_out;
		}
	}

	/* Tell the OS ... */
	if (tfmr & SPR_TFMR_HDEC_PARITY_ERROR)
		*out_flags |= OPAL_HMI_FLAGS_HDEC_LOST;

	/* Cleanup bad HDEC or TB on all threads or subcures before we clear
	 * the error conditions
	 */
	tfmr_cleanup_core_errors(tfmr);

	/* Unlock before next rendez-vous */
	unlock(&hmi_lock);

	/* Second rendez vous, ensure the above cleanups are all done before
	 * we proceed further
	 */
	hmi_rendez_vous(2);

	/* We can now clear the error conditions in the core. */
	recover = tfmr_clear_core_errors(tfmr);
	if (recover == 0)
		goto error_out;

	/* Third rendez-vous. We could in theory do the timebase resync as
	 * part of the previous one, but I prefer having all the error
	 * conditions cleared before we start trying.
	 */
	hmi_rendez_vous(3);

	/* Now perform the actual TB recovery on thread 0 */
	if (t == t0)
		recover = chiptod_recover_tb_errors(&this_cpu()->tb_resynced);

error_out:
	/* Last rendez-vous */
	hmi_rendez_vous(4);

	/* Now all threads have gone past rendez-vous 3 and not yet past another
	 * rendez-vous 1, so the value of tb_resynced of thread 0 of the core
	 * contains an accurate indication as to whether the timebase was lost.
	 */
	if (t0->tb_resynced)
		*out_flags |= OPAL_HMI_FLAGS_TB_RESYNC;

	if (t == t0 && hmi_jobs) {
		int i;
		for (i = 1; i < cpu_thread_count; i++)
			if (hmi_jobs[i])
				cpu_wait_job(hmi_jobs[i], true);
		free(hmi_jobs);
	}

	return recover;
}

static uint64_t read_tfmr_t0(void)
{
	uint64_t tfmr_t0;
	uint32_t chip_id = this_cpu()->chip_id;
	uint32_t core_id = pir_to_core_id(this_cpu()->pir);

	lock(&hmi_lock);

	xscom_write(chip_id, XSCOM_ADDR_P9_EC(core_id, P9_SCOM_SPRC),
			SETFIELD(P9_SCOMC_SPR_SELECT, 0, P9_SCOMC_TFMR_T0));
	xscom_read(chip_id, XSCOM_ADDR_P9_EC(core_id, P9_SCOM_SPRD),
				&tfmr_t0);
	unlock(&hmi_lock);
	return tfmr_t0;
}

/* P9 errata: In theory, an HDEC error is sent to all threads. However,
 * due to an errata on P9 where TFMR bit 26 (HDEC parity) cannot be
 * cleared on thread 1..3, I am not confident we can do a rendez-vous
 * in all cases.
 *
 * Our current approach is to ignore that error unless it is present
 * on thread 0 TFMR. Also, ignore TB residue error due to a similar
 * errata as above.
 */
static void validate_latched_errors(uint64_t *tfmr)
{
	if ((*tfmr & (SPR_TFMR_HDEC_PARITY_ERROR | SPR_TFMR_TB_RESIDUE_ERR))
				&& this_cpu()->is_secondary) {
		uint64_t tfmr_t0 = read_tfmr_t0();

		if (!(tfmr_t0 & SPR_TFMR_HDEC_PARITY_ERROR))
			*tfmr &= ~SPR_TFMR_HDEC_PARITY_ERROR;

		if (!(tfmr_t0 & SPR_TFMR_TB_RESIDUE_ERR))
			*tfmr &= ~SPR_TFMR_TB_RESIDUE_ERR;
	}
}

static int handle_tfac_errors(struct OpalHMIEvent *hmi_evt, uint64_t *out_flags)
{
	int recover = -1;
	uint64_t tfmr = mfspr(SPR_TFMR);

	/* Initialize the hmi event with old value of TFMR */
	hmi_evt->tfmr = cpu_to_be64(tfmr);

	/* A TFMR parity/corrupt error makes us ignore all the local stuff.*/
	if (tfmr & SPR_TFMR_TFMR_CORRUPT) {
		/* Mark TB as invalid for now as we don't trust TFMR, we'll fix
		 * it up later
		 */
		this_cpu()->tb_invalid = true;
		goto bad_tfmr;
	}

	this_cpu()->tb_invalid = !(tfmr & SPR_TFMR_TB_VALID);

	if (proc_gen == proc_gen_p9)
		validate_latched_errors(&tfmr);

	/* First, handle thread local errors */
	if (tfmr & SPR_TFMR_THREAD_ERRORS) {
		recover = handle_thread_tfac_error(tfmr, out_flags);
		tfmr &= ~SPR_TFMR_THREAD_ERRORS;
	}

 bad_tfmr:

	/* Let's see if we still have a all-core error to deal with, if
	 * not, we just bail out
	 */
	if (tfmr & SPR_TFMR_CORE_ERRORS) {
		int recover2;

		/* Only update "recover" if it's not already 0 (non-recovered)
		 */
		recover2 = handle_all_core_tfac_error(tfmr, out_flags);
		if (recover != 0)
			recover = recover2;
	} else if (tfmr & SPR_TFMR_CHIP_TOD_INTERRUPT) {
		int recover2;

		/*
		 * There are some TOD errors which do not affect working of
		 * TOD and TB. They stay in valid state. Hence we don't need
		 * rendez vous.
		 *
		 * TOD errors that affects TOD/TB will report a global error
		 * on TFMR alongwith bit 51, and they will go in rendez vous.
		 */
		recover2 = chiptod_recover_tod_errors();
		if (recover != 0)
			recover = recover2;
	} else if (this_cpu()->tb_invalid) {
		int recover2;

		/*
		 * This shouldn't happen, TB is invalid and no global error was
		 * reported. However, On p10, in a very rare situation when
		 * core is waking up from stop2 or higher stop state, timer
		 * facility goes into error state due to Missing step, causing
		 * an HMI with no error reason set in TFMR register other than
		 * TFMR[41]=0 (tb_valid) and TFMR[28:31]=9 (tbst_encoded).
		 * Ideally, "Missing step" error should be reported in
		 * TFMR[44]=1. It looks like in this rare case, while
		 * generating HMI, HW fails to sync up the TFMR register with
		 * the core which is waking up from stop2.
		 *
		 * To be able to recover, follow down to recovery method as if
		 * we got core level TB error and treat this as TFMR corrupt
		 * error and reset all core errors including Missing step.
		 */

		prlog(PR_ERR, "HMI: TB invalid without core error reported ! "
			"CPU=%x, TFMR=0x%016lx\n", this_cpu()->pir,
						mfspr(SPR_TFMR));
		recover2 = handle_all_core_tfac_error(tfmr, out_flags);
		if (recover != 0)
			recover = recover2;
	}

	if (recover != -1 && hmi_evt) {
		hmi_evt->severity = OpalHMI_SEV_ERROR_SYNC;
		hmi_evt->type = OpalHMI_ERROR_TFAC;
		queue_hmi_event(hmi_evt, recover, out_flags);
	}

	/* Set the TB state looking at TFMR register before we head out. */
	this_cpu()->tb_invalid = !(mfspr(SPR_TFMR) & SPR_TFMR_TB_VALID);

	if (this_cpu()->tb_invalid) {
		*out_flags |= OPAL_HMI_FLAGS_TOD_TB_FAIL;
		prlog(PR_WARNING, "Failed to get TB in running state! "
			"CPU=%x, TFMR=%016lx\n", this_cpu()->pir,
					mfspr(SPR_TFMR));
	}

	return recover;
}

static int handle_hmi_exception(uint64_t hmer, struct OpalHMIEvent *hmi_evt,
				uint64_t *out_flags)
{
	struct cpu_thread *cpu = this_cpu();
	int recover = 1;
	uint64_t handled = 0;

	prlog(PR_DEBUG, "Received HMI interrupt: HMER = 0x%016llx\n", hmer);
	/* Initialize the hmi event with old value of HMER */
	if (hmi_evt)
		hmi_evt->hmer = cpu_to_be64(hmer);

	/* Handle Timer/TOD errors separately */
	if (hmer & (SPR_HMER_TFAC_ERROR | SPR_HMER_TFMR_PARITY_ERROR)) {
		hmi_print_debug("Timer Facility Error", hmer);
		handled = hmer & (SPR_HMER_TFAC_ERROR | SPR_HMER_TFMR_PARITY_ERROR);
		mtspr(SPR_HMER, ~handled);
		recover = handle_tfac_errors(hmi_evt, out_flags);
		handled = 0;
	}

	lock(&hmi_lock);
	/*
	 * Not all HMIs would move TB into invalid state. Set the TB state
	 * looking at TFMR register. TFMR will tell us correct state of
	 * TB register.
	 */
	if (hmer & SPR_HMER_PROC_RECV_DONE) {
		uint32_t chip_id = pir_to_chip_id(cpu->pir);
		uint32_t core_id = pir_to_core_id(cpu->pir);
		uint64_t core_wof;

		hmi_print_debug("Processor recovery occurred.", hmer);
		if (!read_core_wof(chip_id, core_id, &core_wof)) {
			int i;

			prlog(PR_DEBUG, "Core WOF = 0x%016llx recovered error:\n", core_wof);
			if (proc_gen <= proc_gen_p9) {
				for (i = 0; i < ARRAY_SIZE(p9_recoverable_bits); i++) {
					if (core_wof & PPC_BIT(p9_recoverable_bits[i].bit))
						prlog(PR_DEBUG, "    %s\n", p9_recoverable_bits[i].reason);
				}
			} else if (proc_gen == proc_gen_p10) {
				for (i = 0; i < ARRAY_SIZE(p10_core_fir_bits); i++) {
					if (core_wof & PPC_BIT(p10_core_fir_bits[i].bit))
						prlog(PR_DEBUG, "    %s\n", p10_core_fir_bits[i].reason);
				}
			}
		}

		handled |= SPR_HMER_PROC_RECV_DONE;
		if (cpu_is_thread0(cpu) && hmi_evt) {
			hmi_evt->severity = OpalHMI_SEV_NO_ERROR;
			hmi_evt->type = OpalHMI_ERROR_PROC_RECOV_DONE;
			queue_hmi_event(hmi_evt, recover, out_flags);
		}
	}

	if ((proc_gen <= proc_gen_p9) && (hmer & SPR_HMER_PROC_RECV_ERROR_MASKED)) {
		handled |= SPR_HMER_PROC_RECV_ERROR_MASKED;
		if (cpu_is_thread0(cpu) && hmi_evt) {
			hmi_evt->severity = OpalHMI_SEV_NO_ERROR;
			hmi_evt->type = OpalHMI_ERROR_PROC_RECOV_MASKED;
			queue_hmi_event(hmi_evt, recover, out_flags);
		}
		hmi_print_debug("Processor recovery Done (masked).", hmer);
	}

	if (hmer & SPR_HMER_PROC_RECV_AGAIN) {
		handled |= SPR_HMER_PROC_RECV_AGAIN;
		if (cpu_is_thread0(cpu) && hmi_evt) {
			hmi_evt->severity = OpalHMI_SEV_NO_ERROR;
			hmi_evt->type = OpalHMI_ERROR_PROC_RECOV_DONE_AGAIN;
			queue_hmi_event(hmi_evt, recover, out_flags);
		}
		hmi_print_debug("Processor recovery occurred again before"
				"bit2 was cleared\n", hmer);
	}

	/* XXX: what to do with this? */
	if (hmer & SPR_HMER_SPURR_SCALE_LIMIT) {
		handled |= SPR_HMER_SPURR_SCALE_LIMIT;
		if (cpu_is_thread0(cpu) && hmi_evt) {
			hmi_evt->severity = OpalHMI_SEV_NO_ERROR;
			hmi_evt->type = OpalHMI_ERROR_PROC_RECOV_DONE;
			queue_hmi_event(hmi_evt, recover, out_flags);
		}
		hmi_print_debug("Turbo versus nominal frequency exceeded limit.", hmer);
	}

	/* Assert if we see malfunction alert, we can not continue. */
	if (hmer & SPR_HMER_MALFUNCTION_ALERT) {
		handled |= SPR_HMER_MALFUNCTION_ALERT;

		hmi_print_debug("Malfunction Alert", hmer);
		recover = 0;
		if (hmi_evt)
			decode_malfunction(hmi_evt, out_flags);
	}

	/* Assert if we see Hypervisor resource error, we can not continue. */
	if ((proc_gen <= proc_gen_p9) && (hmer & SPR_HMER_HYP_RESOURCE_ERR)) {
		handled |= SPR_HMER_HYP_RESOURCE_ERR;

		hmi_print_debug("Hypervisor resource error", hmer);
		recover = 0;
		if (hmi_evt) {
			hmi_evt->severity = OpalHMI_SEV_FATAL;
			hmi_evt->type = OpalHMI_ERROR_HYP_RESOURCE;
			queue_hmi_event(hmi_evt, recover, out_flags);
		}
	}

	/* XXX: what to do with this? */
	if ((proc_gen <= proc_gen_p9) && (hmer & SPR_HMER_THD_WAKE_BLOCKED_TM_SUSPEND)) {
		handled |= SPR_HMER_THD_WAKE_BLOCKED_TM_SUSPEND;
		hmer &= ~SPR_HMER_THD_WAKE_BLOCKED_TM_SUSPEND;

		hmi_print_debug("Attempted to wake thread when threads in TM suspend mode.", hmer);
		if (hmi_evt) {
			hmi_evt->severity = OpalHMI_SEV_NO_ERROR;
			hmi_evt->type = OpalHMI_ERROR_PROC_RECOV_DONE,
				queue_hmi_event(hmi_evt, recover, out_flags);
		}
	}

	if ((proc_gen <= proc_gen_p9) && (hmer & SPR_HMER_TRIG_FIR_HMI)) {
		handled |= SPR_HMER_TRIG_FIR_HMI;
		hmer &= ~SPR_HMER_TRIG_FIR_HMI;

		hmi_print_debug("Clearing unknown debug trigger", hmer);
		if (hmi_evt) {
			hmi_evt->severity = OpalHMI_SEV_NO_ERROR;
			hmi_evt->type = OpalHMI_ERROR_DEBUG_TRIG_FIR,
				queue_hmi_event(hmi_evt, recover, out_flags);
		}
	}
	if ((proc_gen == proc_gen_p10) && (hmer & SPR_HMER_P10_TRIG_FIR_HMI)) {
		handled |= SPR_HMER_P10_TRIG_FIR_HMI;
		hmer &= ~SPR_HMER_P10_TRIG_FIR_HMI;

		hmi_print_debug("Clearing unknown debug trigger", hmer);
		if (hmi_evt) {
			hmi_evt->severity = OpalHMI_SEV_NO_ERROR;
			hmi_evt->type = OpalHMI_ERROR_DEBUG_TRIG_FIR,
				queue_hmi_event(hmi_evt, recover, out_flags);
		}
	}

	if (recover == 0)
		disable_fast_reboot("Unrecoverable HMI");
	/*
	 * HMER bits are sticky, once set to 1 they remain set to 1 until
	 * they are set to 0. Reset the error source bit to 0, otherwise
	 * we keep getting HMI interrupt again and again. Writing to HMER
	 * acts as an AND, so we write mask of all 1's except for the bits
	 * we want to clear.
	 */
	mtspr(SPR_HMER, ~handled);
	unlock(&hmi_lock);
	return recover;
}

static int64_t opal_handle_hmi(void)
{
	uint64_t hmer, dummy_flags;
	struct OpalHMIEvent hmi_evt;

	/*
	 * Compiled time check to see size of OpalHMIEvent do not exceed
	 * that of struct opal_msg.
	 */
	BUILD_ASSERT(sizeof(struct opal_msg) >= sizeof(struct OpalHMIEvent));

	memset(&hmi_evt, 0, sizeof(struct OpalHMIEvent));
	hmi_evt.version = OpalHMIEvt_V2;

	hmer = mfspr(SPR_HMER);		/* Get HMER register value */
	handle_hmi_exception(hmer, &hmi_evt, &dummy_flags);

	return OPAL_SUCCESS;
}
opal_call(OPAL_HANDLE_HMI, opal_handle_hmi, 0);

static int64_t opal_handle_hmi2(__be64 *out_flags)
{
	uint64_t hmer, flags = 0;
	struct OpalHMIEvent hmi_evt;

	/*
	 * Compiled time check to see size of OpalHMIEvent do not exceed
	 * that of struct opal_msg.
	 */
	BUILD_ASSERT(sizeof(struct opal_msg) >= sizeof(struct OpalHMIEvent));

	memset(&hmi_evt, 0, sizeof(struct OpalHMIEvent));
	hmi_evt.version = OpalHMIEvt_V2;

	hmer = mfspr(SPR_HMER);		/* Get HMER register value */
	handle_hmi_exception(hmer, &hmi_evt, &flags);
	*out_flags = cpu_to_be64(flags);

	return OPAL_SUCCESS;
}
opal_call(OPAL_HANDLE_HMI2, opal_handle_hmi2, 1);