Free Electrons

Embedded Linux Experts

   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
/*
 * Performance event support - powerpc architecture code
 *
 * Copyright 2008-2009 Paul Mackerras, 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.
 */
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/perf_event.h>
#include <linux/percpu.h>
#include <linux/hardirq.h>
#include <asm/reg.h>
#include <asm/pmc.h>
#include <asm/machdep.h>
#include <asm/firmware.h>
#include <asm/ptrace.h>

struct cpu_hw_events {
	int n_events;
	int n_percpu;
	int disabled;
	int n_added;
	int n_limited;
	u8  pmcs_enabled;
	struct perf_event *event[MAX_HWEVENTS];
	u64 events[MAX_HWEVENTS];
	unsigned int flags[MAX_HWEVENTS];
	unsigned long mmcr[3];
	struct perf_event *limited_counter[MAX_LIMITED_HWCOUNTERS];
	u8  limited_hwidx[MAX_LIMITED_HWCOUNTERS];
	u64 alternatives[MAX_HWEVENTS][MAX_EVENT_ALTERNATIVES];
	unsigned long amasks[MAX_HWEVENTS][MAX_EVENT_ALTERNATIVES];
	unsigned long avalues[MAX_HWEVENTS][MAX_EVENT_ALTERNATIVES];

	unsigned int group_flag;
	int n_txn_start;
};
DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events);

struct power_pmu *ppmu;

/*
 * Normally, to ignore kernel events we set the FCS (freeze counters
 * in supervisor mode) bit in MMCR0, but if the kernel runs with the
 * hypervisor bit set in the MSR, or if we are running on a processor
 * where the hypervisor bit is forced to 1 (as on Apple G5 processors),
 * then we need to use the FCHV bit to ignore kernel events.
 */
static unsigned int freeze_events_kernel = MMCR0_FCS;

/*
 * 32-bit doesn't have MMCRA but does have an MMCR2,
 * and a few other names are different.
 */
#ifdef CONFIG_PPC32

#define MMCR0_FCHV		0
#define MMCR0_PMCjCE		MMCR0_PMCnCE

#define SPRN_MMCRA		SPRN_MMCR2
#define MMCRA_SAMPLE_ENABLE	0

static inline unsigned long perf_ip_adjust(struct pt_regs *regs)
{
	return 0;
}
static inline void perf_get_data_addr(struct pt_regs *regs, u64 *addrp) { }
static inline u32 perf_get_misc_flags(struct pt_regs *regs)
{
	return 0;
}
static inline void perf_read_regs(struct pt_regs *regs) { }
static inline int perf_intr_is_nmi(struct pt_regs *regs)
{
	return 0;
}

#endif /* CONFIG_PPC32 */

/*
 * Things that are specific to 64-bit implementations.
 */
#ifdef CONFIG_PPC64

static inline unsigned long perf_ip_adjust(struct pt_regs *regs)
{
	unsigned long mmcra = regs->dsisr;

	if ((mmcra & MMCRA_SAMPLE_ENABLE) && !(ppmu->flags & PPMU_ALT_SIPR)) {
		unsigned long slot = (mmcra & MMCRA_SLOT) >> MMCRA_SLOT_SHIFT;
		if (slot > 1)
			return 4 * (slot - 1);
	}
	return 0;
}

/*
 * The user wants a data address recorded.
 * If we're not doing instruction sampling, give them the SDAR
 * (sampled data address).  If we are doing instruction sampling, then
 * only give them the SDAR if it corresponds to the instruction
 * pointed to by SIAR; this is indicated by the [POWER6_]MMCRA_SDSYNC
 * bit in MMCRA.
 */
static inline void perf_get_data_addr(struct pt_regs *regs, u64 *addrp)
{
	unsigned long mmcra = regs->dsisr;
	unsigned long sdsync = (ppmu->flags & PPMU_ALT_SIPR) ?
		POWER6_MMCRA_SDSYNC : MMCRA_SDSYNC;

	if (!(mmcra & MMCRA_SAMPLE_ENABLE) || (mmcra & sdsync))
		*addrp = mfspr(SPRN_SDAR);
}

static inline u32 perf_flags_from_msr(struct pt_regs *regs)
{
	if (regs->msr & MSR_PR)
		return PERF_RECORD_MISC_USER;
	if ((regs->msr & MSR_HV) && freeze_events_kernel != MMCR0_FCHV)
		return PERF_RECORD_MISC_HYPERVISOR;
	return PERF_RECORD_MISC_KERNEL;
}

static inline u32 perf_get_misc_flags(struct pt_regs *regs)
{
	unsigned long mmcra = regs->dsisr;
	unsigned long sihv = MMCRA_SIHV;
	unsigned long sipr = MMCRA_SIPR;

	/* Not a PMU interrupt: Make up flags from regs->msr */
	if (TRAP(regs) != 0xf00)
		return perf_flags_from_msr(regs);

	/*
	 * If we don't support continuous sampling and this
	 * is not a marked event, same deal
	 */
	if ((ppmu->flags & PPMU_NO_CONT_SAMPLING) &&
	    !(mmcra & MMCRA_SAMPLE_ENABLE))
		return perf_flags_from_msr(regs);

	/*
	 * If we don't have flags in MMCRA, rather than using
	 * the MSR, we intuit the flags from the address in
	 * SIAR which should give slightly more reliable
	 * results
	 */
	if (ppmu->flags & PPMU_NO_SIPR) {
		unsigned long siar = mfspr(SPRN_SIAR);
		if (siar >= PAGE_OFFSET)
			return PERF_RECORD_MISC_KERNEL;
		return PERF_RECORD_MISC_USER;
	}

	if (ppmu->flags & PPMU_ALT_SIPR) {
		sihv = POWER6_MMCRA_SIHV;
		sipr = POWER6_MMCRA_SIPR;
	}

	/* PR has priority over HV, so order below is important */
	if (mmcra & sipr)
		return PERF_RECORD_MISC_USER;
	if ((mmcra & sihv) && (freeze_events_kernel != MMCR0_FCHV))
		return PERF_RECORD_MISC_HYPERVISOR;
	return PERF_RECORD_MISC_KERNEL;
}

/*
 * Overload regs->dsisr to store MMCRA so we only need to read it once
 * on each interrupt.
 */
static inline void perf_read_regs(struct pt_regs *regs)
{
	regs->dsisr = mfspr(SPRN_MMCRA);
}

/*
 * If interrupts were soft-disabled when a PMU interrupt occurs, treat
 * it as an NMI.
 */
static inline int perf_intr_is_nmi(struct pt_regs *regs)
{
	return !regs->softe;
}

#endif /* CONFIG_PPC64 */

static void perf_event_interrupt(struct pt_regs *regs);

void perf_event_print_debug(void)
{
}

/*
 * Read one performance monitor counter (PMC).
 */
static unsigned long read_pmc(int idx)
{
	unsigned long val;

	switch (idx) {
	case 1:
		val = mfspr(SPRN_PMC1);
		break;
	case 2:
		val = mfspr(SPRN_PMC2);
		break;
	case 3:
		val = mfspr(SPRN_PMC3);
		break;
	case 4:
		val = mfspr(SPRN_PMC4);
		break;
	case 5:
		val = mfspr(SPRN_PMC5);
		break;
	case 6:
		val = mfspr(SPRN_PMC6);
		break;
#ifdef CONFIG_PPC64
	case 7:
		val = mfspr(SPRN_PMC7);
		break;
	case 8:
		val = mfspr(SPRN_PMC8);
		break;
#endif /* CONFIG_PPC64 */
	default:
		printk(KERN_ERR "oops trying to read PMC%d\n", idx);
		val = 0;
	}
	return val;
}

/*
 * Write one PMC.
 */
static void write_pmc(int idx, unsigned long val)
{
	switch (idx) {
	case 1:
		mtspr(SPRN_PMC1, val);
		break;
	case 2:
		mtspr(SPRN_PMC2, val);
		break;
	case 3:
		mtspr(SPRN_PMC3, val);
		break;
	case 4:
		mtspr(SPRN_PMC4, val);
		break;
	case 5:
		mtspr(SPRN_PMC5, val);
		break;
	case 6:
		mtspr(SPRN_PMC6, val);
		break;
#ifdef CONFIG_PPC64
	case 7:
		mtspr(SPRN_PMC7, val);
		break;
	case 8:
		mtspr(SPRN_PMC8, val);
		break;
#endif /* CONFIG_PPC64 */
	default:
		printk(KERN_ERR "oops trying to write PMC%d\n", idx);
	}
}

/*
 * Check if a set of events can all go on the PMU at once.
 * If they can't, this will look at alternative codes for the events
 * and see if any combination of alternative codes is feasible.
 * The feasible set is returned in event_id[].
 */
static int power_check_constraints(struct cpu_hw_events *cpuhw,
				   u64 event_id[], unsigned int cflags[],
				   int n_ev)
{
	unsigned long mask, value, nv;
	unsigned long smasks[MAX_HWEVENTS], svalues[MAX_HWEVENTS];
	int n_alt[MAX_HWEVENTS], choice[MAX_HWEVENTS];
	int i, j;
	unsigned long addf = ppmu->add_fields;
	unsigned long tadd = ppmu->test_adder;

	if (n_ev > ppmu->n_counter)
		return -1;

	/* First see if the events will go on as-is */
	for (i = 0; i < n_ev; ++i) {
		if ((cflags[i] & PPMU_LIMITED_PMC_REQD)
		    && !ppmu->limited_pmc_event(event_id[i])) {
			ppmu->get_alternatives(event_id[i], cflags[i],
					       cpuhw->alternatives[i]);
			event_id[i] = cpuhw->alternatives[i][0];
		}
		if (ppmu->get_constraint(event_id[i], &cpuhw->amasks[i][0],
					 &cpuhw->avalues[i][0]))
			return -1;
	}
	value = mask = 0;
	for (i = 0; i < n_ev; ++i) {
		nv = (value | cpuhw->avalues[i][0]) +
			(value & cpuhw->avalues[i][0] & addf);
		if ((((nv + tadd) ^ value) & mask) != 0 ||
		    (((nv + tadd) ^ cpuhw->avalues[i][0]) &
		     cpuhw->amasks[i][0]) != 0)
			break;
		value = nv;
		mask |= cpuhw->amasks[i][0];
	}
	if (i == n_ev)
		return 0;	/* all OK */

	/* doesn't work, gather alternatives... */
	if (!ppmu->get_alternatives)
		return -1;
	for (i = 0; i < n_ev; ++i) {
		choice[i] = 0;
		n_alt[i] = ppmu->get_alternatives(event_id[i], cflags[i],
						  cpuhw->alternatives[i]);
		for (j = 1; j < n_alt[i]; ++j)
			ppmu->get_constraint(cpuhw->alternatives[i][j],
					     &cpuhw->amasks[i][j],
					     &cpuhw->avalues[i][j]);
	}

	/* enumerate all possibilities and see if any will work */
	i = 0;
	j = -1;
	value = mask = nv = 0;
	while (i < n_ev) {
		if (j >= 0) {
			/* we're backtracking, restore context */
			value = svalues[i];
			mask = smasks[i];
			j = choice[i];
		}
		/*
		 * See if any alternative k for event_id i,
		 * where k > j, will satisfy the constraints.
		 */
		while (++j < n_alt[i]) {
			nv = (value | cpuhw->avalues[i][j]) +
				(value & cpuhw->avalues[i][j] & addf);
			if ((((nv + tadd) ^ value) & mask) == 0 &&
			    (((nv + tadd) ^ cpuhw->avalues[i][j])
			     & cpuhw->amasks[i][j]) == 0)
				break;
		}
		if (j >= n_alt[i]) {
			/*
			 * No feasible alternative, backtrack
			 * to event_id i-1 and continue enumerating its
			 * alternatives from where we got up to.
			 */
			if (--i < 0)
				return -1;
		} else {
			/*
			 * Found a feasible alternative for event_id i,
			 * remember where we got up to with this event_id,
			 * go on to the next event_id, and start with
			 * the first alternative for it.
			 */
			choice[i] = j;
			svalues[i] = value;
			smasks[i] = mask;
			value = nv;
			mask |= cpuhw->amasks[i][j];
			++i;
			j = -1;
		}
	}

	/* OK, we have a feasible combination, tell the caller the solution */
	for (i = 0; i < n_ev; ++i)
		event_id[i] = cpuhw->alternatives[i][choice[i]];
	return 0;
}

/*
 * Check if newly-added events have consistent settings for
 * exclude_{user,kernel,hv} with each other and any previously
 * added events.
 */
static int check_excludes(struct perf_event **ctrs, unsigned int cflags[],
			  int n_prev, int n_new)
{
	int eu = 0, ek = 0, eh = 0;
	int i, n, first;
	struct perf_event *event;

	n = n_prev + n_new;
	if (n <= 1)
		return 0;

	first = 1;
	for (i = 0; i < n; ++i) {
		if (cflags[i] & PPMU_LIMITED_PMC_OK) {
			cflags[i] &= ~PPMU_LIMITED_PMC_REQD;
			continue;
		}
		event = ctrs[i];
		if (first) {
			eu = event->attr.exclude_user;
			ek = event->attr.exclude_kernel;
			eh = event->attr.exclude_hv;
			first = 0;
		} else if (event->attr.exclude_user != eu ||
			   event->attr.exclude_kernel != ek ||
			   event->attr.exclude_hv != eh) {
			return -EAGAIN;
		}
	}

	if (eu || ek || eh)
		for (i = 0; i < n; ++i)
			if (cflags[i] & PPMU_LIMITED_PMC_OK)
				cflags[i] |= PPMU_LIMITED_PMC_REQD;

	return 0;
}

static u64 check_and_compute_delta(u64 prev, u64 val)
{
	u64 delta = (val - prev) & 0xfffffffful;

	/*
	 * POWER7 can roll back counter values, if the new value is smaller
	 * than the previous value it will cause the delta and the counter to
	 * have bogus values unless we rolled a counter over.  If a coutner is
	 * rolled back, it will be smaller, but within 256, which is the maximum
	 * number of events to rollback at once.  If we dectect a rollback
	 * return 0.  This can lead to a small lack of precision in the
	 * counters.
	 */
	if (prev > val && (prev - val) < 256)
		delta = 0;

	return delta;
}

static void power_pmu_read(struct perf_event *event)
{
	s64 val, delta, prev;

	if (event->hw.state & PERF_HES_STOPPED)
		return;

	if (!event->hw.idx)
		return;
	/*
	 * Performance monitor interrupts come even when interrupts
	 * are soft-disabled, as long as interrupts are hard-enabled.
	 * Therefore we treat them like NMIs.
	 */
	do {
		prev = local64_read(&event->hw.prev_count);
		barrier();
		val = read_pmc(event->hw.idx);
		delta = check_and_compute_delta(prev, val);
		if (!delta)
			return;
	} while (local64_cmpxchg(&event->hw.prev_count, prev, val) != prev);

	local64_add(delta, &event->count);
	local64_sub(delta, &event->hw.period_left);
}

/*
 * On some machines, PMC5 and PMC6 can't be written, don't respect
 * the freeze conditions, and don't generate interrupts.  This tells
 * us if `event' is using such a PMC.
 */
static int is_limited_pmc(int pmcnum)
{
	return (ppmu->flags & PPMU_LIMITED_PMC5_6)
		&& (pmcnum == 5 || pmcnum == 6);
}

static void freeze_limited_counters(struct cpu_hw_events *cpuhw,
				    unsigned long pmc5, unsigned long pmc6)
{
	struct perf_event *event;
	u64 val, prev, delta;
	int i;

	for (i = 0; i < cpuhw->n_limited; ++i) {
		event = cpuhw->limited_counter[i];
		if (!event->hw.idx)
			continue;
		val = (event->hw.idx == 5) ? pmc5 : pmc6;
		prev = local64_read(&event->hw.prev_count);
		event->hw.idx = 0;
		delta = check_and_compute_delta(prev, val);
		if (delta)
			local64_add(delta, &event->count);
	}
}

static void thaw_limited_counters(struct cpu_hw_events *cpuhw,
				  unsigned long pmc5, unsigned long pmc6)
{
	struct perf_event *event;
	u64 val, prev;
	int i;

	for (i = 0; i < cpuhw->n_limited; ++i) {
		event = cpuhw->limited_counter[i];
		event->hw.idx = cpuhw->limited_hwidx[i];
		val = (event->hw.idx == 5) ? pmc5 : pmc6;
		prev = local64_read(&event->hw.prev_count);
		if (check_and_compute_delta(prev, val))
			local64_set(&event->hw.prev_count, val);
		perf_event_update_userpage(event);
	}
}

/*
 * Since limited events don't respect the freeze conditions, we
 * have to read them immediately after freezing or unfreezing the
 * other events.  We try to keep the values from the limited
 * events as consistent as possible by keeping the delay (in
 * cycles and instructions) between freezing/unfreezing and reading
 * the limited events as small and consistent as possible.
 * Therefore, if any limited events are in use, we read them
 * both, and always in the same order, to minimize variability,
 * and do it inside the same asm that writes MMCR0.
 */
static void write_mmcr0(struct cpu_hw_events *cpuhw, unsigned long mmcr0)
{
	unsigned long pmc5, pmc6;

	if (!cpuhw->n_limited) {
		mtspr(SPRN_MMCR0, mmcr0);
		return;
	}

	/*
	 * Write MMCR0, then read PMC5 and PMC6 immediately.
	 * To ensure we don't get a performance monitor interrupt
	 * between writing MMCR0 and freezing/thawing the limited
	 * events, we first write MMCR0 with the event overflow
	 * interrupt enable bits turned off.
	 */
	asm volatile("mtspr %3,%2; mfspr %0,%4; mfspr %1,%5"
		     : "=&r" (pmc5), "=&r" (pmc6)
		     : "r" (mmcr0 & ~(MMCR0_PMC1CE | MMCR0_PMCjCE)),
		       "i" (SPRN_MMCR0),
		       "i" (SPRN_PMC5), "i" (SPRN_PMC6));

	if (mmcr0 & MMCR0_FC)
		freeze_limited_counters(cpuhw, pmc5, pmc6);
	else
		thaw_limited_counters(cpuhw, pmc5, pmc6);

	/*
	 * Write the full MMCR0 including the event overflow interrupt
	 * enable bits, if necessary.
	 */
	if (mmcr0 & (MMCR0_PMC1CE | MMCR0_PMCjCE))
		mtspr(SPRN_MMCR0, mmcr0);
}

/*
 * Disable all events to prevent PMU interrupts and to allow
 * events to be added or removed.
 */
static void power_pmu_disable(struct pmu *pmu)
{
	struct cpu_hw_events *cpuhw;
	unsigned long flags;

	if (!ppmu)
		return;
	local_irq_save(flags);
	cpuhw = &__get_cpu_var(cpu_hw_events);

	if (!cpuhw->disabled) {
		cpuhw->disabled = 1;
		cpuhw->n_added = 0;

		/*
		 * Check if we ever enabled the PMU on this cpu.
		 */
		if (!cpuhw->pmcs_enabled) {
			ppc_enable_pmcs();
			cpuhw->pmcs_enabled = 1;
		}

		/*
		 * Disable instruction sampling if it was enabled
		 */
		if (cpuhw->mmcr[2] & MMCRA_SAMPLE_ENABLE) {
			mtspr(SPRN_MMCRA,
			      cpuhw->mmcr[2] & ~MMCRA_SAMPLE_ENABLE);
			mb();
		}

		/*
		 * Set the 'freeze counters' bit.
		 * The barrier is to make sure the mtspr has been
		 * executed and the PMU has frozen the events
		 * before we return.
		 */
		write_mmcr0(cpuhw, mfspr(SPRN_MMCR0) | MMCR0_FC);
		mb();
	}
	local_irq_restore(flags);
}

/*
 * Re-enable all events if disable == 0.
 * If we were previously disabled and events were added, then
 * put the new config on the PMU.
 */
static void power_pmu_enable(struct pmu *pmu)
{
	struct perf_event *event;
	struct cpu_hw_events *cpuhw;
	unsigned long flags;
	long i;
	unsigned long val;
	s64 left;
	unsigned int hwc_index[MAX_HWEVENTS];
	int n_lim;
	int idx;

	if (!ppmu)
		return;
	local_irq_save(flags);
	cpuhw = &__get_cpu_var(cpu_hw_events);
	if (!cpuhw->disabled) {
		local_irq_restore(flags);
		return;
	}
	cpuhw->disabled = 0;

	/*
	 * If we didn't change anything, or only removed events,
	 * no need to recalculate MMCR* settings and reset the PMCs.
	 * Just reenable the PMU with the current MMCR* settings
	 * (possibly updated for removal of events).
	 */
	if (!cpuhw->n_added) {
		mtspr(SPRN_MMCRA, cpuhw->mmcr[2] & ~MMCRA_SAMPLE_ENABLE);
		mtspr(SPRN_MMCR1, cpuhw->mmcr[1]);
		if (cpuhw->n_events == 0)
			ppc_set_pmu_inuse(0);
		goto out_enable;
	}

	/*
	 * Compute MMCR* values for the new set of events
	 */
	if (ppmu->compute_mmcr(cpuhw->events, cpuhw->n_events, hwc_index,
			       cpuhw->mmcr)) {
		/* shouldn't ever get here */
		printk(KERN_ERR "oops compute_mmcr failed\n");
		goto out;
	}

	/*
	 * Add in MMCR0 freeze bits corresponding to the
	 * attr.exclude_* bits for the first event.
	 * We have already checked that all events have the
	 * same values for these bits as the first event.
	 */
	event = cpuhw->event[0];
	if (event->attr.exclude_user)
		cpuhw->mmcr[0] |= MMCR0_FCP;
	if (event->attr.exclude_kernel)
		cpuhw->mmcr[0] |= freeze_events_kernel;
	if (event->attr.exclude_hv)
		cpuhw->mmcr[0] |= MMCR0_FCHV;

	/*
	 * Write the new configuration to MMCR* with the freeze
	 * bit set and set the hardware events to their initial values.
	 * Then unfreeze the events.
	 */
	ppc_set_pmu_inuse(1);
	mtspr(SPRN_MMCRA, cpuhw->mmcr[2] & ~MMCRA_SAMPLE_ENABLE);
	mtspr(SPRN_MMCR1, cpuhw->mmcr[1]);
	mtspr(SPRN_MMCR0, (cpuhw->mmcr[0] & ~(MMCR0_PMC1CE | MMCR0_PMCjCE))
				| MMCR0_FC);

	/*
	 * Read off any pre-existing events that need to move
	 * to another PMC.
	 */
	for (i = 0; i < cpuhw->n_events; ++i) {
		event = cpuhw->event[i];
		if (event->hw.idx && event->hw.idx != hwc_index[i] + 1) {
			power_pmu_read(event);
			write_pmc(event->hw.idx, 0);
			event->hw.idx = 0;
		}
	}

	/*
	 * Initialize the PMCs for all the new and moved events.
	 */
	cpuhw->n_limited = n_lim = 0;
	for (i = 0; i < cpuhw->n_events; ++i) {
		event = cpuhw->event[i];
		if (event->hw.idx)
			continue;
		idx = hwc_index[i] + 1;
		if (is_limited_pmc(idx)) {
			cpuhw->limited_counter[n_lim] = event;
			cpuhw->limited_hwidx[n_lim] = idx;
			++n_lim;
			continue;
		}
		val = 0;
		if (event->hw.sample_period) {
			left = local64_read(&event->hw.period_left);
			if (left < 0x80000000L)
				val = 0x80000000L - left;
		}
		local64_set(&event->hw.prev_count, val);
		event->hw.idx = idx;
		if (event->hw.state & PERF_HES_STOPPED)
			val = 0;
		write_pmc(idx, val);
		perf_event_update_userpage(event);
	}
	cpuhw->n_limited = n_lim;
	cpuhw->mmcr[0] |= MMCR0_PMXE | MMCR0_FCECE;

 out_enable:
	mb();
	write_mmcr0(cpuhw, cpuhw->mmcr[0]);

	/*
	 * Enable instruction sampling if necessary
	 */
	if (cpuhw->mmcr[2] & MMCRA_SAMPLE_ENABLE) {
		mb();
		mtspr(SPRN_MMCRA, cpuhw->mmcr[2]);
	}

 out:
	local_irq_restore(flags);
}

static int collect_events(struct perf_event *group, int max_count,
			  struct perf_event *ctrs[], u64 *events,
			  unsigned int *flags)
{
	int n = 0;
	struct perf_event *event;

	if (!is_software_event(group)) {
		if (n >= max_count)
			return -1;
		ctrs[n] = group;
		flags[n] = group->hw.event_base;
		events[n++] = group->hw.config;
	}
	list_for_each_entry(event, &group->sibling_list, group_entry) {
		if (!is_software_event(event) &&
		    event->state != PERF_EVENT_STATE_OFF) {
			if (n >= max_count)
				return -1;
			ctrs[n] = event;
			flags[n] = event->hw.event_base;
			events[n++] = event->hw.config;
		}
	}
	return n;
}

/*
 * Add a event to the PMU.
 * If all events are not already frozen, then we disable and
 * re-enable the PMU in order to get hw_perf_enable to do the
 * actual work of reconfiguring the PMU.
 */
static int power_pmu_add(struct perf_event *event, int ef_flags)
{
	struct cpu_hw_events *cpuhw;
	unsigned long flags;
	int n0;
	int ret = -EAGAIN;

	local_irq_save(flags);
	perf_pmu_disable(event->pmu);

	/*
	 * Add the event to the list (if there is room)
	 * and check whether the total set is still feasible.
	 */
	cpuhw = &__get_cpu_var(cpu_hw_events);
	n0 = cpuhw->n_events;
	if (n0 >= ppmu->n_counter)
		goto out;
	cpuhw->event[n0] = event;
	cpuhw->events[n0] = event->hw.config;
	cpuhw->flags[n0] = event->hw.event_base;

	if (!(ef_flags & PERF_EF_START))
		event->hw.state = PERF_HES_STOPPED | PERF_HES_UPTODATE;

	/*
	 * If group events scheduling transaction was started,
	 * skip the schedulability test here, it will be performed
	 * at commit time(->commit_txn) as a whole
	 */
	if (cpuhw->group_flag & PERF_EVENT_TXN)
		goto nocheck;

	if (check_excludes(cpuhw->event, cpuhw->flags, n0, 1))
		goto out;
	if (power_check_constraints(cpuhw, cpuhw->events, cpuhw->flags, n0 + 1))
		goto out;
	event->hw.config = cpuhw->events[n0];

nocheck:
	++cpuhw->n_events;
	++cpuhw->n_added;

	ret = 0;
 out:
	perf_pmu_enable(event->pmu);
	local_irq_restore(flags);
	return ret;
}

/*
 * Remove a event from the PMU.
 */
static void power_pmu_del(struct perf_event *event, int ef_flags)
{
	struct cpu_hw_events *cpuhw;
	long i;
	unsigned long flags;

	local_irq_save(flags);
	perf_pmu_disable(event->pmu);

	power_pmu_read(event);

	cpuhw = &__get_cpu_var(cpu_hw_events);
	for (i = 0; i < cpuhw->n_events; ++i) {
		if (event == cpuhw->event[i]) {
			while (++i < cpuhw->n_events) {
				cpuhw->event[i-1] = cpuhw->event[i];
				cpuhw->events[i-1] = cpuhw->events[i];
				cpuhw->flags[i-1] = cpuhw->flags[i];
			}
			--cpuhw->n_events;
			ppmu->disable_pmc(event->hw.idx - 1, cpuhw->mmcr);
			if (event->hw.idx) {
				write_pmc(event->hw.idx, 0);
				event->hw.idx = 0;
			}
			perf_event_update_userpage(event);
			break;
		}
	}
	for (i = 0; i < cpuhw->n_limited; ++i)
		if (event == cpuhw->limited_counter[i])
			break;
	if (i < cpuhw->n_limited) {
		while (++i < cpuhw->n_limited) {
			cpuhw->limited_counter[i-1] = cpuhw->limited_counter[i];
			cpuhw->limited_hwidx[i-1] = cpuhw->limited_hwidx[i];
		}
		--cpuhw->n_limited;
	}
	if (cpuhw->n_events == 0) {
		/* disable exceptions if no events are running */
		cpuhw->mmcr[0] &= ~(MMCR0_PMXE | MMCR0_FCECE);
	}

	perf_pmu_enable(event->pmu);
	local_irq_restore(flags);
}

/*
 * POWER-PMU does not support disabling individual counters, hence
 * program their cycle counter to their max value and ignore the interrupts.
 */

static void power_pmu_start(struct perf_event *event, int ef_flags)
{
	unsigned long flags;
	s64 left;
	unsigned long val;

	if (!event->hw.idx || !event->hw.sample_period)
		return;

	if (!(event->hw.state & PERF_HES_STOPPED))
		return;

	if (ef_flags & PERF_EF_RELOAD)
		WARN_ON_ONCE(!(event->hw.state & PERF_HES_UPTODATE));

	local_irq_save(flags);
	perf_pmu_disable(event->pmu);

	event->hw.state = 0;
	left = local64_read(&event->hw.period_left);

	val = 0;
	if (left < 0x80000000L)
		val = 0x80000000L - left;

	write_pmc(event->hw.idx, val);

	perf_event_update_userpage(event);
	perf_pmu_enable(event->pmu);
	local_irq_restore(flags);
}

static void power_pmu_stop(struct perf_event *event, int ef_flags)
{
	unsigned long flags;

	if (!event->hw.idx || !event->hw.sample_period)
		return;

	if (event->hw.state & PERF_HES_STOPPED)
		return;

	local_irq_save(flags);
	perf_pmu_disable(event->pmu);

	power_pmu_read(event);
	event->hw.state |= PERF_HES_STOPPED | PERF_HES_UPTODATE;
	write_pmc(event->hw.idx, 0);

	perf_event_update_userpage(event);
	perf_pmu_enable(event->pmu);
	local_irq_restore(flags);
}

/*
 * Start group events scheduling transaction
 * Set the flag to make pmu::enable() not perform the
 * schedulability test, it will be performed at commit time
 */
void power_pmu_start_txn(struct pmu *pmu)
{
	struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events);

	perf_pmu_disable(pmu);
	cpuhw->group_flag |= PERF_EVENT_TXN;
	cpuhw->n_txn_start = cpuhw->n_events;
}

/*
 * Stop group events scheduling transaction
 * Clear the flag and pmu::enable() will perform the
 * schedulability test.
 */
void power_pmu_cancel_txn(struct pmu *pmu)
{
	struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events);

	cpuhw->group_flag &= ~PERF_EVENT_TXN;
	perf_pmu_enable(pmu);
}

/*
 * Commit group events scheduling transaction
 * Perform the group schedulability test as a whole
 * Return 0 if success
 */
int power_pmu_commit_txn(struct pmu *pmu)
{
	struct cpu_hw_events *cpuhw;
	long i, n;

	if (!ppmu)
		return -EAGAIN;
	cpuhw = &__get_cpu_var(cpu_hw_events);
	n = cpuhw->n_events;
	if (check_excludes(cpuhw->event, cpuhw->flags, 0, n))
		return -EAGAIN;
	i = power_check_constraints(cpuhw, cpuhw->events, cpuhw->flags, n);
	if (i < 0)
		return -EAGAIN;

	for (i = cpuhw->n_txn_start; i < n; ++i)
		cpuhw->event[i]->hw.config = cpuhw->events[i];

	cpuhw->group_flag &= ~PERF_EVENT_TXN;
	perf_pmu_enable(pmu);
	return 0;
}

/*
 * Return 1 if we might be able to put event on a limited PMC,
 * or 0 if not.
 * A event can only go on a limited PMC if it counts something
 * that a limited PMC can count, doesn't require interrupts, and
 * doesn't exclude any processor mode.
 */
static int can_go_on_limited_pmc(struct perf_event *event, u64 ev,
				 unsigned int flags)
{
	int n;
	u64 alt[MAX_EVENT_ALTERNATIVES];

	if (event->attr.exclude_user
	    || event->attr.exclude_kernel
	    || event->attr.exclude_hv
	    || event->attr.sample_period)
		return 0;

	if (ppmu->limited_pmc_event(ev))
		return 1;

	/*
	 * The requested event_id isn't on a limited PMC already;
	 * see if any alternative code goes on a limited PMC.
	 */
	if (!ppmu->get_alternatives)
		return 0;

	flags |= PPMU_LIMITED_PMC_OK | PPMU_LIMITED_PMC_REQD;
	n = ppmu->get_alternatives(ev, flags, alt);

	return n > 0;
}

/*
 * Find an alternative event_id that goes on a normal PMC, if possible,
 * and return the event_id code, or 0 if there is no such alternative.
 * (Note: event_id code 0 is "don't count" on all machines.)
 */
static u64 normal_pmc_alternative(u64 ev, unsigned long flags)
{
	u64 alt[MAX_EVENT_ALTERNATIVES];
	int n;

	flags &= ~(PPMU_LIMITED_PMC_OK | PPMU_LIMITED_PMC_REQD);
	n = ppmu->get_alternatives(ev, flags, alt);
	if (!n)
		return 0;
	return alt[0];
}

/* Number of perf_events counting hardware events */
static atomic_t num_events;
/* Used to avoid races in calling reserve/release_pmc_hardware */
static DEFINE_MUTEX(pmc_reserve_mutex);

/*
 * Release the PMU if this is the last perf_event.
 */
static void hw_perf_event_destroy(struct perf_event *event)
{
	if (!atomic_add_unless(&num_events, -1, 1)) {
		mutex_lock(&pmc_reserve_mutex);
		if (atomic_dec_return(&num_events) == 0)
			release_pmc_hardware();
		mutex_unlock(&pmc_reserve_mutex);
	}
}

/*
 * Translate a generic cache event_id config to a raw event_id code.
 */
static int hw_perf_cache_event(u64 config, u64 *eventp)
{
	unsigned long type, op, result;
	int ev;

	if (!ppmu->cache_events)
		return -EINVAL;

	/* unpack config */
	type = config & 0xff;
	op = (config >> 8) & 0xff;
	result = (config >> 16) & 0xff;

	if (type >= PERF_COUNT_HW_CACHE_MAX ||
	    op >= PERF_COUNT_HW_CACHE_OP_MAX ||
	    result >= PERF_COUNT_HW_CACHE_RESULT_MAX)
		return -EINVAL;

	ev = (*ppmu->cache_events)[type][op][result];
	if (ev == 0)
		return -EOPNOTSUPP;
	if (ev == -1)
		return -EINVAL;
	*eventp = ev;
	return 0;
}

static int power_pmu_event_init(struct perf_event *event)
{
	u64 ev;
	unsigned long flags;
	struct perf_event *ctrs[MAX_HWEVENTS];
	u64 events[MAX_HWEVENTS];
	unsigned int cflags[MAX_HWEVENTS];
	int n;
	int err;
	struct cpu_hw_events *cpuhw;

	if (!ppmu)
		return -ENOENT;

	/* does not support taken branch sampling */
	if (has_branch_stack(event))
		return -EOPNOTSUPP;

	switch (event->attr.type) {
	case PERF_TYPE_HARDWARE:
		ev = event->attr.config;
		if (ev >= ppmu->n_generic || ppmu->generic_events[ev] == 0)
			return -EOPNOTSUPP;
		ev = ppmu->generic_events[ev];
		break;
	case PERF_TYPE_HW_CACHE:
		err = hw_perf_cache_event(event->attr.config, &ev);
		if (err)
			return err;
		break;
	case PERF_TYPE_RAW:
		ev = event->attr.config;
		break;
	default:
		return -ENOENT;
	}

	event->hw.config_base = ev;
	event->hw.idx = 0;

	/*
	 * If we are not running on a hypervisor, force the
	 * exclude_hv bit to 0 so that we don't care what
	 * the user set it to.
	 */
	if (!firmware_has_feature(FW_FEATURE_LPAR))
		event->attr.exclude_hv = 0;

	/*
	 * If this is a per-task event, then we can use
	 * PM_RUN_* events interchangeably with their non RUN_*
	 * equivalents, e.g. PM_RUN_CYC instead of PM_CYC.
	 * XXX we should check if the task is an idle task.
	 */
	flags = 0;
	if (event->attach_state & PERF_ATTACH_TASK)
		flags |= PPMU_ONLY_COUNT_RUN;

	/*
	 * If this machine has limited events, check whether this
	 * event_id could go on a limited event.
	 */
	if (ppmu->flags & PPMU_LIMITED_PMC5_6) {
		if (can_go_on_limited_pmc(event, ev, flags)) {
			flags |= PPMU_LIMITED_PMC_OK;
		} else if (ppmu->limited_pmc_event(ev)) {
			/*
			 * The requested event_id is on a limited PMC,
			 * but we can't use a limited PMC; see if any
			 * alternative goes on a normal PMC.
			 */
			ev = normal_pmc_alternative(ev, flags);
			if (!ev)
				return -EINVAL;
		}
	}

	/*
	 * If this is in a group, check if it can go on with all the
	 * other hardware events in the group.  We assume the event
	 * hasn't been linked into its leader's sibling list at this point.
	 */
	n = 0;
	if (event->group_leader != event) {
		n = collect_events(event->group_leader, ppmu->n_counter - 1,
				   ctrs, events, cflags);
		if (n < 0)
			return -EINVAL;
	}
	events[n] = ev;
	ctrs[n] = event;
	cflags[n] = flags;
	if (check_excludes(ctrs, cflags, n, 1))
		return -EINVAL;

	cpuhw = &get_cpu_var(cpu_hw_events);
	err = power_check_constraints(cpuhw, events, cflags, n + 1);
	put_cpu_var(cpu_hw_events);
	if (err)
		return -EINVAL;

	event->hw.config = events[n];
	event->hw.event_base = cflags[n];
	event->hw.last_period = event->hw.sample_period;
	local64_set(&event->hw.period_left, event->hw.last_period);

	/*
	 * See if we need to reserve the PMU.
	 * If no events are currently in use, then we have to take a
	 * mutex to ensure that we don't race with another task doing
	 * reserve_pmc_hardware or release_pmc_hardware.
	 */
	err = 0;
	if (!atomic_inc_not_zero(&num_events)) {
		mutex_lock(&pmc_reserve_mutex);
		if (atomic_read(&num_events) == 0 &&
		    reserve_pmc_hardware(perf_event_interrupt))
			err = -EBUSY;
		else
			atomic_inc(&num_events);
		mutex_unlock(&pmc_reserve_mutex);
	}
	event->destroy = hw_perf_event_destroy;

	return err;
}

static int power_pmu_event_idx(struct perf_event *event)
{
	return event->hw.idx;
}

struct pmu power_pmu = {
	.pmu_enable	= power_pmu_enable,
	.pmu_disable	= power_pmu_disable,
	.event_init	= power_pmu_event_init,
	.add		= power_pmu_add,
	.del		= power_pmu_del,
	.start		= power_pmu_start,
	.stop		= power_pmu_stop,
	.read		= power_pmu_read,
	.start_txn	= power_pmu_start_txn,
	.cancel_txn	= power_pmu_cancel_txn,
	.commit_txn	= power_pmu_commit_txn,
	.event_idx	= power_pmu_event_idx,
};

/*
 * A counter has overflowed; update its count and record
 * things if requested.  Note that interrupts are hard-disabled
 * here so there is no possibility of being interrupted.
 */
static void record_and_restart(struct perf_event *event, unsigned long val,
			       struct pt_regs *regs)
{
	u64 period = event->hw.sample_period;
	s64 prev, delta, left;
	int record = 0;

	if (event->hw.state & PERF_HES_STOPPED) {
		write_pmc(event->hw.idx, 0);
		return;
	}

	/* we don't have to worry about interrupts here */
	prev = local64_read(&event->hw.prev_count);
	delta = check_and_compute_delta(prev, val);
	local64_add(delta, &event->count);

	/*
	 * See if the total period for this event has expired,
	 * and update for the next period.
	 */
	val = 0;
	left = local64_read(&event->hw.period_left) - delta;
	if (period) {
		if (left <= 0) {
			left += period;
			if (left <= 0)
				left = period;
			record = 1;
			event->hw.last_period = event->hw.sample_period;
		}
		if (left < 0x80000000LL)
			val = 0x80000000LL - left;
	}

	write_pmc(event->hw.idx, val);
	local64_set(&event->hw.prev_count, val);
	local64_set(&event->hw.period_left, left);
	perf_event_update_userpage(event);

	/*
	 * Finally record data if requested.
	 */
	if (record) {
		struct perf_sample_data data;

		perf_sample_data_init(&data, ~0ULL);
		data.period = event->hw.last_period;

		if (event->attr.sample_type & PERF_SAMPLE_ADDR)
			perf_get_data_addr(regs, &data.addr);

		if (perf_event_overflow(event, &data, regs))
			power_pmu_stop(event, 0);
	}
}

/*
 * Called from generic code to get the misc flags (i.e. processor mode)
 * for an event_id.
 */
unsigned long perf_misc_flags(struct pt_regs *regs)
{
	u32 flags = perf_get_misc_flags(regs);

	if (flags)
		return flags;
	return user_mode(regs) ? PERF_RECORD_MISC_USER :
		PERF_RECORD_MISC_KERNEL;
}

/*
 * Called from generic code to get the instruction pointer
 * for an event_id.
 */
unsigned long perf_instruction_pointer(struct pt_regs *regs)
{
	unsigned long mmcra = regs->dsisr;

	/* Not a PMU interrupt */
	if (TRAP(regs) != 0xf00)
		return regs->nip;

	/* Processor doesn't support sampling non marked events */
	if ((ppmu->flags & PPMU_NO_CONT_SAMPLING) &&
	    !(mmcra & MMCRA_SAMPLE_ENABLE))
		return regs->nip;

	return mfspr(SPRN_SIAR) + perf_ip_adjust(regs);
}

static bool pmc_overflow(unsigned long val)
{
	if ((int)val < 0)
		return true;

	/*
	 * Events on POWER7 can roll back if a speculative event doesn't
	 * eventually complete. Unfortunately in some rare cases they will
	 * raise a performance monitor exception. We need to catch this to
	 * ensure we reset the PMC. In all cases the PMC will be 256 or less
	 * cycles from overflow.
	 *
	 * We only do this if the first pass fails to find any overflowing
	 * PMCs because a user might set a period of less than 256 and we
	 * don't want to mistakenly reset them.
	 */
	if (__is_processor(PV_POWER7) && ((0x80000000 - val) <= 256))
		return true;

	return false;
}

/*
 * Performance monitor interrupt stuff
 */
static void perf_event_interrupt(struct pt_regs *regs)
{
	int i;
	struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events);
	struct perf_event *event;
	unsigned long val;
	int found = 0;
	int nmi;

	if (cpuhw->n_limited)
		freeze_limited_counters(cpuhw, mfspr(SPRN_PMC5),
					mfspr(SPRN_PMC6));

	perf_read_regs(regs);

	nmi = perf_intr_is_nmi(regs);
	if (nmi)
		nmi_enter();
	else
		irq_enter();

	for (i = 0; i < cpuhw->n_events; ++i) {
		event = cpuhw->event[i];
		if (!event->hw.idx || is_limited_pmc(event->hw.idx))
			continue;
		val = read_pmc(event->hw.idx);
		if ((int)val < 0) {
			/* event has overflowed */
			found = 1;
			record_and_restart(event, val, regs);
		}
	}

	/*
	 * In case we didn't find and reset the event that caused
	 * the interrupt, scan all events and reset any that are
	 * negative, to avoid getting continual interrupts.
	 * Any that we processed in the previous loop will not be negative.
	 */
	if (!found) {
		for (i = 0; i < ppmu->n_counter; ++i) {
			if (is_limited_pmc(i + 1))
				continue;
			val = read_pmc(i + 1);
			if (pmc_overflow(val))
				write_pmc(i + 1, 0);
		}
	}

	/*
	 * Reset MMCR0 to its normal value.  This will set PMXE and
	 * clear FC (freeze counters) and PMAO (perf mon alert occurred)
	 * and thus allow interrupts to occur again.
	 * XXX might want to use MSR.PM to keep the events frozen until
	 * we get back out of this interrupt.
	 */
	write_mmcr0(cpuhw, cpuhw->mmcr[0]);

	if (nmi)
		nmi_exit();
	else
		irq_exit();
}

static void power_pmu_setup(int cpu)
{
	struct cpu_hw_events *cpuhw = &per_cpu(cpu_hw_events, cpu);

	if (!ppmu)
		return;
	memset(cpuhw, 0, sizeof(*cpuhw));
	cpuhw->mmcr[0] = MMCR0_FC;
}

static int __cpuinit
power_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
{
	unsigned int cpu = (long)hcpu;

	switch (action & ~CPU_TASKS_FROZEN) {
	case CPU_UP_PREPARE:
		power_pmu_setup(cpu);
		break;

	default:
		break;
	}

	return NOTIFY_OK;
}

int __cpuinit register_power_pmu(struct power_pmu *pmu)
{
	if (ppmu)
		return -EBUSY;		/* something's already registered */

	ppmu = pmu;
	pr_info("%s performance monitor hardware support registered\n",
		pmu->name);

#ifdef MSR_HV
	/*
	 * Use FCHV to ignore kernel events if MSR.HV is set.
	 */
	if (mfmsr() & MSR_HV)
		freeze_events_kernel = MMCR0_FCHV;
#endif /* CONFIG_PPC64 */

	perf_pmu_register(&power_pmu, "cpu", PERF_TYPE_RAW);
	perf_cpu_notifier(power_pmu_notifier);

	return 0;
}