summaryrefslogtreecommitdiffstats
path: root/arch/powerpc/perf/hv-24x7.c
blob: db213eb7cb02866a9779872960fa2edbe2a45b24 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
// SPDX-License-Identifier: GPL-2.0-or-later
/*
 * Hypervisor supplied "24x7" performance counter support
 *
 * Author: Cody P Schafer <cody@linux.vnet.ibm.com>
 * Copyright 2014 IBM Corporation.
 */

#define pr_fmt(fmt) "hv-24x7: " fmt

#include <linux/perf_event.h>
#include <linux/rbtree.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>

#include <asm/cputhreads.h>
#include <asm/firmware.h>
#include <asm/hvcall.h>
#include <asm/io.h>
#include <linux/byteorder/generic.h>

#include <asm/rtas.h>
#include "hv-24x7.h"
#include "hv-24x7-catalog.h"
#include "hv-common.h"

/* Version of the 24x7 hypervisor API that we should use in this machine. */
static int interface_version;

/* Whether we have to aggregate result data for some domains. */
static bool aggregate_result_elements;

static bool domain_is_valid(unsigned domain)
{
	switch (domain) {
#define DOMAIN(n, v, x, c)		\
	case HV_PERF_DOMAIN_##n:	\
		/* fall through */
#include "hv-24x7-domains.h"
#undef DOMAIN
		return true;
	default:
		return false;
	}
}

static bool is_physical_domain(unsigned domain)
{
	switch (domain) {
#define DOMAIN(n, v, x, c)		\
	case HV_PERF_DOMAIN_##n:	\
		return c;
#include "hv-24x7-domains.h"
#undef DOMAIN
	default:
		return false;
	}
}

/*
 * The Processor Module Information system parameter allows transferring
 * of certain processor module information from the platform to the OS.
 * Refer PAPR+ document to get parameter token value as '43'.
 */

#define PROCESSOR_MODULE_INFO   43

static u32 phys_sockets;	/* Physical sockets */
static u32 phys_chipspersocket;	/* Physical chips per socket*/
static u32 phys_coresperchip; /* Physical cores per chip */

/*
 * read_24x7_sys_info()
 * Retrieve the number of sockets and chips per socket and cores per
 * chip details through the get-system-parameter rtas call.
 */
void read_24x7_sys_info(void)
{
	int call_status, len, ntypes;

	spin_lock(&rtas_data_buf_lock);

	/*
	 * Making system parameter: chips and sockets and cores per chip
	 * default to 1.
	 */
	phys_sockets = 1;
	phys_chipspersocket = 1;
	phys_coresperchip = 1;

	call_status = rtas_call(rtas_token("ibm,get-system-parameter"), 3, 1,
				NULL,
				PROCESSOR_MODULE_INFO,
				__pa(rtas_data_buf),
				RTAS_DATA_BUF_SIZE);

	if (call_status != 0) {
		pr_err("Error calling get-system-parameter %d\n",
		       call_status);
	} else {
		len = be16_to_cpup((__be16 *)&rtas_data_buf[0]);
		if (len < 8)
			goto out;

		ntypes = be16_to_cpup((__be16 *)&rtas_data_buf[2]);

		if (!ntypes)
			goto out;

		phys_sockets = be16_to_cpup((__be16 *)&rtas_data_buf[4]);
		phys_chipspersocket = be16_to_cpup((__be16 *)&rtas_data_buf[6]);
		phys_coresperchip = be16_to_cpup((__be16 *)&rtas_data_buf[8]);
	}

out:
	spin_unlock(&rtas_data_buf_lock);
}

/* Domains for which more than one result element are returned for each event. */
static bool domain_needs_aggregation(unsigned int domain)
{
	return aggregate_result_elements &&
			(domain == HV_PERF_DOMAIN_PHYS_CORE ||
			 (domain >= HV_PERF_DOMAIN_VCPU_HOME_CORE &&
			  domain <= HV_PERF_DOMAIN_VCPU_REMOTE_NODE));
}

static const char *domain_name(unsigned domain)
{
	if (!domain_is_valid(domain))
		return NULL;

	switch (domain) {
	case HV_PERF_DOMAIN_PHYS_CHIP:		return "Physical Chip";
	case HV_PERF_DOMAIN_PHYS_CORE:		return "Physical Core";
	case HV_PERF_DOMAIN_VCPU_HOME_CORE:	return "VCPU Home Core";
	case HV_PERF_DOMAIN_VCPU_HOME_CHIP:	return "VCPU Home Chip";
	case HV_PERF_DOMAIN_VCPU_HOME_NODE:	return "VCPU Home Node";
	case HV_PERF_DOMAIN_VCPU_REMOTE_NODE:	return "VCPU Remote Node";
	}

	WARN_ON_ONCE(domain);
	return NULL;
}

static bool catalog_entry_domain_is_valid(unsigned domain)
{
	/* POWER8 doesn't support virtual domains. */
	if (interface_version == 1)
		return is_physical_domain(domain);
	else
		return domain_is_valid(domain);
}

/*
 * TODO: Merging events:
 * - Think of the hcall as an interface to a 4d array of counters:
 *   - x = domains
 *   - y = indexes in the domain (core, chip, vcpu, node, etc)
 *   - z = offset into the counter space
 *   - w = lpars (guest vms, "logical partitions")
 * - A single request is: x,y,y_last,z,z_last,w,w_last
 *   - this means we can retrieve a rectangle of counters in y,z for a single x.
 *
 * - Things to consider (ignoring w):
 *   - input  cost_per_request = 16
 *   - output cost_per_result(ys,zs)  = 8 + 8 * ys + ys * zs
 *   - limited number of requests per hcall (must fit into 4K bytes)
 *     - 4k = 16 [buffer header] - 16 [request size] * request_count
 *     - 255 requests per hcall
 *   - sometimes it will be more efficient to read extra data and discard
 */

/*
 * Example usage:
 *  perf stat -e 'hv_24x7/domain=2,offset=8,vcpu=0,lpar=0xffffffff/'
 */

/* u3 0-6, one of HV_24X7_PERF_DOMAIN */
EVENT_DEFINE_RANGE_FORMAT(domain, config, 0, 3);
/* u16 */
EVENT_DEFINE_RANGE_FORMAT(core, config, 16, 31);
EVENT_DEFINE_RANGE_FORMAT(chip, config, 16, 31);
EVENT_DEFINE_RANGE_FORMAT(vcpu, config, 16, 31);
/* u32, see "data_offset" */
EVENT_DEFINE_RANGE_FORMAT(offset, config, 32, 63);
/* u16 */
EVENT_DEFINE_RANGE_FORMAT(lpar, config1, 0, 15);

EVENT_DEFINE_RANGE(reserved1, config,   4, 15);
EVENT_DEFINE_RANGE(reserved2, config1, 16, 63);
EVENT_DEFINE_RANGE(reserved3, config2,  0, 63);

static struct attribute *format_attrs[] = {
	&format_attr_domain.attr,
	&format_attr_offset.attr,
	&format_attr_core.attr,
	&format_attr_chip.attr,
	&format_attr_vcpu.attr,
	&format_attr_lpar.attr,
	NULL,
};

static struct attribute_group format_group = {
	.name = "format",
	.attrs = format_attrs,
};

static struct attribute_group event_group = {
	.name = "events",
	/* .attrs is set in init */
};

static struct attribute_group event_desc_group = {
	.name = "event_descs",
	/* .attrs is set in init */
};

static struct attribute_group event_long_desc_group = {
	.name = "event_long_descs",
	/* .attrs is set in init */
};

static struct kmem_cache *hv_page_cache;

DEFINE_PER_CPU(int, hv_24x7_txn_flags);
DEFINE_PER_CPU(int, hv_24x7_txn_err);

struct hv_24x7_hw {
	struct perf_event *events[255];
};

DEFINE_PER_CPU(struct hv_24x7_hw, hv_24x7_hw);

/*
 * request_buffer and result_buffer are not required to be 4k aligned,
 * but are not allowed to cross any 4k boundary. Aligning them to 4k is
 * the simplest way to ensure that.
 */
#define H24x7_DATA_BUFFER_SIZE	4096
DEFINE_PER_CPU(char, hv_24x7_reqb[H24x7_DATA_BUFFER_SIZE]) __aligned(4096);
DEFINE_PER_CPU(char, hv_24x7_resb[H24x7_DATA_BUFFER_SIZE]) __aligned(4096);

static unsigned int max_num_requests(int interface_version)
{
	return (H24x7_DATA_BUFFER_SIZE - sizeof(struct hv_24x7_request_buffer))
		/ H24x7_REQUEST_SIZE(interface_version);
}

static char *event_name(struct hv_24x7_event_data *ev, int *len)
{
	*len = be16_to_cpu(ev->event_name_len) - 2;
	return (char *)ev->remainder;
}

static char *event_desc(struct hv_24x7_event_data *ev, int *len)
{
	unsigned nl = be16_to_cpu(ev->event_name_len);
	__be16 *desc_len = (__be16 *)(ev->remainder + nl - 2);

	*len = be16_to_cpu(*desc_len) - 2;
	return (char *)ev->remainder + nl;
}

static char *event_long_desc(struct hv_24x7_event_data *ev, int *len)
{
	unsigned nl = be16_to_cpu(ev->event_name_len);
	__be16 *desc_len_ = (__be16 *)(ev->remainder + nl - 2);
	unsigned desc_len = be16_to_cpu(*desc_len_);
	__be16 *long_desc_len = (__be16 *)(ev->remainder + nl + desc_len - 2);

	*len = be16_to_cpu(*long_desc_len) - 2;
	return (char *)ev->remainder + nl + desc_len;
}

static bool event_fixed_portion_is_within(struct hv_24x7_event_data *ev,
					  void *end)
{
	void *start = ev;

	return (start + offsetof(struct hv_24x7_event_data, remainder)) < end;
}

/*
 * Things we don't check:
 *  - padding for desc, name, and long/detailed desc is required to be '\0'
 *    bytes.
 *
 *  Return NULL if we pass end,
 *  Otherwise return the address of the byte just following the event.
 */
static void *event_end(struct hv_24x7_event_data *ev, void *end)
{
	void *start = ev;
	__be16 *dl_, *ldl_;
	unsigned dl, ldl;
	unsigned nl = be16_to_cpu(ev->event_name_len);

	if (nl < 2) {
		pr_debug("%s: name length too short: %d", __func__, nl);
		return NULL;
	}

	if (start + nl > end) {
		pr_debug("%s: start=%p + nl=%u > end=%p",
				__func__, start, nl, end);
		return NULL;
	}

	dl_ = (__be16 *)(ev->remainder + nl - 2);
	if (!IS_ALIGNED((uintptr_t)dl_, 2))
		pr_warn("desc len not aligned %p", dl_);
	dl = be16_to_cpu(*dl_);
	if (dl < 2) {
		pr_debug("%s: desc len too short: %d", __func__, dl);
		return NULL;
	}

	if (start + nl + dl > end) {
		pr_debug("%s: (start=%p + nl=%u + dl=%u)=%p > end=%p",
				__func__, start, nl, dl, start + nl + dl, end);
		return NULL;
	}

	ldl_ = (__be16 *)(ev->remainder + nl + dl - 2);
	if (!IS_ALIGNED((uintptr_t)ldl_, 2))
		pr_warn("long desc len not aligned %p", ldl_);
	ldl = be16_to_cpu(*ldl_);
	if (ldl < 2) {
		pr_debug("%s: long desc len too short (ldl=%u)",
				__func__, ldl);
		return NULL;
	}

	if (start + nl + dl + ldl > end) {
		pr_debug("%s: start=%p + nl=%u + dl=%u + ldl=%u > end=%p",
				__func__, start, nl, dl, ldl, end);
		return NULL;
	}

	return start + nl + dl + ldl;
}

static long h_get_24x7_catalog_page_(unsigned long phys_4096,
				     unsigned long version, unsigned long index)
{
	pr_devel("h_get_24x7_catalog_page(0x%lx, %lu, %lu)",
			phys_4096, version, index);

	WARN_ON(!IS_ALIGNED(phys_4096, 4096));

	return plpar_hcall_norets(H_GET_24X7_CATALOG_PAGE,
			phys_4096, version, index);
}

static long h_get_24x7_catalog_page(char page[], u64 version, u32 index)
{
	return h_get_24x7_catalog_page_(virt_to_phys(page),
					version, index);
}

/*
 * Each event we find in the catalog, will have a sysfs entry. Format the
 * data for this sysfs entry based on the event's domain.
 *
 * Events belonging to the Chip domain can only be monitored in that domain.
 * i.e the domain for these events is a fixed/knwon value.
 *
 * Events belonging to the Core domain can be monitored either in the physical
 * core or in one of the virtual CPU domains. So the domain value for these
 * events must be specified by the user (i.e is a required parameter). Format
 * the Core events with 'domain=?' so the perf-tool can error check required
 * parameters.
 *
 * NOTE: For the Core domain events, rather than making domain a required
 *	 parameter we could default it to PHYS_CORE and allowe users to
 *	 override the domain to one of the VCPU domains.
 *
 *	 However, this can make the interface a little inconsistent.
 *
 *	 If we set domain=2 (PHYS_CHIP) and allow user to override this field
 *	 the user may be tempted to also modify the "offset=x" field in which
 *	 can lead to confusing usage. Consider the HPM_PCYC (offset=0x18) and
 *	 HPM_INST (offset=0x20) events. With:
 *
 *		perf stat -e hv_24x7/HPM_PCYC,offset=0x20/
 *
 *	we end up monitoring HPM_INST, while the command line has HPM_PCYC.
 *
 *	By not assigning a default value to the domain for the Core events,
 *	we can have simple guidelines:
 *
 *		- Specifying values for parameters with "=?" is required.
 *
 *		- Specifying (i.e overriding) values for other parameters
 *		  is undefined.
 */
static char *event_fmt(struct hv_24x7_event_data *event, unsigned domain)
{
	const char *sindex;
	const char *lpar;
	const char *domain_str;
	char buf[8];

	switch (domain) {
	case HV_PERF_DOMAIN_PHYS_CHIP:
		snprintf(buf, sizeof(buf), "%d", domain);
		domain_str = buf;
		lpar = "0x0";
		sindex = "chip";
		break;
	case HV_PERF_DOMAIN_PHYS_CORE:
		domain_str = "?";
		lpar = "0x0";
		sindex = "core";
		break;
	default:
		domain_str = "?";
		lpar = "?";
		sindex = "vcpu";
	}

	return kasprintf(GFP_KERNEL,
			"domain=%s,offset=0x%x,%s=?,lpar=%s",
			domain_str,
			be16_to_cpu(event->event_counter_offs) +
				be16_to_cpu(event->event_group_record_offs),
			sindex,
			lpar);
}

/* Avoid trusting fw to NUL terminate strings */
static char *memdup_to_str(char *maybe_str, int max_len, gfp_t gfp)
{
	return kasprintf(gfp, "%.*s", max_len, maybe_str);
}

static ssize_t device_show_string(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	struct dev_ext_attribute *d;

	d = container_of(attr, struct dev_ext_attribute, attr);

	return sprintf(buf, "%s\n", (char *)d->var);
}

static ssize_t sockets_show(struct device *dev,
			    struct device_attribute *attr, char *buf)
{
	return sprintf(buf, "%d\n", phys_sockets);
}

static ssize_t chipspersocket_show(struct device *dev,
				   struct device_attribute *attr, char *buf)
{
	return sprintf(buf, "%d\n", phys_chipspersocket);
}

static ssize_t coresperchip_show(struct device *dev,
				 struct device_attribute *attr, char *buf)
{
	return sprintf(buf, "%d\n", phys_coresperchip);
}

static struct attribute *device_str_attr_create_(char *name, char *str)
{
	struct dev_ext_attribute *attr = kzalloc(sizeof(*attr), GFP_KERNEL);

	if (!attr)
		return NULL;

	sysfs_attr_init(&attr->attr.attr);

	attr->var = str;
	attr->attr.attr.name = name;
	attr->attr.attr.mode = 0444;
	attr->attr.show = device_show_string;

	return &attr->attr.attr;
}

/*
 * Allocate and initialize strings representing event attributes.
 *
 * NOTE: The strings allocated here are never destroyed and continue to
 *	 exist till shutdown. This is to allow us to create as many events
 *	 from the catalog as possible, even if we encounter errors with some.
 *	 In case of changes to error paths in future, these may need to be
 *	 freed by the caller.
 */
static struct attribute *device_str_attr_create(char *name, int name_max,
						int name_nonce,
						char *str, size_t str_max)
{
	char *n;
	char *s = memdup_to_str(str, str_max, GFP_KERNEL);
	struct attribute *a;

	if (!s)
		return NULL;

	if (!name_nonce)
		n = kasprintf(GFP_KERNEL, "%.*s", name_max, name);
	else
		n = kasprintf(GFP_KERNEL, "%.*s__%d", name_max, name,
					name_nonce);
	if (!n)
		goto out_s;

	a = device_str_attr_create_(n, s);
	if (!a)
		goto out_n;

	return a;
out_n:
	kfree(n);
out_s:
	kfree(s);
	return NULL;
}

static struct attribute *event_to_attr(unsigned ix,
				       struct hv_24x7_event_data *event,
				       unsigned domain,
				       int nonce)
{
	int event_name_len;
	char *ev_name, *a_ev_name, *val;
	struct attribute *attr;

	if (!domain_is_valid(domain)) {
		pr_warn("catalog event %u has invalid domain %u\n",
				ix, domain);
		return NULL;
	}

	val = event_fmt(event, domain);
	if (!val)
		return NULL;

	ev_name = event_name(event, &event_name_len);
	if (!nonce)
		a_ev_name = kasprintf(GFP_KERNEL, "%.*s",
				(int)event_name_len, ev_name);
	else
		a_ev_name = kasprintf(GFP_KERNEL, "%.*s__%d",
				(int)event_name_len, ev_name, nonce);

	if (!a_ev_name)
		goto out_val;

	attr = device_str_attr_create_(a_ev_name, val);
	if (!attr)
		goto out_name;

	return attr;
out_name:
	kfree(a_ev_name);
out_val:
	kfree(val);
	return NULL;
}

static struct attribute *event_to_desc_attr(struct hv_24x7_event_data *event,
					    int nonce)
{
	int nl, dl;
	char *name = event_name(event, &nl);
	char *desc = event_desc(event, &dl);

	/* If there isn't a description, don't create the sysfs file */
	if (!dl)
		return NULL;

	return device_str_attr_create(name, nl, nonce, desc, dl);
}

static struct attribute *
event_to_long_desc_attr(struct hv_24x7_event_data *event, int nonce)
{
	int nl, dl;
	char *name = event_name(event, &nl);
	char *desc = event_long_desc(event, &dl);

	/* If there isn't a description, don't create the sysfs file */
	if (!dl)
		return NULL;

	return device_str_attr_create(name, nl, nonce, desc, dl);
}

static int event_data_to_attrs(unsigned ix, struct attribute **attrs,
				   struct hv_24x7_event_data *event, int nonce)
{
	*attrs = event_to_attr(ix, event, event->domain, nonce);
	if (!*attrs)
		return -1;

	return 0;
}

/* */
struct event_uniq {
	struct rb_node node;
	const char *name;
	int nl;
	unsigned ct;
	unsigned domain;
};

static int memord(const void *d1, size_t s1, const void *d2, size_t s2)
{
	if (s1 < s2)
		return 1;
	if (s1 > s2)
		return -1;

	return memcmp(d1, d2, s1);
}

static int ev_uniq_ord(const void *v1, size_t s1, unsigned d1, const void *v2,
		       size_t s2, unsigned d2)
{
	int r = memord(v1, s1, v2, s2);

	if (r)
		return r;
	if (d1 > d2)
		return 1;
	if (d2 > d1)
		return -1;
	return 0;
}

static int event_uniq_add(struct rb_root *root, const char *name, int nl,
			  unsigned domain)
{
	struct rb_node **new = &(root->rb_node), *parent = NULL;
	struct event_uniq *data;

	/* Figure out where to put new node */
	while (*new) {
		struct event_uniq *it;
		int result;

		it = rb_entry(*new, struct event_uniq, node);
		result = ev_uniq_ord(name, nl, domain, it->name, it->nl,
					it->domain);

		parent = *new;
		if (result < 0)
			new = &((*new)->rb_left);
		else if (result > 0)
			new = &((*new)->rb_right);
		else {
			it->ct++;
			pr_info("found a duplicate event %.*s, ct=%u\n", nl,
						name, it->ct);
			return it->ct;
		}
	}

	data = kmalloc(sizeof(*data), GFP_KERNEL);
	if (!data)
		return -ENOMEM;

	*data = (struct event_uniq) {
		.name = name,
		.nl = nl,
		.ct = 0,
		.domain = domain,
	};

	/* Add new node and rebalance tree. */
	rb_link_node(&data->node, parent, new);
	rb_insert_color(&data->node, root);

	/* data->ct */
	return 0;
}

static void event_uniq_destroy(struct rb_root *root)
{
	/*
	 * the strings we point to are in the giant block of memory filled by
	 * the catalog, and are freed separately.
	 */
	struct event_uniq *pos, *n;

	rbtree_postorder_for_each_entry_safe(pos, n, root, node)
		kfree(pos);
}


/*
 * ensure the event structure's sizes are self consistent and don't cause us to
 * read outside of the event
 *
 * On success, return the event length in bytes.
 * Otherwise, return -1 (and print as appropriate).
 */
static ssize_t catalog_event_len_validate(struct hv_24x7_event_data *event,
					  size_t event_idx,
					  size_t event_data_bytes,
					  size_t event_entry_count,
					  size_t offset, void *end)
{
	ssize_t ev_len;
	void *ev_end, *calc_ev_end;

	if (offset >= event_data_bytes)
		return -1;

	if (event_idx >= event_entry_count) {
		pr_devel("catalog event data has %zu bytes of padding after last event\n",
				event_data_bytes - offset);
		return -1;
	}

	if (!event_fixed_portion_is_within(event, end)) {
		pr_warn("event %zu fixed portion is not within range\n",
				event_idx);
		return -1;
	}

	ev_len = be16_to_cpu(event->length);

	if (ev_len % 16)
		pr_info("event %zu has length %zu not divisible by 16: event=%pK\n",
				event_idx, ev_len, event);

	ev_end = (__u8 *)event + ev_len;
	if (ev_end > end) {
		pr_warn("event %zu has .length=%zu, ends after buffer end: ev_end=%pK > end=%pK, offset=%zu\n",
				event_idx, ev_len, ev_end, end,
				offset);
		return -1;
	}

	calc_ev_end = event_end(event, end);
	if (!calc_ev_end) {
		pr_warn("event %zu has a calculated length which exceeds buffer length %zu: event=%pK end=%pK, offset=%zu\n",
			event_idx, event_data_bytes, event, end,
			offset);
		return -1;
	}

	if (calc_ev_end > ev_end) {
		pr_warn("event %zu exceeds it's own length: event=%pK, end=%pK, offset=%zu, calc_ev_end=%pK\n",
			event_idx, event, ev_end, offset, calc_ev_end);
		return -1;
	}

	return ev_len;
}

#define MAX_4K (SIZE_MAX / 4096)

static int create_events_from_catalog(struct attribute ***events_,
				      struct attribute ***event_descs_,
				      struct attribute ***event_long_descs_)
{
	long hret;
	size_t catalog_len, catalog_page_len, event_entry_count,
	       event_data_len, event_data_offs,
	       event_data_bytes, junk_events, event_idx, event_attr_ct, i,
	       attr_max, event_idx_last, desc_ct, long_desc_ct;
	ssize_t ct, ev_len;
	uint64_t catalog_version_num;
	struct attribute **events, **event_descs, **event_long_descs;
	struct hv_24x7_catalog_page_0 *page_0 =
		kmem_cache_alloc(hv_page_cache, GFP_KERNEL);
	void *page = page_0;
	void *event_data, *end;
	struct hv_24x7_event_data *event;
	struct rb_root ev_uniq = RB_ROOT;
	int ret = 0;

	if (!page) {
		ret = -ENOMEM;
		goto e_out;
	}

	hret = h_get_24x7_catalog_page(page, 0, 0);
	if (hret) {
		ret = -EIO;
		goto e_free;
	}

	catalog_version_num = be64_to_cpu(page_0->version);
	catalog_page_len = be32_to_cpu(page_0->length);

	if (MAX_4K < catalog_page_len) {
		pr_err("invalid page count: %zu\n", catalog_page_len);
		ret = -EIO;
		goto e_free;
	}

	catalog_len = catalog_page_len * 4096;

	event_entry_count = be16_to_cpu(page_0->event_entry_count);
	event_data_offs   = be16_to_cpu(page_0->event_data_offs);
	event_data_len    = be16_to_cpu(page_0->event_data_len);

	pr_devel("cv %llu cl %zu eec %zu edo %zu edl %zu\n",
			catalog_version_num, catalog_len,
			event_entry_count, event_data_offs, event_data_len);

	if ((MAX_4K < event_data_len)
			|| (MAX_4K < event_data_offs)
			|| (MAX_4K - event_data_offs < event_data_len)) {
		pr_err("invalid event data offs %zu and/or len %zu\n",
				event_data_offs, event_data_len);
		ret = -EIO;
		goto e_free;
	}

	if ((event_data_offs + event_data_len) > catalog_page_len) {
		pr_err("event data %zu-%zu does not fit inside catalog 0-%zu\n",
				event_data_offs,
				event_data_offs + event_data_len,
				catalog_page_len);
		ret = -EIO;
		goto e_free;
	}

	if (SIZE_MAX - 1 < event_entry_count) {
		pr_err("event_entry_count %zu is invalid\n", event_entry_count);
		ret = -EIO;
		goto e_free;
	}

	event_data_bytes = event_data_len * 4096;

	/*
	 * event data can span several pages, events can cross between these
	 * pages. Use vmalloc to make this easier.
	 */
	event_data = vmalloc(event_data_bytes);
	if (!event_data) {
		pr_err("could not allocate event data\n");
		ret = -ENOMEM;
		goto e_free;
	}

	end = event_data + event_data_bytes;

	/*
	 * using vmalloc_to_phys() like this only works if PAGE_SIZE is
	 * divisible by 4096
	 */
	BUILD_BUG_ON(PAGE_SIZE % 4096);

	for (i = 0; i < event_data_len; i++) {
		hret = h_get_24x7_catalog_page_(
				vmalloc_to_phys(event_data + i * 4096),
				catalog_version_num,
				i + event_data_offs);
		if (hret) {
			pr_err("Failed to get event data in page %zu: rc=%ld\n",
			       i + event_data_offs, hret);
			ret = -EIO;
			goto e_event_data;
		}
	}

	/*
	 * scan the catalog to determine the number of attributes we need, and
	 * verify it at the same time.
	 */
	for (junk_events = 0, event = event_data, event_idx = 0, attr_max = 0;
	     ;
	     event_idx++, event = (void *)event + ev_len) {
		size_t offset = (void *)event - (void *)event_data;
		char *name;
		int nl;

		ev_len = catalog_event_len_validate(event, event_idx,
						    event_data_bytes,
						    event_entry_count,
						    offset, end);
		if (ev_len < 0)
			break;

		name = event_name(event, &nl);

		if (event->event_group_record_len == 0) {
			pr_devel("invalid event %zu (%.*s): group_record_len == 0, skipping\n",
					event_idx, nl, name);
			junk_events++;
			continue;
		}

		if (!catalog_entry_domain_is_valid(event->domain)) {
			pr_info("event %zu (%.*s) has invalid domain %d\n",
					event_idx, nl, name, event->domain);
			junk_events++;
			continue;
		}

		attr_max++;
	}

	event_idx_last = event_idx;
	if (event_idx_last != event_entry_count)
		pr_warn("event buffer ended before listed # of events were parsed (got %zu, wanted %zu, junk %zu)\n",
				event_idx_last, event_entry_count, junk_events);

	events = kmalloc_array(attr_max + 1, sizeof(*events), GFP_KERNEL);
	if (!events) {
		ret = -ENOMEM;
		goto e_event_data;
	}

	event_descs = kmalloc_array(event_idx + 1, sizeof(*event_descs),
				GFP_KERNEL);
	if (!event_descs) {
		ret = -ENOMEM;
		goto e_event_attrs;
	}

	event_long_descs = kmalloc_array(event_idx + 1,
			sizeof(*event_long_descs), GFP_KERNEL);
	if (!event_long_descs) {
		ret = -ENOMEM;
		goto e_event_descs;
	}

	/* Iterate over the catalog filling in the attribute vector */
	for (junk_events = 0, event_attr_ct = 0, desc_ct = 0, long_desc_ct = 0,
				event = event_data, event_idx = 0;
			event_idx < event_idx_last;
			event_idx++, ev_len = be16_to_cpu(event->length),
				event = (void *)event + ev_len) {
		char *name;
		int nl;
		int nonce;
		/*
		 * these are the only "bad" events that are intermixed and that
		 * we can ignore without issue. make sure to skip them here
		 */
		if (event->event_group_record_len == 0)
			continue;
		if (!catalog_entry_domain_is_valid(event->domain))
			continue;

		name  = event_name(event, &nl);
		nonce = event_uniq_add(&ev_uniq, name, nl, event->domain);
		ct    = event_data_to_attrs(event_idx, events + event_attr_ct,
					    event, nonce);
		if (ct < 0) {
			pr_warn("event %zu (%.*s) creation failure, skipping\n",
				event_idx, nl, name);
			junk_events++;
		} else {
			event_attr_ct++;
			event_descs[desc_ct] = event_to_desc_attr(event, nonce);
			if (event_descs[desc_ct])
				desc_ct++;
			event_long_descs[long_desc_ct] =
					event_to_long_desc_attr(event, nonce);
			if (event_long_descs[long_desc_ct])
				long_desc_ct++;
		}
	}

	pr_info("read %zu catalog entries, created %zu event attrs (%zu failures), %zu descs\n",
			event_idx, event_attr_ct, junk_events, desc_ct);

	events[event_attr_ct] = NULL;
	event_descs[desc_ct] = NULL;
	event_long_descs[long_desc_ct] = NULL;

	event_uniq_destroy(&ev_uniq);
	vfree(event_data);
	kmem_cache_free(hv_page_cache, page);

	*events_ = events;
	*event_descs_ = event_descs;
	*event_long_descs_ = event_long_descs;
	return 0;

e_event_descs:
	kfree(event_descs);
e_event_attrs:
	kfree(events);
e_event_data:
	vfree(event_data);
e_free:
	kmem_cache_free(hv_page_cache, page);
e_out:
	*events_ = NULL;
	*event_descs_ = NULL;
	*event_long_descs_ = NULL;
	return ret;
}

static ssize_t catalog_read(struct file *filp, struct kobject *kobj,
			    struct bin_attribute *bin_attr, char *buf,
			    loff_t offset, size_t count)
{
	long hret;
	ssize_t ret = 0;
	size_t catalog_len = 0, catalog_page_len = 0;
	loff_t page_offset = 0;
	loff_t offset_in_page;
	size_t copy_len;
	uint64_t catalog_version_num = 0;
	void *page = kmem_cache_alloc(hv_page_cache, GFP_USER);
	struct hv_24x7_catalog_page_0 *page_0 = page;

	if (!page)
		return -ENOMEM;

	hret = h_get_24x7_catalog_page(page, 0, 0);
	if (hret) {
		ret = -EIO;
		goto e_free;
	}

	catalog_version_num = be64_to_cpu(page_0->version);
	catalog_page_len = be32_to_cpu(page_0->length);
	catalog_len = catalog_page_len * 4096;

	page_offset = offset / 4096;
	offset_in_page = offset % 4096;

	if (page_offset >= catalog_page_len)
		goto e_free;

	if (page_offset != 0) {
		hret = h_get_24x7_catalog_page(page, catalog_version_num,
					       page_offset);
		if (hret) {
			ret = -EIO;
			goto e_free;
		}
	}

	copy_len = 4096 - offset_in_page;
	if (copy_len > count)
		copy_len = count;

	memcpy(buf, page+offset_in_page, copy_len);
	ret = copy_len;

e_free:
	if (hret)
		pr_err("h_get_24x7_catalog_page(ver=%lld, page=%lld) failed:"
		       " rc=%ld\n",
		       catalog_version_num, page_offset, hret);
	kmem_cache_free(hv_page_cache, page);

	pr_devel("catalog_read: offset=%lld(%lld) count=%zu "
			"catalog_len=%zu(%zu) => %zd\n", offset, page_offset,
			count, catalog_len, catalog_page_len, ret);

	return ret;
}

static ssize_t domains_show(struct device *dev, struct device_attribute *attr,
			    char *page)
{
	int d, n, count = 0;
	const char *str;

	for (d = 0; d < HV_PERF_DOMAIN_MAX; d++) {
		str = domain_name(d);
		if (!str)
			continue;

		n = sprintf(page, "%d: %s\n", d, str);
		if (n < 0)
			break;

		count += n;
		page += n;
	}
	return count;
}

#define PAGE_0_ATTR(_name, _fmt, _expr)				\
static ssize_t _name##_show(struct device *dev,			\
			    struct device_attribute *dev_attr,	\
			    char *buf)				\
{								\
	long hret;						\
	ssize_t ret = 0;					\
	void *page = kmem_cache_alloc(hv_page_cache, GFP_USER);	\
	struct hv_24x7_catalog_page_0 *page_0 = page;		\
	if (!page)						\
		return -ENOMEM;					\
	hret = h_get_24x7_catalog_page(page, 0, 0);		\
	if (hret) {						\
		ret = -EIO;					\
		goto e_free;					\
	}							\
	ret = sprintf(buf, _fmt, _expr);			\
e_free:								\
	kmem_cache_free(hv_page_cache, page);			\
	return ret;						\
}								\
static DEVICE_ATTR_RO(_name)

PAGE_0_ATTR(catalog_version, "%lld\n",
		(unsigned long long)be64_to_cpu(page_0->version));
PAGE_0_ATTR(catalog_len, "%lld\n",
		(unsigned long long)be32_to_cpu(page_0->length) * 4096);
static BIN_ATTR_RO(catalog, 0/* real length varies */);
static DEVICE_ATTR_RO(domains);
static DEVICE_ATTR_RO(sockets);
static DEVICE_ATTR_RO(chipspersocket);
static DEVICE_ATTR_RO(coresperchip);

static struct bin_attribute *if_bin_attrs[] = {
	&bin_attr_catalog,
	NULL,
};

static struct attribute *if_attrs[] = {
	&dev_attr_catalog_len.attr,
	&dev_attr_catalog_version.attr,
	&dev_attr_domains.attr,
	&dev_attr_sockets.attr,
	&dev_attr_chipspersocket.attr,
	&dev_attr_coresperchip.attr,
	NULL,
};

static struct attribute_group if_group = {
	.name = "interface",
	.bin_attrs = if_bin_attrs,
	.attrs = if_attrs,
};

static const struct attribute_group *attr_groups[] = {
	&format_group,
	&event_group,
	&event_desc_group,
	&event_long_desc_group,
	&if_group,
	NULL,
};

/*
 * Start the process for a new H_GET_24x7_DATA hcall.
 */
static void init_24x7_request(struct hv_24x7_request_buffer *request_buffer,
			      struct hv_24x7_data_result_buffer *result_buffer)
{

	memset(request_buffer, 0, H24x7_DATA_BUFFER_SIZE);
	memset(result_buffer, 0, H24x7_DATA_BUFFER_SIZE);

	request_buffer->interface_version = interface_version;
	/* memset above set request_buffer->num_requests to 0 */
}

/*
 * Commit (i.e perform) the H_GET_24x7_DATA hcall using the data collected
 * by 'init_24x7_request()' and 'add_event_to_24x7_request()'.
 */
static int make_24x7_request(struct hv_24x7_request_buffer *request_buffer,
			     struct hv_24x7_data_result_buffer *result_buffer)
{
	long ret;

	/*
	 * NOTE: Due to variable number of array elements in request and
	 *	 result buffer(s), sizeof() is not reliable. Use the actual
	 *	 allocated buffer size, H24x7_DATA_BUFFER_SIZE.
	 */
	ret = plpar_hcall_norets(H_GET_24X7_DATA,
			virt_to_phys(request_buffer), H24x7_DATA_BUFFER_SIZE,
			virt_to_phys(result_buffer),  H24x7_DATA_BUFFER_SIZE);

	if (ret) {
		struct hv_24x7_request *req;

		req = request_buffer->requests;
		pr_notice_ratelimited("hcall failed: [%d %#x %#x %d] => ret 0x%lx (%ld) detail=0x%x failing ix=%x\n",
				      req->performance_domain, req->data_offset,
				      req->starting_ix, req->starting_lpar_ix,
				      ret, ret, result_buffer->detailed_rc,
				      result_buffer->failing_request_ix);
		return -EIO;
	}

	return 0;
}

/*
 * Add the given @event to the next slot in the 24x7 request_buffer.
 *
 * Note that H_GET_24X7_DATA hcall allows reading several counters'
 * values in a single HCALL. We expect the caller to add events to the
 * request buffer one by one, make the HCALL and process the results.
 */
static int add_event_to_24x7_request(struct perf_event *event,
				struct hv_24x7_request_buffer *request_buffer)
{
	u16 idx;
	int i;
	size_t req_size;
	struct hv_24x7_request *req;

	if (request_buffer->num_requests >=
	    max_num_requests(request_buffer->interface_version)) {
		pr_devel("Too many requests for 24x7 HCALL %d\n",
				request_buffer->num_requests);
		return -EINVAL;
	}

	switch (event_get_domain(event)) {
	case HV_PERF_DOMAIN_PHYS_CHIP:
		idx = event_get_chip(event);
		break;
	case HV_PERF_DOMAIN_PHYS_CORE:
		idx = event_get_core(event);
		break;
	default:
		idx = event_get_vcpu(event);
	}

	req_size = H24x7_REQUEST_SIZE(request_buffer->interface_version);

	i = request_buffer->num_requests++;
	req = (void *) request_buffer->requests + i * req_size;

	req->performance_domain = event_get_domain(event);
	req->data_size = cpu_to_be16(8);
	req->data_offset = cpu_to_be32(event_get_offset(event));
	req->starting_lpar_ix = cpu_to_be16(event_get_lpar(event));
	req->max_num_lpars = cpu_to_be16(1);
	req->starting_ix = cpu_to_be16(idx);
	req->max_ix = cpu_to_be16(1);

	if (request_buffer->interface_version > 1) {
		if (domain_needs_aggregation(req->performance_domain))
			req->max_num_thread_groups = -1;
		else if (req->performance_domain != HV_PERF_DOMAIN_PHYS_CHIP) {
			req->starting_thread_group_ix = idx % 2;
			req->max_num_thread_groups = 1;
		}
	}

	return 0;
}

/**
 * get_count_from_result - get event count from all result elements in result
 *
 * If the event corresponding to this result needs aggregation of the result
 * element values, then this function does that.
 *
 * @event:	Event associated with @res.
 * @resb:	Result buffer containing @res.
 * @res:	Result to work on.
 * @countp:	Output variable containing the event count.
 * @next:	Optional output variable pointing to the next result in @resb.
 */
static int get_count_from_result(struct perf_event *event,
				 struct hv_24x7_data_result_buffer *resb,
				 struct hv_24x7_result *res, u64 *countp,
				 struct hv_24x7_result **next)
{
	u16 num_elements = be16_to_cpu(res->num_elements_returned);
	u16 data_size = be16_to_cpu(res->result_element_data_size);
	unsigned int data_offset;
	void *element_data;
	int i;
	u64 count;

	/*
	 * We can bail out early if the result is empty.
	 */
	if (!num_elements) {
		pr_debug("Result of request %hhu is empty, nothing to do\n",
			 res->result_ix);

		if (next)
			*next = (struct hv_24x7_result *) res->elements;

		return -ENODATA;
	}

	/*
	 * Since we always specify 1 as the maximum for the smallest resource
	 * we're requesting, there should to be only one element per result.
	 * Except when an event needs aggregation, in which case there are more.
	 */
	if (num_elements != 1 &&
	    !domain_needs_aggregation(event_get_domain(event))) {
		pr_err("Error: result of request %hhu has %hu elements\n",
		       res->result_ix, num_elements);

		return -EIO;
	}

	if (data_size != sizeof(u64)) {
		pr_debug("Error: result of request %hhu has data of %hu bytes\n",
			 res->result_ix, data_size);

		return -ENOTSUPP;
	}

	if (resb->interface_version == 1)
		data_offset = offsetof(struct hv_24x7_result_element_v1,
				       element_data);
	else
		data_offset = offsetof(struct hv_24x7_result_element_v2,
				       element_data);

	/* Go through the result elements in the result. */
	for (i = count = 0, element_data = res->elements + data_offset;
	     i < num_elements;
	     i++, element_data += data_size + data_offset)
		count += be64_to_cpu(*((u64 *) element_data));

	*countp = count;

	/* The next result is after the last result element. */
	if (next)
		*next = element_data - data_offset;

	return 0;
}

static int single_24x7_request(struct perf_event *event, u64 *count)
{
	int ret;
	struct hv_24x7_request_buffer *request_buffer;
	struct hv_24x7_data_result_buffer *result_buffer;

	BUILD_BUG_ON(sizeof(*request_buffer) > 4096);
	BUILD_BUG_ON(sizeof(*result_buffer) > 4096);

	request_buffer = (void *)get_cpu_var(hv_24x7_reqb);
	result_buffer = (void *)get_cpu_var(hv_24x7_resb);

	init_24x7_request(request_buffer, result_buffer);

	ret = add_event_to_24x7_request(event, request_buffer);
	if (ret)
		goto out;

	ret = make_24x7_request(request_buffer, result_buffer);
	if (ret)
		goto out;

	/* process result from hcall */
	ret = get_count_from_result(event, result_buffer,
				    result_buffer->results, count, NULL);

out:
	put_cpu_var(hv_24x7_reqb);
	put_cpu_var(hv_24x7_resb);
	return ret;
}


static int h_24x7_event_init(struct perf_event *event)
{
	struct hv_perf_caps caps;
	unsigned domain;
	unsigned long hret;
	u64 ct;

	/* Not our event */
	if (event->attr.type != event->pmu->type)
		return -ENOENT;

	/* Unused areas must be 0 */
	if (event_get_reserved1(event) ||
	    event_get_reserved2(event) ||
	    event_get_reserved3(event)) {
		pr_devel("reserved set when forbidden 0x%llx(0x%llx) 0x%llx(0x%llx) 0x%llx(0x%llx)\n",
				event->attr.config,
				event_get_reserved1(event),
				event->attr.config1,
				event_get_reserved2(event),
				event->attr.config2,
				event_get_reserved3(event));
		return -EINVAL;
	}

	/* no branch sampling */
	if (has_branch_stack(event))
		return -EOPNOTSUPP;

	/* offset must be 8 byte aligned */
	if (event_get_offset(event) % 8) {
		pr_devel("bad alignment\n");
		return -EINVAL;
	}

	domain = event_get_domain(event);
	if (domain >= HV_PERF_DOMAIN_MAX) {
		pr_devel("invalid domain %d\n", domain);
		return -EINVAL;
	}

	hret = hv_perf_caps_get(&caps);
	if (hret) {
		pr_devel("could not get capabilities: rc=%ld\n", hret);
		return -EIO;
	}

	/* Physical domains & other lpars require extra capabilities */
	if (!caps.collect_privileged && (is_physical_domain(domain) ||
		(event_get_lpar(event) != event_get_lpar_max()))) {
		pr_devel("hv permissions disallow: is_physical_domain:%d, lpar=0x%llx\n",
				is_physical_domain(domain),
				event_get_lpar(event));
		return -EACCES;
	}

	/* Get the initial value of the counter for this event */
	if (single_24x7_request(event, &ct)) {
		pr_devel("test hcall failed\n");
		return -EIO;
	}
	(void)local64_xchg(&event->hw.prev_count, ct);

	return 0;
}

static u64 h_24x7_get_value(struct perf_event *event)
{
	u64 ct;

	if (single_24x7_request(event, &ct))
		/* We checked this in event init, shouldn't fail here... */
		return 0;

	return ct;
}

static void update_event_count(struct perf_event *event, u64 now)
{
	s64 prev;

	prev = local64_xchg(&event->hw.prev_count, now);
	local64_add(now - prev, &event->count);
}

static void h_24x7_event_read(struct perf_event *event)
{
	u64 now;
	struct hv_24x7_request_buffer *request_buffer;
	struct hv_24x7_hw *h24x7hw;
	int txn_flags;

	txn_flags = __this_cpu_read(hv_24x7_txn_flags);

	/*
	 * If in a READ transaction, add this counter to the list of
	 * counters to read during the next HCALL (i.e commit_txn()).
	 * If not in a READ transaction, go ahead and make the HCALL
	 * to read this counter by itself.
	 */

	if (txn_flags & PERF_PMU_TXN_READ) {
		int i;
		int ret;

		if (__this_cpu_read(hv_24x7_txn_err))
			return;

		request_buffer = (void *)get_cpu_var(hv_24x7_reqb);

		ret = add_event_to_24x7_request(event, request_buffer);
		if (ret) {
			__this_cpu_write(hv_24x7_txn_err, ret);
		} else {
			/*
			 * Associate the event with the HCALL request index,
			 * so ->commit_txn() can quickly find/update count.
			 */
			i = request_buffer->num_requests - 1;

			h24x7hw = &get_cpu_var(hv_24x7_hw);
			h24x7hw->events[i] = event;
			put_cpu_var(h24x7hw);
		}

		put_cpu_var(hv_24x7_reqb);
	} else {
		now = h_24x7_get_value(event);
		update_event_count(event, now);
	}
}

static void h_24x7_event_start(struct perf_event *event, int flags)
{
	if (flags & PERF_EF_RELOAD)
		local64_set(&event->hw.prev_count, h_24x7_get_value(event));
}

static void h_24x7_event_stop(struct perf_event *event, int flags)
{
	h_24x7_event_read(event);
}

static int h_24x7_event_add(struct perf_event *event, int flags)
{
	if (flags & PERF_EF_START)
		h_24x7_event_start(event, flags);

	return 0;
}

/*
 * 24x7 counters only support READ transactions. They are
 * always counting and dont need/support ADD transactions.
 * Cache the flags, but otherwise ignore transactions that
 * are not PERF_PMU_TXN_READ.
 */
static void h_24x7_event_start_txn(struct pmu *pmu, unsigned int flags)
{
	struct hv_24x7_request_buffer *request_buffer;
	struct hv_24x7_data_result_buffer *result_buffer;

	/* We should not be called if we are already in a txn */
	WARN_ON_ONCE(__this_cpu_read(hv_24x7_txn_flags));

	__this_cpu_write(hv_24x7_txn_flags, flags);
	if (flags & ~PERF_PMU_TXN_READ)
		return;

	request_buffer = (void *)get_cpu_var(hv_24x7_reqb);
	result_buffer = (void *)get_cpu_var(hv_24x7_resb);

	init_24x7_request(request_buffer, result_buffer);

	put_cpu_var(hv_24x7_resb);
	put_cpu_var(hv_24x7_reqb);
}

/*
 * Clean up transaction state.
 *
 * NOTE: Ignore state of request and result buffers for now.
 *	 We will initialize them during the next read/txn.
 */
static void reset_txn(void)
{
	__this_cpu_write(hv_24x7_txn_flags, 0);
	__this_cpu_write(hv_24x7_txn_err, 0);
}

/*
 * 24x7 counters only support READ transactions. They are always counting
 * and dont need/support ADD transactions. Clear ->txn_flags but otherwise
 * ignore transactions that are not of type PERF_PMU_TXN_READ.
 *
 * For READ transactions, submit all pending 24x7 requests (i.e requests
 * that were queued by h_24x7_event_read()), to the hypervisor and update
 * the event counts.
 */
static int h_24x7_event_commit_txn(struct pmu *pmu)
{
	struct hv_24x7_request_buffer *request_buffer;
	struct hv_24x7_data_result_buffer *result_buffer;
	struct hv_24x7_result *res, *next_res;
	u64 count;
	int i, ret, txn_flags;
	struct hv_24x7_hw *h24x7hw;

	txn_flags = __this_cpu_read(hv_24x7_txn_flags);
	WARN_ON_ONCE(!txn_flags);

	ret = 0;
	if (txn_flags & ~PERF_PMU_TXN_READ)
		goto out;

	ret = __this_cpu_read(hv_24x7_txn_err);
	if (ret)
		goto out;

	request_buffer = (void *)get_cpu_var(hv_24x7_reqb);
	result_buffer = (void *)get_cpu_var(hv_24x7_resb);

	ret = make_24x7_request(request_buffer, result_buffer);
	if (ret)
		goto put_reqb;

	h24x7hw = &get_cpu_var(hv_24x7_hw);

	/* Go through results in the result buffer to update event counts. */
	for (i = 0, res = result_buffer->results;
	     i < result_buffer->num_results; i++, res = next_res) {
		struct perf_event *event = h24x7hw->events[res->result_ix];

		ret = get_count_from_result(event, result_buffer, res, &count,
					    &next_res);
		if (ret)
			break;

		update_event_count(event, count);
	}

	put_cpu_var(hv_24x7_hw);

put_reqb:
	put_cpu_var(hv_24x7_resb);
	put_cpu_var(hv_24x7_reqb);
out:
	reset_txn();
	return ret;
}

/*
 * 24x7 counters only support READ transactions. They are always counting
 * and dont need/support ADD transactions. However, regardless of type
 * of transaction, all we need to do is cleanup, so we don't have to check
 * the type of transaction.
 */
static void h_24x7_event_cancel_txn(struct pmu *pmu)
{
	WARN_ON_ONCE(!__this_cpu_read(hv_24x7_txn_flags));
	reset_txn();
}

static struct pmu h_24x7_pmu = {
	.task_ctx_nr = perf_invalid_context,

	.name = "hv_24x7",
	.attr_groups = attr_groups,
	.event_init  = h_24x7_event_init,
	.add         = h_24x7_event_add,
	.del         = h_24x7_event_stop,
	.start       = h_24x7_event_start,
	.stop        = h_24x7_event_stop,
	.read        = h_24x7_event_read,
	.start_txn   = h_24x7_event_start_txn,
	.commit_txn  = h_24x7_event_commit_txn,
	.cancel_txn  = h_24x7_event_cancel_txn,
	.capabilities = PERF_PMU_CAP_NO_EXCLUDE,
};

static int hv_24x7_init(void)
{
	int r;
	unsigned long hret;
	struct hv_perf_caps caps;

	if (!firmware_has_feature(FW_FEATURE_LPAR)) {
		pr_debug("not a virtualized system, not enabling\n");
		return -ENODEV;
	} else if (!cur_cpu_spec->oprofile_cpu_type)
		return -ENODEV;

	/* POWER8 only supports v1, while POWER9 only supports v2. */
	if (!strcmp(cur_cpu_spec->oprofile_cpu_type, "ppc64/power8"))
		interface_version = 1;
	else {
		interface_version = 2;

		/* SMT8 in POWER9 needs to aggregate result elements. */
		if (threads_per_core == 8)
			aggregate_result_elements = true;
	}

	hret = hv_perf_caps_get(&caps);
	if (hret) {
		pr_debug("could not obtain capabilities, not enabling, rc=%ld\n",
				hret);
		return -ENODEV;
	}

	hv_page_cache = kmem_cache_create("hv-page-4096", 4096, 4096, 0, NULL);
	if (!hv_page_cache)
		return -ENOMEM;

	/* sampling not supported */
	h_24x7_pmu.capabilities |= PERF_PMU_CAP_NO_INTERRUPT;

	r = create_events_from_catalog(&event_group.attrs,
				   &event_desc_group.attrs,
				   &event_long_desc_group.attrs);

	if (r)
		return r;

	r = perf_pmu_register(&h_24x7_pmu, h_24x7_pmu.name, -1);
	if (r)
		return r;

	read_24x7_sys_info();

	return 0;
}

device_initcall(hv_24x7_init);