summaryrefslogtreecommitdiff
path: root/arch/powerpc/sysdev/xive/common.c
blob: 3459015092fa62e3f5a4c61cb7d37a812e391cf8 (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
/*
 * Copyright 2016,2017 IBM Corporation.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version
 * 2 of the License, or (at your option) any later version.
 */

#define pr_fmt(fmt) "xive: " fmt

#include <linux/types.h>
#include <linux/threads.h>
#include <linux/kernel.h>
#include <linux/irq.h>
#include <linux/debugfs.h>
#include <linux/smp.h>
#include <linux/interrupt.h>
#include <linux/seq_file.h>
#include <linux/init.h>
#include <linux/cpu.h>
#include <linux/of.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/msi.h>

#include <asm/prom.h>
#include <asm/io.h>
#include <asm/smp.h>
#include <asm/machdep.h>
#include <asm/irq.h>
#include <asm/errno.h>
#include <asm/xive.h>
#include <asm/xive-regs.h>
#include <asm/xmon.h>

#include "xive-internal.h"

#undef DEBUG_FLUSH
#undef DEBUG_ALL

#ifdef DEBUG_ALL
#define DBG_VERBOSE(fmt, ...)	pr_devel("cpu %d - " fmt, \
					 smp_processor_id(), ## __VA_ARGS__)
#else
#define DBG_VERBOSE(fmt...)	do { } while(0)
#endif

bool __xive_enabled;
EXPORT_SYMBOL_GPL(__xive_enabled);
bool xive_cmdline_disabled;

/* We use only one priority for now */
static u8 xive_irq_priority;

/* TIMA exported to KVM */
void __iomem *xive_tima;
EXPORT_SYMBOL_GPL(xive_tima);
u32 xive_tima_offset;

/* Backend ops */
static const struct xive_ops *xive_ops;

/* Our global interrupt domain */
static struct irq_domain *xive_irq_domain;

#ifdef CONFIG_SMP
/* The IPIs all use the same logical irq number */
static u32 xive_ipi_irq;
#endif

/* Xive state for each CPU */
static DEFINE_PER_CPU(struct xive_cpu *, xive_cpu);

/*
 * A "disabled" interrupt should never fire, to catch problems
 * we set its logical number to this
 */
#define XIVE_BAD_IRQ		0x7fffffff
#define XIVE_MAX_IRQ		(XIVE_BAD_IRQ - 1)

/* An invalid CPU target */
#define XIVE_INVALID_TARGET	(-1)

/*
 * Read the next entry in a queue, return its content if it's valid
 * or 0 if there is no new entry.
 *
 * The queue pointer is moved forward unless "just_peek" is set
 */
static u32 xive_read_eq(struct xive_q *q, bool just_peek)
{
	u32 cur;

	if (!q->qpage)
		return 0;
	cur = be32_to_cpup(q->qpage + q->idx);

	/* Check valid bit (31) vs current toggle polarity */
	if ((cur >> 31) == q->toggle)
		return 0;

	/* If consuming from the queue ... */
	if (!just_peek) {
		/* Next entry */
		q->idx = (q->idx + 1) & q->msk;

		/* Wrap around: flip valid toggle */
		if (q->idx == 0)
			q->toggle ^= 1;
	}
	/* Mask out the valid bit (31) */
	return cur & 0x7fffffff;
}

/*
 * Scans all the queue that may have interrupts in them
 * (based on "pending_prio") in priority order until an
 * interrupt is found or all the queues are empty.
 *
 * Then updates the CPPR (Current Processor Priority
 * Register) based on the most favored interrupt found
 * (0xff if none) and return what was found (0 if none).
 *
 * If just_peek is set, return the most favored pending
 * interrupt if any but don't update the queue pointers.
 *
 * Note: This function can operate generically on any number
 * of queues (up to 8). The current implementation of the XIVE
 * driver only uses a single queue however.
 *
 * Note2: This will also "flush" "the pending_count" of a queue
 * into the "count" when that queue is observed to be empty.
 * This is used to keep track of the amount of interrupts
 * targetting a queue. When an interrupt is moved away from
 * a queue, we only decrement that queue count once the queue
 * has been observed empty to avoid races.
 */
static u32 xive_scan_interrupts(struct xive_cpu *xc, bool just_peek)
{
	u32 irq = 0;
	u8 prio;

	/* Find highest pending priority */
	while (xc->pending_prio != 0) {
		struct xive_q *q;

		prio = ffs(xc->pending_prio) - 1;
		DBG_VERBOSE("scan_irq: trying prio %d\n", prio);

		/* Try to fetch */
		irq = xive_read_eq(&xc->queue[prio], just_peek);

		/* Found something ? That's it */
		if (irq)
			break;

		/* Clear pending bits */
		xc->pending_prio &= ~(1 << prio);

		/*
		 * Check if the queue count needs adjusting due to
		 * interrupts being moved away. See description of
		 * xive_dec_target_count()
		 */
		q = &xc->queue[prio];
		if (atomic_read(&q->pending_count)) {
			int p = atomic_xchg(&q->pending_count, 0);
			if (p) {
				WARN_ON(p > atomic_read(&q->count));
				atomic_sub(p, &q->count);
			}
		}
	}

	/* If nothing was found, set CPPR to 0xff */
	if (irq == 0)
		prio = 0xff;

	/* Update HW CPPR to match if necessary */
	if (prio != xc->cppr) {
		DBG_VERBOSE("scan_irq: adjusting CPPR to %d\n", prio);
		xc->cppr = prio;
		out_8(xive_tima + xive_tima_offset + TM_CPPR, prio);
	}

	return irq;
}

/*
 * This is used to perform the magic loads from an ESB
 * described in xive.h
 */
static notrace u8 xive_esb_read(struct xive_irq_data *xd, u32 offset)
{
	u64 val;

	/* Handle HW errata */
	if (xd->flags & XIVE_IRQ_FLAG_SHIFT_BUG)
		offset |= offset << 4;

	if ((xd->flags & XIVE_IRQ_FLAG_H_INT_ESB) && xive_ops->esb_rw)
		val = xive_ops->esb_rw(xd->hw_irq, offset, 0, 0);
	else
		val = in_be64(xd->eoi_mmio + offset);

	return (u8)val;
}

static void xive_esb_write(struct xive_irq_data *xd, u32 offset, u64 data)
{
	/* Handle HW errata */
	if (xd->flags & XIVE_IRQ_FLAG_SHIFT_BUG)
		offset |= offset << 4;

	if ((xd->flags & XIVE_IRQ_FLAG_H_INT_ESB) && xive_ops->esb_rw)
		xive_ops->esb_rw(xd->hw_irq, offset, data, 1);
	else
		out_be64(xd->eoi_mmio + offset, data);
}

#ifdef CONFIG_XMON
static notrace void xive_dump_eq(const char *name, struct xive_q *q)
{
	u32 i0, i1, idx;

	if (!q->qpage)
		return;
	idx = q->idx;
	i0 = be32_to_cpup(q->qpage + idx);
	idx = (idx + 1) & q->msk;
	i1 = be32_to_cpup(q->qpage + idx);
	xmon_printf("  %s Q T=%d %08x %08x ...\n", name,
		    q->toggle, i0, i1);
}

notrace void xmon_xive_do_dump(int cpu)
{
	struct xive_cpu *xc = per_cpu(xive_cpu, cpu);

	xmon_printf("XIVE state for CPU %d:\n", cpu);
	xmon_printf("  pp=%02x cppr=%02x\n", xc->pending_prio, xc->cppr);
	xive_dump_eq("IRQ", &xc->queue[xive_irq_priority]);
#ifdef CONFIG_SMP
	{
		u64 val = xive_esb_read(&xc->ipi_data, XIVE_ESB_GET);
		xmon_printf("  IPI state: %x:%c%c\n", xc->hw_ipi,
			val & XIVE_ESB_VAL_P ? 'P' : 'p',
			val & XIVE_ESB_VAL_Q ? 'Q' : 'q');
	}
#endif
}
#endif /* CONFIG_XMON */

static unsigned int xive_get_irq(void)
{
	struct xive_cpu *xc = __this_cpu_read(xive_cpu);
	u32 irq;

	/*
	 * This can be called either as a result of a HW interrupt or
	 * as a "replay" because EOI decided there was still something
	 * in one of the queues.
	 *
	 * First we perform an ACK cycle in order to update our mask
	 * of pending priorities. This will also have the effect of
	 * updating the CPPR to the most favored pending interrupts.
	 *
	 * In the future, if we have a way to differenciate a first
	 * entry (on HW interrupt) from a replay triggered by EOI,
	 * we could skip this on replays unless we soft-mask tells us
	 * that a new HW interrupt occurred.
	 */
	xive_ops->update_pending(xc);

	DBG_VERBOSE("get_irq: pending=%02x\n", xc->pending_prio);

	/* Scan our queue(s) for interrupts */
	irq = xive_scan_interrupts(xc, false);

	DBG_VERBOSE("get_irq: got irq 0x%x, new pending=0x%02x\n",
	    irq, xc->pending_prio);

	/* Return pending interrupt if any */
	if (irq == XIVE_BAD_IRQ)
		return 0;
	return irq;
}

/*
 * After EOI'ing an interrupt, we need to re-check the queue
 * to see if another interrupt is pending since multiple
 * interrupts can coalesce into a single notification to the
 * CPU.
 *
 * If we find that there is indeed more in there, we call
 * force_external_irq_replay() to make Linux synthetize an
 * external interrupt on the next call to local_irq_restore().
 */
static void xive_do_queue_eoi(struct xive_cpu *xc)
{
	if (xive_scan_interrupts(xc, true) != 0) {
		DBG_VERBOSE("eoi: pending=0x%02x\n", xc->pending_prio);
		force_external_irq_replay();
	}
}

/*
 * EOI an interrupt at the source. There are several methods
 * to do this depending on the HW version and source type
 */
void xive_do_source_eoi(u32 hw_irq, struct xive_irq_data *xd)
{
	/* If the XIVE supports the new "store EOI facility, use it */
	if (xd->flags & XIVE_IRQ_FLAG_STORE_EOI)
		xive_esb_write(xd, XIVE_ESB_STORE_EOI, 0);
	else if (hw_irq && xd->flags & XIVE_IRQ_FLAG_EOI_FW) {
		/*
		 * The FW told us to call it. This happens for some
		 * interrupt sources that need additional HW whacking
		 * beyond the ESB manipulation. For example LPC interrupts
		 * on P9 DD1.0 need a latch to be clared in the LPC bridge
		 * itself. The Firmware will take care of it.
		 */
		if (WARN_ON_ONCE(!xive_ops->eoi))
			return;
		xive_ops->eoi(hw_irq);
	} else {
		u8 eoi_val;

		/*
		 * Otherwise for EOI, we use the special MMIO that does
		 * a clear of both P and Q and returns the old Q,
		 * except for LSIs where we use the "EOI cycle" special
		 * load.
		 *
		 * This allows us to then do a re-trigger if Q was set
		 * rather than synthesizing an interrupt in software
		 *
		 * For LSIs, using the HW EOI cycle works around a problem
		 * on P9 DD1 PHBs where the other ESB accesses don't work
		 * properly.
		 */
		if (xd->flags & XIVE_IRQ_FLAG_LSI)
			xive_esb_read(xd, XIVE_ESB_LOAD_EOI);
		else {
			eoi_val = xive_esb_read(xd, XIVE_ESB_SET_PQ_00);
			DBG_VERBOSE("eoi_val=%x\n", eoi_val);

			/* Re-trigger if needed */
			if ((eoi_val & XIVE_ESB_VAL_Q) && xd->trig_mmio)
				out_be64(xd->trig_mmio, 0);
		}
	}
}

/* irq_chip eoi callback */
static void xive_irq_eoi(struct irq_data *d)
{
	struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);
	struct xive_cpu *xc = __this_cpu_read(xive_cpu);

	DBG_VERBOSE("eoi_irq: irq=%d [0x%lx] pending=%02x\n",
		    d->irq, irqd_to_hwirq(d), xc->pending_prio);

	/*
	 * EOI the source if it hasn't been disabled and hasn't
	 * been passed-through to a KVM guest
	 */
	if (!irqd_irq_disabled(d) && !irqd_is_forwarded_to_vcpu(d) &&
	    !(xd->flags & XIVE_IRQ_NO_EOI))
		xive_do_source_eoi(irqd_to_hwirq(d), xd);

	/*
	 * Clear saved_p to indicate that it's no longer occupying
	 * a queue slot on the target queue
	 */
	xd->saved_p = false;

	/* Check for more work in the queue */
	xive_do_queue_eoi(xc);
}

/*
 * Helper used to mask and unmask an interrupt source. This
 * is only called for normal interrupts that do not require
 * masking/unmasking via firmware.
 */
static void xive_do_source_set_mask(struct xive_irq_data *xd,
				    bool mask)
{
	u64 val;

	/*
	 * If the interrupt had P set, it may be in a queue.
	 *
	 * We need to make sure we don't re-enable it until it
	 * has been fetched from that queue and EOId. We keep
	 * a copy of that P state and use it to restore the
	 * ESB accordingly on unmask.
	 */
	if (mask) {
		val = xive_esb_read(xd, XIVE_ESB_SET_PQ_01);
		xd->saved_p = !!(val & XIVE_ESB_VAL_P);
	} else if (xd->saved_p)
		xive_esb_read(xd, XIVE_ESB_SET_PQ_10);
	else
		xive_esb_read(xd, XIVE_ESB_SET_PQ_00);
}

/*
 * Try to chose "cpu" as a new interrupt target. Increments
 * the queue accounting for that target if it's not already
 * full.
 */
static bool xive_try_pick_target(int cpu)
{
	struct xive_cpu *xc = per_cpu(xive_cpu, cpu);
	struct xive_q *q = &xc->queue[xive_irq_priority];
	int max;

	/*
	 * Calculate max number of interrupts in that queue.
	 *
	 * We leave a gap of 1 just in case...
	 */
	max = (q->msk + 1) - 1;
	return !!atomic_add_unless(&q->count, 1, max);
}

/*
 * Un-account an interrupt for a target CPU. We don't directly
 * decrement q->count since the interrupt might still be present
 * in the queue.
 *
 * Instead increment a separate counter "pending_count" which
 * will be substracted from "count" later when that CPU observes
 * the queue to be empty.
 */
static void xive_dec_target_count(int cpu)
{
	struct xive_cpu *xc = per_cpu(xive_cpu, cpu);
	struct xive_q *q = &xc->queue[xive_irq_priority];

	if (unlikely(WARN_ON(cpu < 0 || !xc))) {
		pr_err("%s: cpu=%d xc=%p\n", __func__, cpu, xc);
		return;
	}

	/*
	 * We increment the "pending count" which will be used
	 * to decrement the target queue count whenever it's next
	 * processed and found empty. This ensure that we don't
	 * decrement while we still have the interrupt there
	 * occupying a slot.
	 */
	atomic_inc(&q->pending_count);
}

/* Find a tentative CPU target in a CPU mask */
static int xive_find_target_in_mask(const struct cpumask *mask,
				    unsigned int fuzz)
{
	int cpu, first, num, i;

	/* Pick up a starting point CPU in the mask based on  fuzz */
	num = min_t(int, cpumask_weight(mask), nr_cpu_ids);
	first = fuzz % num;

	/* Locate it */
	cpu = cpumask_first(mask);
	for (i = 0; i < first && cpu < nr_cpu_ids; i++)
		cpu = cpumask_next(cpu, mask);

	/* Sanity check */
	if (WARN_ON(cpu >= nr_cpu_ids))
		cpu = cpumask_first(cpu_online_mask);

	/* Remember first one to handle wrap-around */
	first = cpu;

	/*
	 * Now go through the entire mask until we find a valid
	 * target.
	 */
	for (;;) {
		/*
		 * We re-check online as the fallback case passes us
		 * an untested affinity mask
		 */
		if (cpu_online(cpu) && xive_try_pick_target(cpu))
			return cpu;
		cpu = cpumask_next(cpu, mask);
		if (cpu == first)
			break;
		/* Wrap around */
		if (cpu >= nr_cpu_ids)
			cpu = cpumask_first(mask);
	}
	return -1;
}

/*
 * Pick a target CPU for an interrupt. This is done at
 * startup or if the affinity is changed in a way that
 * invalidates the current target.
 */
static int xive_pick_irq_target(struct irq_data *d,
				const struct cpumask *affinity)
{
	static unsigned int fuzz;
	struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);
	cpumask_var_t mask;
	int cpu = -1;

	/*
	 * If we have chip IDs, first we try to build a mask of
	 * CPUs matching the CPU and find a target in there
	 */
	if (xd->src_chip != XIVE_INVALID_CHIP_ID &&
		zalloc_cpumask_var(&mask, GFP_ATOMIC)) {
		/* Build a mask of matching chip IDs */
		for_each_cpu_and(cpu, affinity, cpu_online_mask) {
			struct xive_cpu *xc = per_cpu(xive_cpu, cpu);
			if (xc->chip_id == xd->src_chip)
				cpumask_set_cpu(cpu, mask);
		}
		/* Try to find a target */
		if (cpumask_empty(mask))
			cpu = -1;
		else
			cpu = xive_find_target_in_mask(mask, fuzz++);
		free_cpumask_var(mask);
		if (cpu >= 0)
			return cpu;
		fuzz--;
	}

	/* No chip IDs, fallback to using the affinity mask */
	return xive_find_target_in_mask(affinity, fuzz++);
}

static unsigned int xive_irq_startup(struct irq_data *d)
{
	struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);
	unsigned int hw_irq = (unsigned int)irqd_to_hwirq(d);
	int target, rc;

	pr_devel("xive_irq_startup: irq %d [0x%x] data @%p\n",
		 d->irq, hw_irq, d);

#ifdef CONFIG_PCI_MSI
	/*
	 * The generic MSI code returns with the interrupt disabled on the
	 * card, using the MSI mask bits. Firmware doesn't appear to unmask
	 * at that level, so we do it here by hand.
	 */
	if (irq_data_get_msi_desc(d))
		pci_msi_unmask_irq(d);
#endif

	/* Pick a target */
	target = xive_pick_irq_target(d, irq_data_get_affinity_mask(d));
	if (target == XIVE_INVALID_TARGET) {
		/* Try again breaking affinity */
		target = xive_pick_irq_target(d, cpu_online_mask);
		if (target == XIVE_INVALID_TARGET)
			return -ENXIO;
		pr_warn("irq %d started with broken affinity\n", d->irq);
	}

	/* Sanity check */
	if (WARN_ON(target == XIVE_INVALID_TARGET ||
		    target >= nr_cpu_ids))
		target = smp_processor_id();

	xd->target = target;

	/*
	 * Configure the logical number to be the Linux IRQ number
	 * and set the target queue
	 */
	rc = xive_ops->configure_irq(hw_irq,
				     get_hard_smp_processor_id(target),
				     xive_irq_priority, d->irq);
	if (rc)
		return rc;

	/* Unmask the ESB */
	xive_do_source_set_mask(xd, false);

	return 0;
}

static void xive_irq_shutdown(struct irq_data *d)
{
	struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);
	unsigned int hw_irq = (unsigned int)irqd_to_hwirq(d);

	pr_devel("xive_irq_shutdown: irq %d [0x%x] data @%p\n",
		 d->irq, hw_irq, d);

	if (WARN_ON(xd->target == XIVE_INVALID_TARGET))
		return;

	/* Mask the interrupt at the source */
	xive_do_source_set_mask(xd, true);

	/*
	 * The above may have set saved_p. We clear it otherwise it
	 * will prevent re-enabling later on. It is ok to forget the
	 * fact that the interrupt might be in a queue because we are
	 * accounting that already in xive_dec_target_count() and will
	 * be re-routing it to a new queue with proper accounting when
	 * it's started up again
	 */
	xd->saved_p = false;

	/*
	 * Mask the interrupt in HW in the IVT/EAS and set the number
	 * to be the "bad" IRQ number
	 */
	xive_ops->configure_irq(hw_irq,
				get_hard_smp_processor_id(xd->target),
				0xff, XIVE_BAD_IRQ);

	xive_dec_target_count(xd->target);
	xd->target = XIVE_INVALID_TARGET;
}

static void xive_irq_unmask(struct irq_data *d)
{
	struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);

	pr_devel("xive_irq_unmask: irq %d data @%p\n", d->irq, xd);

	/*
	 * This is a workaround for PCI LSI problems on P9, for
	 * these, we call FW to set the mask. The problems might
	 * be fixed by P9 DD2.0, if that is the case, firmware
	 * will no longer set that flag.
	 */
	if (xd->flags & XIVE_IRQ_FLAG_MASK_FW) {
		unsigned int hw_irq = (unsigned int)irqd_to_hwirq(d);
		xive_ops->configure_irq(hw_irq,
					get_hard_smp_processor_id(xd->target),
					xive_irq_priority, d->irq);
		return;
	}

	xive_do_source_set_mask(xd, false);
}

static void xive_irq_mask(struct irq_data *d)
{
	struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);

	pr_devel("xive_irq_mask: irq %d data @%p\n", d->irq, xd);

	/*
	 * This is a workaround for PCI LSI problems on P9, for
	 * these, we call OPAL to set the mask. The problems might
	 * be fixed by P9 DD2.0, if that is the case, firmware
	 * will no longer set that flag.
	 */
	if (xd->flags & XIVE_IRQ_FLAG_MASK_FW) {
		unsigned int hw_irq = (unsigned int)irqd_to_hwirq(d);
		xive_ops->configure_irq(hw_irq,
					get_hard_smp_processor_id(xd->target),
					0xff, d->irq);
		return;
	}

	xive_do_source_set_mask(xd, true);
}

static int xive_irq_set_affinity(struct irq_data *d,
				 const struct cpumask *cpumask,
				 bool force)
{
	struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);
	unsigned int hw_irq = (unsigned int)irqd_to_hwirq(d);
	u32 target, old_target;
	int rc = 0;

	pr_devel("xive_irq_set_affinity: irq %d\n", d->irq);

	/* Is this valid ? */
	if (cpumask_any_and(cpumask, cpu_online_mask) >= nr_cpu_ids)
		return -EINVAL;

	/* Don't do anything if the interrupt isn't started */
	if (!irqd_is_started(d))
		return IRQ_SET_MASK_OK;

	/*
	 * If existing target is already in the new mask, and is
	 * online then do nothing.
	 */
	if (xd->target != XIVE_INVALID_TARGET &&
	    cpu_online(xd->target) &&
	    cpumask_test_cpu(xd->target, cpumask))
		return IRQ_SET_MASK_OK;

	/* Pick a new target */
	target = xive_pick_irq_target(d, cpumask);

	/* No target found */
	if (target == XIVE_INVALID_TARGET)
		return -ENXIO;

	/* Sanity check */
	if (WARN_ON(target >= nr_cpu_ids))
		target = smp_processor_id();

	old_target = xd->target;

	/*
	 * Only configure the irq if it's not currently passed-through to
	 * a KVM guest
	 */
	if (!irqd_is_forwarded_to_vcpu(d))
		rc = xive_ops->configure_irq(hw_irq,
					     get_hard_smp_processor_id(target),
					     xive_irq_priority, d->irq);
	if (rc < 0) {
		pr_err("Error %d reconfiguring irq %d\n", rc, d->irq);
		return rc;
	}

	pr_devel("  target: 0x%x\n", target);
	xd->target = target;

	/* Give up previous target */
	if (old_target != XIVE_INVALID_TARGET)
	    xive_dec_target_count(old_target);

	return IRQ_SET_MASK_OK;
}

static int xive_irq_set_type(struct irq_data *d, unsigned int flow_type)
{
	struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);

	/*
	 * We only support these. This has really no effect other than setting
	 * the corresponding descriptor bits mind you but those will in turn
	 * affect the resend function when re-enabling an edge interrupt.
	 *
	 * Set set the default to edge as explained in map().
	 */
	if (flow_type == IRQ_TYPE_DEFAULT || flow_type == IRQ_TYPE_NONE)
		flow_type = IRQ_TYPE_EDGE_RISING;

	if (flow_type != IRQ_TYPE_EDGE_RISING &&
	    flow_type != IRQ_TYPE_LEVEL_LOW)
		return -EINVAL;

	irqd_set_trigger_type(d, flow_type);

	/*
	 * Double check it matches what the FW thinks
	 *
	 * NOTE: We don't know yet if the PAPR interface will provide
	 * the LSI vs MSI information apart from the device-tree so
	 * this check might have to move into an optional backend call
	 * that is specific to the native backend
	 */
	if ((flow_type == IRQ_TYPE_LEVEL_LOW) !=
	    !!(xd->flags & XIVE_IRQ_FLAG_LSI)) {
		pr_warn("Interrupt %d (HW 0x%x) type mismatch, Linux says %s, FW says %s\n",
			d->irq, (u32)irqd_to_hwirq(d),
			(flow_type == IRQ_TYPE_LEVEL_LOW) ? "Level" : "Edge",
			(xd->flags & XIVE_IRQ_FLAG_LSI) ? "Level" : "Edge");
	}

	return IRQ_SET_MASK_OK_NOCOPY;
}

static int xive_irq_retrigger(struct irq_data *d)
{
	struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);

	/* This should be only for MSIs */
	if (WARN_ON(xd->flags & XIVE_IRQ_FLAG_LSI))
		return 0;

	/*
	 * To perform a retrigger, we first set the PQ bits to
	 * 11, then perform an EOI.
	 */
	xive_esb_read(xd, XIVE_ESB_SET_PQ_11);

	/*
	 * Note: We pass "0" to the hw_irq argument in order to
	 * avoid calling into the backend EOI code which we don't
	 * want to do in the case of a re-trigger. Backends typically
	 * only do EOI for LSIs anyway.
	 */
	xive_do_source_eoi(0, xd);

	return 1;
}

static int xive_irq_set_vcpu_affinity(struct irq_data *d, void *state)
{
	struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);
	unsigned int hw_irq = (unsigned int)irqd_to_hwirq(d);
	int rc;
	u8 pq;

	/*
	 * We only support this on interrupts that do not require
	 * firmware calls for masking and unmasking
	 */
	if (xd->flags & XIVE_IRQ_FLAG_MASK_FW)
		return -EIO;

	/*
	 * This is called by KVM with state non-NULL for enabling
	 * pass-through or NULL for disabling it
	 */
	if (state) {
		irqd_set_forwarded_to_vcpu(d);

		/* Set it to PQ=10 state to prevent further sends */
		pq = xive_esb_read(xd, XIVE_ESB_SET_PQ_10);

		/* No target ? nothing to do */
		if (xd->target == XIVE_INVALID_TARGET) {
			/*
			 * An untargetted interrupt should have been
			 * also masked at the source
			 */
			WARN_ON(pq & 2);

			return 0;
		}

		/*
		 * If P was set, adjust state to PQ=11 to indicate
		 * that a resend is needed for the interrupt to reach
		 * the guest. Also remember the value of P.
		 *
		 * This also tells us that it's in flight to a host queue
		 * or has already been fetched but hasn't been EOIed yet
		 * by the host. This it's potentially using up a host
		 * queue slot. This is important to know because as long
		 * as this is the case, we must not hard-unmask it when
		 * "returning" that interrupt to the host.
		 *
		 * This saved_p is cleared by the host EOI, when we know
		 * for sure the queue slot is no longer in use.
		 */
		if (pq & 2) {
			pq = xive_esb_read(xd, XIVE_ESB_SET_PQ_11);
			xd->saved_p = true;

			/*
			 * Sync the XIVE source HW to ensure the interrupt
			 * has gone through the EAS before we change its
			 * target to the guest. That should guarantee us
			 * that we *will* eventually get an EOI for it on
			 * the host. Otherwise there would be a small window
			 * for P to be seen here but the interrupt going
			 * to the guest queue.
			 */
			if (xive_ops->sync_source)
				xive_ops->sync_source(hw_irq);
		} else
			xd->saved_p = false;
	} else {
		irqd_clr_forwarded_to_vcpu(d);

		/* No host target ? hard mask and return */
		if (xd->target == XIVE_INVALID_TARGET) {
			xive_do_source_set_mask(xd, true);
			return 0;
		}

		/*
		 * Sync the XIVE source HW to ensure the interrupt
		 * has gone through the EAS before we change its
		 * target to the host.
		 */
		if (xive_ops->sync_source)
			xive_ops->sync_source(hw_irq);

		/*
		 * By convention we are called with the interrupt in
		 * a PQ=10 or PQ=11 state, ie, it won't fire and will
		 * have latched in Q whether there's a pending HW
		 * interrupt or not.
		 *
		 * First reconfigure the target.
		 */
		rc = xive_ops->configure_irq(hw_irq,
					     get_hard_smp_processor_id(xd->target),
					     xive_irq_priority, d->irq);
		if (rc)
			return rc;

		/*
		 * Then if saved_p is not set, effectively re-enable the
		 * interrupt with an EOI. If it is set, we know there is
		 * still a message in a host queue somewhere that will be
		 * EOId eventually.
		 *
		 * Note: We don't check irqd_irq_disabled(). Effectively,
		 * we *will* let the irq get through even if masked if the
		 * HW is still firing it in order to deal with the whole
		 * saved_p business properly. If the interrupt triggers
		 * while masked, the generic code will re-mask it anyway.
		 */
		if (!xd->saved_p)
			xive_do_source_eoi(hw_irq, xd);

	}
	return 0;
}

static struct irq_chip xive_irq_chip = {
	.name = "XIVE-IRQ",
	.irq_startup = xive_irq_startup,
	.irq_shutdown = xive_irq_shutdown,
	.irq_eoi = xive_irq_eoi,
	.irq_mask = xive_irq_mask,
	.irq_unmask = xive_irq_unmask,
	.irq_set_affinity = xive_irq_set_affinity,
	.irq_set_type = xive_irq_set_type,
	.irq_retrigger = xive_irq_retrigger,
	.irq_set_vcpu_affinity = xive_irq_set_vcpu_affinity,
};

bool is_xive_irq(struct irq_chip *chip)
{
	return chip == &xive_irq_chip;
}
EXPORT_SYMBOL_GPL(is_xive_irq);

void xive_cleanup_irq_data(struct xive_irq_data *xd)
{
	if (xd->eoi_mmio) {
		iounmap(xd->eoi_mmio);
		if (xd->eoi_mmio == xd->trig_mmio)
			xd->trig_mmio = NULL;
		xd->eoi_mmio = NULL;
	}
	if (xd->trig_mmio) {
		iounmap(xd->trig_mmio);
		xd->trig_mmio = NULL;
	}
}
EXPORT_SYMBOL_GPL(xive_cleanup_irq_data);

static int xive_irq_alloc_data(unsigned int virq, irq_hw_number_t hw)
{
	struct xive_irq_data *xd;
	int rc;

	xd = kzalloc(sizeof(struct xive_irq_data), GFP_KERNEL);
	if (!xd)
		return -ENOMEM;
	rc = xive_ops->populate_irq_data(hw, xd);
	if (rc) {
		kfree(xd);
		return rc;
	}
	xd->target = XIVE_INVALID_TARGET;
	irq_set_handler_data(virq, xd);

	return 0;
}

static void xive_irq_free_data(unsigned int virq)
{
	struct xive_irq_data *xd = irq_get_handler_data(virq);

	if (!xd)
		return;
	irq_set_handler_data(virq, NULL);
	xive_cleanup_irq_data(xd);
	kfree(xd);
}

#ifdef CONFIG_SMP

static void xive_cause_ipi(int cpu)
{
	struct xive_cpu *xc;
	struct xive_irq_data *xd;

	xc = per_cpu(xive_cpu, cpu);

	DBG_VERBOSE("IPI CPU %d -> %d (HW IRQ 0x%x)\n",
		    smp_processor_id(), cpu, xc->hw_ipi);

	xd = &xc->ipi_data;
	if (WARN_ON(!xd->trig_mmio))
		return;
	out_be64(xd->trig_mmio, 0);
}

static irqreturn_t xive_muxed_ipi_action(int irq, void *dev_id)
{
	return smp_ipi_demux();
}

static void xive_ipi_eoi(struct irq_data *d)
{
	struct xive_cpu *xc = __this_cpu_read(xive_cpu);

	DBG_VERBOSE("IPI eoi: irq=%d [0x%lx] (HW IRQ 0x%x) pending=%02x\n",
		    d->irq, irqd_to_hwirq(d), xc->hw_ipi, xc->pending_prio);

	/* Handle possible race with unplug and drop stale IPIs */
	if (!xc)
		return;
	xive_do_source_eoi(xc->hw_ipi, &xc->ipi_data);
	xive_do_queue_eoi(xc);
}

static void xive_ipi_do_nothing(struct irq_data *d)
{
	/*
	 * Nothing to do, we never mask/unmask IPIs, but the callback
	 * has to exist for the struct irq_chip.
	 */
}

static struct irq_chip xive_ipi_chip = {
	.name = "XIVE-IPI",
	.irq_eoi = xive_ipi_eoi,
	.irq_mask = xive_ipi_do_nothing,
	.irq_unmask = xive_ipi_do_nothing,
};

static void __init xive_request_ipi(void)
{
	unsigned int virq;

	/*
	 * Initialization failed, move on, we might manage to
	 * reach the point where we display our errors before
	 * the system falls appart
	 */
	if (!xive_irq_domain)
		return;

	/* Initialize it */
	virq = irq_create_mapping(xive_irq_domain, 0);
	xive_ipi_irq = virq;

	WARN_ON(request_irq(virq, xive_muxed_ipi_action,
			    IRQF_PERCPU | IRQF_NO_THREAD, "IPI", NULL));
}

static int xive_setup_cpu_ipi(unsigned int cpu)
{
	struct xive_cpu *xc;
	int rc;

	pr_debug("Setting up IPI for CPU %d\n", cpu);

	xc = per_cpu(xive_cpu, cpu);

	/* Check if we are already setup */
	if (xc->hw_ipi != 0)
		return 0;

	/* Grab an IPI from the backend, this will populate xc->hw_ipi */
	if (xive_ops->get_ipi(cpu, xc))
		return -EIO;

	/*
	 * Populate the IRQ data in the xive_cpu structure and
	 * configure the HW / enable the IPIs.
	 */
	rc = xive_ops->populate_irq_data(xc->hw_ipi, &xc->ipi_data);
	if (rc) {
		pr_err("Failed to populate IPI data on CPU %d\n", cpu);
		return -EIO;
	}
	rc = xive_ops->configure_irq(xc->hw_ipi,
				     get_hard_smp_processor_id(cpu),
				     xive_irq_priority, xive_ipi_irq);
	if (rc) {
		pr_err("Failed to map IPI CPU %d\n", cpu);
		return -EIO;
	}
	pr_devel("CPU %d HW IPI %x, virq %d, trig_mmio=%p\n", cpu,
	    xc->hw_ipi, xive_ipi_irq, xc->ipi_data.trig_mmio);

	/* Unmask it */
	xive_do_source_set_mask(&xc->ipi_data, false);

	return 0;
}

static void xive_cleanup_cpu_ipi(unsigned int cpu, struct xive_cpu *xc)
{
	/* Disable the IPI and free the IRQ data */

	/* Already cleaned up ? */
	if (xc->hw_ipi == 0)
		return;

	/* Mask the IPI */
	xive_do_source_set_mask(&xc->ipi_data, true);

	/*
	 * Note: We don't call xive_cleanup_irq_data() to free
	 * the mappings as this is called from an IPI on kexec
	 * which is not a safe environment to call iounmap()
	 */

	/* Deconfigure/mask in the backend */
	xive_ops->configure_irq(xc->hw_ipi, hard_smp_processor_id(),
				0xff, xive_ipi_irq);

	/* Free the IPIs in the backend */
	xive_ops->put_ipi(cpu, xc);
}

void __init xive_smp_probe(void)
{
	smp_ops->cause_ipi = xive_cause_ipi;

	/* Register the IPI */
	xive_request_ipi();

	/* Allocate and setup IPI for the boot CPU */
	xive_setup_cpu_ipi(smp_processor_id());
}

#endif /* CONFIG_SMP */

static int xive_irq_domain_map(struct irq_domain *h, unsigned int virq,
			       irq_hw_number_t hw)
{
	int rc;

	/*
	 * Mark interrupts as edge sensitive by default so that resend
	 * actually works. Will fix that up below if needed.
	 */
	irq_clear_status_flags(virq, IRQ_LEVEL);

#ifdef CONFIG_SMP
	/* IPIs are special and come up with HW number 0 */
	if (hw == 0) {
		/*
		 * IPIs are marked per-cpu. We use separate HW interrupts under
		 * the hood but associated with the same "linux" interrupt
		 */
		irq_set_chip_and_handler(virq, &xive_ipi_chip,
					 handle_percpu_irq);
		return 0;
	}
#endif

	rc = xive_irq_alloc_data(virq, hw);
	if (rc)
		return rc;

	irq_set_chip_and_handler(virq, &xive_irq_chip, handle_fasteoi_irq);

	return 0;
}

static void xive_irq_domain_unmap(struct irq_domain *d, unsigned int virq)
{
	struct irq_data *data = irq_get_irq_data(virq);
	unsigned int hw_irq;

	/* XXX Assign BAD number */
	if (!data)
		return;
	hw_irq = (unsigned int)irqd_to_hwirq(data);
	if (hw_irq)
		xive_irq_free_data(virq);
}

static int xive_irq_domain_xlate(struct irq_domain *h, struct device_node *ct,
				 const u32 *intspec, unsigned int intsize,
				 irq_hw_number_t *out_hwirq, unsigned int *out_flags)

{
	*out_hwirq = intspec[0];

	/*
	 * If intsize is at least 2, we look for the type in the second cell,
	 * we assume the LSB indicates a level interrupt.
	 */
	if (intsize > 1) {
		if (intspec[1] & 1)
			*out_flags = IRQ_TYPE_LEVEL_LOW;
		else
			*out_flags = IRQ_TYPE_EDGE_RISING;
	} else
		*out_flags = IRQ_TYPE_LEVEL_LOW;

	return 0;
}

static int xive_irq_domain_match(struct irq_domain *h, struct device_node *node,
				 enum irq_domain_bus_token bus_token)
{
	return xive_ops->match(node);
}

static const struct irq_domain_ops xive_irq_domain_ops = {
	.match = xive_irq_domain_match,
	.map = xive_irq_domain_map,
	.unmap = xive_irq_domain_unmap,
	.xlate = xive_irq_domain_xlate,
};

static void __init xive_init_host(void)
{
	xive_irq_domain = irq_domain_add_nomap(NULL, XIVE_MAX_IRQ,
					       &xive_irq_domain_ops, NULL);
	if (WARN_ON(xive_irq_domain == NULL))
		return;
	irq_set_default_host(xive_irq_domain);
}

static void xive_cleanup_cpu_queues(unsigned int cpu, struct xive_cpu *xc)
{
	if (xc->queue[xive_irq_priority].qpage)
		xive_ops->cleanup_queue(cpu, xc, xive_irq_priority);
}

static int xive_setup_cpu_queues(unsigned int cpu, struct xive_cpu *xc)
{
	int rc = 0;

	/* We setup 1 queues for now with a 64k page */
	if (!xc->queue[xive_irq_priority].qpage)
		rc = xive_ops->setup_queue(cpu, xc, xive_irq_priority);

	return rc;
}

static int xive_prepare_cpu(unsigned int cpu)
{
	struct xive_cpu *xc;

	xc = per_cpu(xive_cpu, cpu);
	if (!xc) {
		struct device_node *np;

		xc = kzalloc_node(sizeof(struct xive_cpu),
				  GFP_KERNEL, cpu_to_node(cpu));
		if (!xc)
			return -ENOMEM;
		np = of_get_cpu_node(cpu, NULL);
		if (np)
			xc->chip_id = of_get_ibm_chip_id(np);
		of_node_put(np);

		per_cpu(xive_cpu, cpu) = xc;
	}

	/* Setup EQs if not already */
	return xive_setup_cpu_queues(cpu, xc);
}

static void xive_setup_cpu(void)
{
	struct xive_cpu *xc = __this_cpu_read(xive_cpu);

	/* The backend might have additional things to do */
	if (xive_ops->setup_cpu)
		xive_ops->setup_cpu(smp_processor_id(), xc);

	/* Set CPPR to 0xff to enable flow of interrupts */
	xc->cppr = 0xff;
	out_8(xive_tima + xive_tima_offset + TM_CPPR, 0xff);
}

#ifdef CONFIG_SMP
void xive_smp_setup_cpu(void)
{
	pr_devel("SMP setup CPU %d\n", smp_processor_id());

	/* This will have already been done on the boot CPU */
	if (smp_processor_id() != boot_cpuid)
		xive_setup_cpu();

}

int xive_smp_prepare_cpu(unsigned int cpu)
{
	int rc;

	/* Allocate per-CPU data and queues */
	rc = xive_prepare_cpu(cpu);
	if (rc)
		return rc;

	/* Allocate and setup IPI for the new CPU */
	return xive_setup_cpu_ipi(cpu);
}

#ifdef CONFIG_HOTPLUG_CPU
static void xive_flush_cpu_queue(unsigned int cpu, struct xive_cpu *xc)
{
	u32 irq;

	/* We assume local irqs are disabled */
	WARN_ON(!irqs_disabled());

	/* Check what's already in the CPU queue */
	while ((irq = xive_scan_interrupts(xc, false)) != 0) {
		/*
		 * We need to re-route that interrupt to its new destination.
		 * First get and lock the descriptor
		 */
		struct irq_desc *desc = irq_to_desc(irq);
		struct irq_data *d = irq_desc_get_irq_data(desc);
		struct xive_irq_data *xd;
		unsigned int hw_irq = (unsigned int)irqd_to_hwirq(d);

		/*
		 * Ignore anything that isn't a XIVE irq and ignore
		 * IPIs, so can just be dropped.
		 */
		if (d->domain != xive_irq_domain || hw_irq == 0)
			continue;

		/*
		 * The IRQ should have already been re-routed, it's just a
		 * stale in the old queue, so re-trigger it in order to make
		 * it reach is new destination.
		 */
#ifdef DEBUG_FLUSH
		pr_info("CPU %d: Got irq %d while offline, re-sending...\n",
			cpu, irq);
#endif
		raw_spin_lock(&desc->lock);
		xd = irq_desc_get_handler_data(desc);

		/*
		 * For LSIs, we EOI, this will cause a resend if it's
		 * still asserted. Otherwise do an MSI retrigger.
		 */
		if (xd->flags & XIVE_IRQ_FLAG_LSI)
			xive_do_source_eoi(irqd_to_hwirq(d), xd);
		else
			xive_irq_retrigger(d);

		raw_spin_unlock(&desc->lock);
	}
}

void xive_smp_disable_cpu(void)
{
	struct xive_cpu *xc = __this_cpu_read(xive_cpu);
	unsigned int cpu = smp_processor_id();

	/* Migrate interrupts away from the CPU */
	irq_migrate_all_off_this_cpu();

	/* Set CPPR to 0 to disable flow of interrupts */
	xc->cppr = 0;
	out_8(xive_tima + xive_tima_offset + TM_CPPR, 0);

	/* Flush everything still in the queue */
	xive_flush_cpu_queue(cpu, xc);

	/* Re-enable CPPR  */
	xc->cppr = 0xff;
	out_8(xive_tima + xive_tima_offset + TM_CPPR, 0xff);
}

void xive_flush_interrupt(void)
{
	struct xive_cpu *xc = __this_cpu_read(xive_cpu);
	unsigned int cpu = smp_processor_id();

	/* Called if an interrupt occurs while the CPU is hot unplugged */
	xive_flush_cpu_queue(cpu, xc);
}

#endif /* CONFIG_HOTPLUG_CPU */

#endif /* CONFIG_SMP */

void xive_teardown_cpu(void)
{
	struct xive_cpu *xc = __this_cpu_read(xive_cpu);
	unsigned int cpu = smp_processor_id();

	/* Set CPPR to 0 to disable flow of interrupts */
	xc->cppr = 0;
	out_8(xive_tima + xive_tima_offset + TM_CPPR, 0);

	if (xive_ops->teardown_cpu)
		xive_ops->teardown_cpu(cpu, xc);

#ifdef CONFIG_SMP
	/* Get rid of IPI */
	xive_cleanup_cpu_ipi(cpu, xc);
#endif

	/* Disable and free the queues */
	xive_cleanup_cpu_queues(cpu, xc);
}

void xive_kexec_teardown_cpu(int secondary)
{
	struct xive_cpu *xc = __this_cpu_read(xive_cpu);
	unsigned int cpu = smp_processor_id();

	/* Set CPPR to 0 to disable flow of interrupts */
	xc->cppr = 0;
	out_8(xive_tima + xive_tima_offset + TM_CPPR, 0);

	/* Backend cleanup if any */
	if (xive_ops->teardown_cpu)
		xive_ops->teardown_cpu(cpu, xc);

#ifdef CONFIG_SMP
	/* Get rid of IPI */
	xive_cleanup_cpu_ipi(cpu, xc);
#endif

	/* Disable and free the queues */
	xive_cleanup_cpu_queues(cpu, xc);
}

void xive_shutdown(void)
{
	xive_ops->shutdown();
}

bool __init xive_core_init(const struct xive_ops *ops, void __iomem *area, u32 offset,
			   u8 max_prio)
{
	xive_tima = area;
	xive_tima_offset = offset;
	xive_ops = ops;
	xive_irq_priority = max_prio;

	ppc_md.get_irq = xive_get_irq;
	__xive_enabled = true;

	pr_devel("Initializing host..\n");
	xive_init_host();

	pr_devel("Initializing boot CPU..\n");

	/* Allocate per-CPU data and queues */
	xive_prepare_cpu(smp_processor_id());

	/* Get ready for interrupts */
	xive_setup_cpu();

	pr_info("Interrupt handling initialized with %s backend\n",
		xive_ops->name);
	pr_info("Using priority %d for all interrupts\n", max_prio);

	return true;
}

__be32 *xive_queue_page_alloc(unsigned int cpu, u32 queue_shift)
{
	unsigned int alloc_order;
	struct page *pages;
	__be32 *qpage;

	alloc_order = xive_alloc_order(queue_shift);
	pages = alloc_pages_node(cpu_to_node(cpu), GFP_KERNEL, alloc_order);
	if (!pages)
		return ERR_PTR(-ENOMEM);
	qpage = (__be32 *)page_address(pages);
	memset(qpage, 0, 1 << queue_shift);

	return qpage;
}

static int __init xive_off(char *arg)
{
	xive_cmdline_disabled = true;
	return 0;
}
__setup("xive=off", xive_off);