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
/*
 * I/O SAPIC support.
 *
 * Copyright (C) 1999 Intel Corp.
 * Copyright (C) 1999 Asit Mallick <asit.k.mallick@intel.com>
 * Copyright (C) 2000-2002 J.I. Lee <jung-ik.lee@intel.com>
 * Copyright (C) 1999-2000, 2002-2003 Hewlett-Packard Co.
 *	David Mosberger-Tang <davidm@hpl.hp.com>
 * Copyright (C) 1999 VA Linux Systems
 * Copyright (C) 1999,2000 Walt Drummond <drummond@valinux.com>
 *
 * 00/04/19	D. Mosberger	Rewritten to mirror more closely the x86 I/O
 *				APIC code.  In particular, we now have separate
 *				handlers for edge and level triggered
 *				interrupts.
 * 00/10/27	Asit Mallick, Goutham Rao <goutham.rao@intel.com> IRQ vector
 *				allocation PCI to vector mapping, shared PCI
 *				interrupts.
 * 00/10/27	D. Mosberger	Document things a bit more to make them more
 *				understandable.  Clean up much of the old
 *				IOSAPIC cruft.
 * 01/07/27	J.I. Lee	PCI irq routing, Platform/Legacy interrupts
 *				and fixes for ACPI S5(SoftOff) support.
 * 02/01/23	J.I. Lee	iosapic pgm fixes for PCI irq routing from _PRT
 * 02/01/07     E. Focht        <efocht@ess.nec.de> Redirectable interrupt
 *				vectors in iosapic_set_affinity(),
 *				initializations for /proc/irq/#/smp_affinity
 * 02/04/02	P. Diefenbaugh	Cleaned up ACPI PCI IRQ routing.
 * 02/04/18	J.I. Lee	bug fix in iosapic_init_pci_irq
 * 02/04/30	J.I. Lee	bug fix in find_iosapic to fix ACPI PCI IRQ to
 *				IOSAPIC mapping error
 * 02/07/29	T. Kochi	Allocate interrupt vectors dynamically
 * 02/08/04	T. Kochi	Cleaned up terminology (irq, global system
 *				interrupt, vector, etc.)
 * 02/09/20	D. Mosberger	Simplified by taking advantage of ACPI's
 *				pci_irq code.
 * 03/02/19	B. Helgaas	Make pcat_compat system-wide, not per-IOSAPIC.
 *				Remove iosapic_address & gsi_base from
 *				external interfaces.  Rationalize
 *				__init/__devinit attributes.
 * 04/12/04 Ashok Raj	<ashok.raj@intel.com> Intel Corporation 2004
 *				Updated to work with irq migration necessary
 *				for CPU Hotplug
 */
/*
 * Here is what the interrupt logic between a PCI device and the kernel looks
 * like:
 *
 * (1) A PCI device raises one of the four interrupt pins (INTA, INTB, INTC,
 *     INTD).  The device is uniquely identified by its bus-, and slot-number
 *     (the function number does not matter here because all functions share
 *     the same interrupt lines).
 *
 * (2) The motherboard routes the interrupt line to a pin on a IOSAPIC
 *     controller.  Multiple interrupt lines may have to share the same
 *     IOSAPIC pin (if they're level triggered and use the same polarity).
 *     Each interrupt line has a unique Global System Interrupt (GSI) number
 *     which can be calculated as the sum of the controller's base GSI number
 *     and the IOSAPIC pin number to which the line connects.
 *
 * (3) The IOSAPIC uses an internal routing table entries (RTEs) to map the
 * IOSAPIC pin into the IA-64 interrupt vector.  This interrupt vector is then
 * sent to the CPU.
 *
 * (4) The kernel recognizes an interrupt as an IRQ.  The IRQ interface is
 *     used as architecture-independent interrupt handling mechanism in Linux.
 *     As an IRQ is a number, we have to have
 *     IA-64 interrupt vector number <-> IRQ number mapping.  On smaller
 *     systems, we use one-to-one mapping between IA-64 vector and IRQ.  A
 *     platform can implement platform_irq_to_vector(irq) and
 *     platform_local_vector_to_irq(vector) APIs to differentiate the mapping.
 *     Please see also arch/ia64/include/asm/hw_irq.h for those APIs.
 *
 * To sum up, there are three levels of mappings involved:
 *
 *	PCI pin -> global system interrupt (GSI) -> IA-64 vector <-> IRQ
 *
 * Note: The term "IRQ" is loosely used everywhere in Linux kernel to
 * describeinterrupts.  Now we use "IRQ" only for Linux IRQ's.  ISA IRQ
 * (isa_irq) is the only exception in this source code.
 */

#include <linux/acpi.h>
#include <linux/init.h>
#include <linux/irq.h>
#include <linux/kernel.h>
#include <linux/list.h>
#include <linux/pci.h>
#include <linux/slab.h>
#include <linux/smp.h>
#include <linux/string.h>
#include <linux/bootmem.h>

#include <asm/delay.h>
#include <asm/hw_irq.h>
#include <asm/io.h>
#include <asm/iosapic.h>
#include <asm/machvec.h>
#include <asm/processor.h>
#include <asm/ptrace.h>

#undef DEBUG_INTERRUPT_ROUTING

#ifdef DEBUG_INTERRUPT_ROUTING
#define DBG(fmt...)	printk(fmt)
#else
#define DBG(fmt...)
#endif

static DEFINE_SPINLOCK(iosapic_lock);

/*
 * These tables map IA-64 vectors to the IOSAPIC pin that generates this
 * vector.
 */

#define NO_REF_RTE	0

static struct iosapic {
	char __iomem	*addr;		/* base address of IOSAPIC */
	unsigned int	gsi_base;	/* GSI base */
	unsigned short	num_rte;	/* # of RTEs on this IOSAPIC */
	int		rtes_inuse;	/* # of RTEs in use on this IOSAPIC */
#ifdef CONFIG_NUMA
	unsigned short	node;		/* numa node association via pxm */
#endif
	spinlock_t	lock;		/* lock for indirect reg access */
} iosapic_lists[NR_IOSAPICS];

struct iosapic_rte_info {
	struct list_head rte_list;	/* RTEs sharing the same vector */
	char		rte_index;	/* IOSAPIC RTE index */
	int		refcnt;		/* reference counter */
	struct iosapic	*iosapic;
} ____cacheline_aligned;

static struct iosapic_intr_info {
	struct list_head rtes;		/* RTEs using this vector (empty =>
					 * not an IOSAPIC interrupt) */
	int		count;		/* # of registered RTEs */
	u32		low32;		/* current value of low word of
					 * Redirection table entry */
	unsigned int	dest;		/* destination CPU physical ID */
	unsigned char	dmode	: 3;	/* delivery mode (see iosapic.h) */
	unsigned char 	polarity: 1;	/* interrupt polarity
					 * (see iosapic.h) */
	unsigned char	trigger	: 1;	/* trigger mode (see iosapic.h) */
} iosapic_intr_info[NR_IRQS];

static unsigned char pcat_compat __devinitdata;	/* 8259 compatibility flag */

static inline void
iosapic_write(struct iosapic *iosapic, unsigned int reg, u32 val)
{
	unsigned long flags;

	spin_lock_irqsave(&iosapic->lock, flags);
	__iosapic_write(iosapic->addr, reg, val);
	spin_unlock_irqrestore(&iosapic->lock, flags);
}

/*
 * Find an IOSAPIC associated with a GSI
 */
static inline int
find_iosapic (unsigned int gsi)
{
	int i;

	for (i = 0; i < NR_IOSAPICS; i++) {
		if ((unsigned) (gsi - iosapic_lists[i].gsi_base) <
		    iosapic_lists[i].num_rte)
			return i;
	}

	return -1;
}

static inline int __gsi_to_irq(unsigned int gsi)
{
	int irq;
	struct iosapic_intr_info *info;
	struct iosapic_rte_info *rte;

	for (irq = 0; irq < NR_IRQS; irq++) {
		info = &iosapic_intr_info[irq];
		list_for_each_entry(rte, &info->rtes, rte_list)
			if (rte->iosapic->gsi_base + rte->rte_index == gsi)
				return irq;
	}
	return -1;
}

int
gsi_to_irq (unsigned int gsi)
{
	unsigned long flags;
	int irq;

	spin_lock_irqsave(&iosapic_lock, flags);
	irq = __gsi_to_irq(gsi);
	spin_unlock_irqrestore(&iosapic_lock, flags);
	return irq;
}

static struct iosapic_rte_info *find_rte(unsigned int irq, unsigned int gsi)
{
	struct iosapic_rte_info *rte;

	list_for_each_entry(rte, &iosapic_intr_info[irq].rtes, rte_list)
		if (rte->iosapic->gsi_base + rte->rte_index == gsi)
			return rte;
	return NULL;
}

static void
set_rte (unsigned int gsi, unsigned int irq, unsigned int dest, int mask)
{
	unsigned long pol, trigger, dmode;
	u32 low32, high32;
	int rte_index;
	char redir;
	struct iosapic_rte_info *rte;
	ia64_vector vector = irq_to_vector(irq);

	DBG(KERN_DEBUG"IOSAPIC: routing vector %d to 0x%x\n", vector, dest);

	rte = find_rte(irq, gsi);
	if (!rte)
		return;		/* not an IOSAPIC interrupt */

	rte_index = rte->rte_index;
	pol     = iosapic_intr_info[irq].polarity;
	trigger = iosapic_intr_info[irq].trigger;
	dmode   = iosapic_intr_info[irq].dmode;

	redir = (dmode == IOSAPIC_LOWEST_PRIORITY) ? 1 : 0;

#ifdef CONFIG_SMP
	set_irq_affinity_info(irq, (int)(dest & 0xffff), redir);
#endif

	low32 = ((pol << IOSAPIC_POLARITY_SHIFT) |
		 (trigger << IOSAPIC_TRIGGER_SHIFT) |
		 (dmode << IOSAPIC_DELIVERY_SHIFT) |
		 ((mask ? 1 : 0) << IOSAPIC_MASK_SHIFT) |
		 vector);

	/* dest contains both id and eid */
	high32 = (dest << IOSAPIC_DEST_SHIFT);

	iosapic_write(rte->iosapic, IOSAPIC_RTE_HIGH(rte_index), high32);
	iosapic_write(rte->iosapic, IOSAPIC_RTE_LOW(rte_index), low32);
	iosapic_intr_info[irq].low32 = low32;
	iosapic_intr_info[irq].dest = dest;
}

static void
nop (struct irq_data *data)
{
	/* do nothing... */
}


#ifdef CONFIG_KEXEC
void
kexec_disable_iosapic(void)
{
	struct iosapic_intr_info *info;
	struct iosapic_rte_info *rte;
	ia64_vector vec;
	int irq;

	for (irq = 0; irq < NR_IRQS; irq++) {
		info = &iosapic_intr_info[irq];
		vec = irq_to_vector(irq);
		list_for_each_entry(rte, &info->rtes,
				rte_list) {
			iosapic_write(rte->iosapic,
					IOSAPIC_RTE_LOW(rte->rte_index),
					IOSAPIC_MASK|vec);
			iosapic_eoi(rte->iosapic->addr, vec);
		}
	}
}
#endif

static void
mask_irq (struct irq_data *data)
{
	unsigned int irq = data->irq;
	u32 low32;
	int rte_index;
	struct iosapic_rte_info *rte;

	if (!iosapic_intr_info[irq].count)
		return;			/* not an IOSAPIC interrupt! */

	/* set only the mask bit */
	low32 = iosapic_intr_info[irq].low32 |= IOSAPIC_MASK;
	list_for_each_entry(rte, &iosapic_intr_info[irq].rtes, rte_list) {
		rte_index = rte->rte_index;
		iosapic_write(rte->iosapic, IOSAPIC_RTE_LOW(rte_index), low32);
	}
}

static void
unmask_irq (struct irq_data *data)
{
	unsigned int irq = data->irq;
	u32 low32;
	int rte_index;
	struct iosapic_rte_info *rte;

	if (!iosapic_intr_info[irq].count)
		return;			/* not an IOSAPIC interrupt! */

	low32 = iosapic_intr_info[irq].low32 &= ~IOSAPIC_MASK;
	list_for_each_entry(rte, &iosapic_intr_info[irq].rtes, rte_list) {
		rte_index = rte->rte_index;
		iosapic_write(rte->iosapic, IOSAPIC_RTE_LOW(rte_index), low32);
	}
}


static int
iosapic_set_affinity(struct irq_data *data, const struct cpumask *mask,
		     bool force)
{
#ifdef CONFIG_SMP
	unsigned int irq = data->irq;
	u32 high32, low32;
	int cpu, dest, rte_index;
	int redir = (irq & IA64_IRQ_REDIRECTED) ? 1 : 0;
	struct iosapic_rte_info *rte;
	struct iosapic *iosapic;

	irq &= (~IA64_IRQ_REDIRECTED);

	cpu = cpumask_first_and(cpu_online_mask, mask);
	if (cpu >= nr_cpu_ids)
		return -1;

	if (irq_prepare_move(irq, cpu))
		return -1;

	dest = cpu_physical_id(cpu);

	if (!iosapic_intr_info[irq].count)
		return -1;			/* not an IOSAPIC interrupt */

	set_irq_affinity_info(irq, dest, redir);

	/* dest contains both id and eid */
	high32 = dest << IOSAPIC_DEST_SHIFT;

	low32 = iosapic_intr_info[irq].low32 & ~(7 << IOSAPIC_DELIVERY_SHIFT);
	if (redir)
		/* change delivery mode to lowest priority */
		low32 |= (IOSAPIC_LOWEST_PRIORITY << IOSAPIC_DELIVERY_SHIFT);
	else
		/* change delivery mode to fixed */
		low32 |= (IOSAPIC_FIXED << IOSAPIC_DELIVERY_SHIFT);
	low32 &= IOSAPIC_VECTOR_MASK;
	low32 |= irq_to_vector(irq);

	iosapic_intr_info[irq].low32 = low32;
	iosapic_intr_info[irq].dest = dest;
	list_for_each_entry(rte, &iosapic_intr_info[irq].rtes, rte_list) {
		iosapic = rte->iosapic;
		rte_index = rte->rte_index;
		iosapic_write(iosapic, IOSAPIC_RTE_HIGH(rte_index), high32);
		iosapic_write(iosapic, IOSAPIC_RTE_LOW(rte_index), low32);
	}

#endif
	return 0;
}

/*
 * Handlers for level-triggered interrupts.
 */

static unsigned int
iosapic_startup_level_irq (struct irq_data *data)
{
	unmask_irq(data);
	return 0;
}

static void
iosapic_unmask_level_irq (struct irq_data *data)
{
	unsigned int irq = data->irq;
	ia64_vector vec = irq_to_vector(irq);
	struct iosapic_rte_info *rte;
	int do_unmask_irq = 0;

	irq_complete_move(irq);
	if (unlikely(irqd_is_setaffinity_pending(data))) {
		do_unmask_irq = 1;
		mask_irq(data);
	} else
		unmask_irq(data);

	list_for_each_entry(rte, &iosapic_intr_info[irq].rtes, rte_list)
		iosapic_eoi(rte->iosapic->addr, vec);

	if (unlikely(do_unmask_irq)) {
		irq_move_masked_irq(data);
		unmask_irq(data);
	}
}

#define iosapic_shutdown_level_irq	mask_irq
#define iosapic_enable_level_irq	unmask_irq
#define iosapic_disable_level_irq	mask_irq
#define iosapic_ack_level_irq		nop

static struct irq_chip irq_type_iosapic_level = {
	.name =			"IO-SAPIC-level",
	.irq_startup =		iosapic_startup_level_irq,
	.irq_shutdown =		iosapic_shutdown_level_irq,
	.irq_enable =		iosapic_enable_level_irq,
	.irq_disable =		iosapic_disable_level_irq,
	.irq_ack =		iosapic_ack_level_irq,
	.irq_mask =		mask_irq,
	.irq_unmask =		iosapic_unmask_level_irq,
	.irq_set_affinity =	iosapic_set_affinity
};

/*
 * Handlers for edge-triggered interrupts.
 */

static unsigned int
iosapic_startup_edge_irq (struct irq_data *data)
{
	unmask_irq(data);
	/*
	 * IOSAPIC simply drops interrupts pended while the
	 * corresponding pin was masked, so we can't know if an
	 * interrupt is pending already.  Let's hope not...
	 */
	return 0;
}

static void
iosapic_ack_edge_irq (struct irq_data *data)
{
	irq_complete_move(data->irq);
	irq_move_irq(data);
}

#define iosapic_enable_edge_irq		unmask_irq
#define iosapic_disable_edge_irq	nop

static struct irq_chip irq_type_iosapic_edge = {
	.name =			"IO-SAPIC-edge",
	.irq_startup =		iosapic_startup_edge_irq,
	.irq_shutdown =		iosapic_disable_edge_irq,
	.irq_enable =		iosapic_enable_edge_irq,
	.irq_disable =		iosapic_disable_edge_irq,
	.irq_ack =		iosapic_ack_edge_irq,
	.irq_mask =		mask_irq,
	.irq_unmask =		unmask_irq,
	.irq_set_affinity =	iosapic_set_affinity
};

static unsigned int
iosapic_version (char __iomem *addr)
{
	/*
	 * IOSAPIC Version Register return 32 bit structure like:
	 * {
	 *	unsigned int version   : 8;
	 *	unsigned int reserved1 : 8;
	 *	unsigned int max_redir : 8;
	 *	unsigned int reserved2 : 8;
	 * }
	 */
	return __iosapic_read(addr, IOSAPIC_VERSION);
}

static int iosapic_find_sharable_irq(unsigned long trigger, unsigned long pol)
{
	int i, irq = -ENOSPC, min_count = -1;
	struct iosapic_intr_info *info;

	/*
	 * shared vectors for edge-triggered interrupts are not
	 * supported yet
	 */
	if (trigger == IOSAPIC_EDGE)
		return -EINVAL;

	for (i = 0; i < NR_IRQS; i++) {
		info = &iosapic_intr_info[i];
		if (info->trigger == trigger && info->polarity == pol &&
		    (info->dmode == IOSAPIC_FIXED ||
		     info->dmode == IOSAPIC_LOWEST_PRIORITY) &&
		    can_request_irq(i, IRQF_SHARED)) {
			if (min_count == -1 || info->count < min_count) {
				irq = i;
				min_count = info->count;
			}
		}
	}
	return irq;
}

/*
 * if the given vector is already owned by other,
 *  assign a new vector for the other and make the vector available
 */
static void __init
iosapic_reassign_vector (int irq)
{
	int new_irq;

	if (iosapic_intr_info[irq].count) {
		new_irq = create_irq();
		if (new_irq < 0)
			panic("%s: out of interrupt vectors!\n", __func__);
		printk(KERN_INFO "Reassigning vector %d to %d\n",
		       irq_to_vector(irq), irq_to_vector(new_irq));
		memcpy(&iosapic_intr_info[new_irq], &iosapic_intr_info[irq],
		       sizeof(struct iosapic_intr_info));
		INIT_LIST_HEAD(&iosapic_intr_info[new_irq].rtes);
		list_move(iosapic_intr_info[irq].rtes.next,
			  &iosapic_intr_info[new_irq].rtes);
		memset(&iosapic_intr_info[irq], 0,
		       sizeof(struct iosapic_intr_info));
		iosapic_intr_info[irq].low32 = IOSAPIC_MASK;
		INIT_LIST_HEAD(&iosapic_intr_info[irq].rtes);
	}
}

static inline int irq_is_shared (int irq)
{
	return (iosapic_intr_info[irq].count > 1);
}

struct irq_chip*
ia64_native_iosapic_get_irq_chip(unsigned long trigger)
{
	if (trigger == IOSAPIC_EDGE)
		return &irq_type_iosapic_edge;
	else
		return &irq_type_iosapic_level;
}

static int
register_intr (unsigned int gsi, int irq, unsigned char delivery,
	       unsigned long polarity, unsigned long trigger)
{
	struct irq_chip *chip, *irq_type;
	int index;
	struct iosapic_rte_info *rte;

	index = find_iosapic(gsi);
	if (index < 0) {
		printk(KERN_WARNING "%s: No IOSAPIC for GSI %u\n",
		       __func__, gsi);
		return -ENODEV;
	}

	rte = find_rte(irq, gsi);
	if (!rte) {
		rte = kzalloc(sizeof (*rte), GFP_ATOMIC);
		if (!rte) {
			printk(KERN_WARNING "%s: cannot allocate memory\n",
			       __func__);
			return -ENOMEM;
		}

		rte->iosapic	= &iosapic_lists[index];
		rte->rte_index	= gsi - rte->iosapic->gsi_base;
		rte->refcnt++;
		list_add_tail(&rte->rte_list, &iosapic_intr_info[irq].rtes);
		iosapic_intr_info[irq].count++;
		iosapic_lists[index].rtes_inuse++;
	}
	else if (rte->refcnt == NO_REF_RTE) {
		struct iosapic_intr_info *info = &iosapic_intr_info[irq];
		if (info->count > 0 &&
		    (info->trigger != trigger || info->polarity != polarity)){
			printk (KERN_WARNING
				"%s: cannot override the interrupt\n",
				__func__);
			return -EINVAL;
		}
		rte->refcnt++;
		iosapic_intr_info[irq].count++;
		iosapic_lists[index].rtes_inuse++;
	}

	iosapic_intr_info[irq].polarity = polarity;
	iosapic_intr_info[irq].dmode    = delivery;
	iosapic_intr_info[irq].trigger  = trigger;

	irq_type = iosapic_get_irq_chip(trigger);

	chip = irq_get_chip(irq);
	if (irq_type != NULL && chip != irq_type) {
		if (chip != &no_irq_chip)
			printk(KERN_WARNING
			       "%s: changing vector %d from %s to %s\n",
			       __func__, irq_to_vector(irq),
			       chip->name, irq_type->name);
		chip = irq_type;
	}
	__irq_set_chip_handler_name_locked(irq, chip, trigger == IOSAPIC_EDGE ?
					   handle_edge_irq : handle_level_irq,
					   NULL);
	return 0;
}

static unsigned int
get_target_cpu (unsigned int gsi, int irq)
{
#ifdef CONFIG_SMP
	static int cpu = -1;
	extern int cpe_vector;
	cpumask_t domain = irq_to_domain(irq);

	/*
	 * In case of vector shared by multiple RTEs, all RTEs that
	 * share the vector need to use the same destination CPU.
	 */
	if (iosapic_intr_info[irq].count)
		return iosapic_intr_info[irq].dest;

	/*
	 * If the platform supports redirection via XTP, let it
	 * distribute interrupts.
	 */
	if (smp_int_redirect & SMP_IRQ_REDIRECTION)
		return cpu_physical_id(smp_processor_id());

	/*
	 * Some interrupts (ACPI SCI, for instance) are registered
	 * before the BSP is marked as online.
	 */
	if (!cpu_online(smp_processor_id()))
		return cpu_physical_id(smp_processor_id());

#ifdef CONFIG_ACPI
	if (cpe_vector > 0 && irq_to_vector(irq) == IA64_CPEP_VECTOR)
		return get_cpei_target_cpu();
#endif

#ifdef CONFIG_NUMA
	{
		int num_cpus, cpu_index, iosapic_index, numa_cpu, i = 0;
		const struct cpumask *cpu_mask;

		iosapic_index = find_iosapic(gsi);
		if (iosapic_index < 0 ||
		    iosapic_lists[iosapic_index].node == MAX_NUMNODES)
			goto skip_numa_setup;

		cpu_mask = cpumask_of_node(iosapic_lists[iosapic_index].node);
		num_cpus = 0;
		for_each_cpu_and(numa_cpu, cpu_mask, &domain) {
			if (cpu_online(numa_cpu))
				num_cpus++;
		}

		if (!num_cpus)
			goto skip_numa_setup;

		/* Use irq assignment to distribute across cpus in node */
		cpu_index = irq % num_cpus;

		for_each_cpu_and(numa_cpu, cpu_mask, &domain)
			if (cpu_online(numa_cpu) && i++ >= cpu_index)
				break;

		if (numa_cpu < nr_cpu_ids)
			return cpu_physical_id(numa_cpu);
	}
skip_numa_setup:
#endif
	/*
	 * Otherwise, round-robin interrupt vectors across all the
	 * processors.  (It'd be nice if we could be smarter in the
	 * case of NUMA.)
	 */
	do {
		if (++cpu >= nr_cpu_ids)
			cpu = 0;
	} while (!cpu_online(cpu) || !cpu_isset(cpu, domain));

	return cpu_physical_id(cpu);
#else  /* CONFIG_SMP */
	return cpu_physical_id(smp_processor_id());
#endif
}

static inline unsigned char choose_dmode(void)
{
#ifdef CONFIG_SMP
	if (smp_int_redirect & SMP_IRQ_REDIRECTION)
		return IOSAPIC_LOWEST_PRIORITY;
#endif
	return IOSAPIC_FIXED;
}

/*
 * ACPI can describe IOSAPIC interrupts via static tables and namespace
 * methods.  This provides an interface to register those interrupts and
 * program the IOSAPIC RTE.
 */
int
iosapic_register_intr (unsigned int gsi,
		       unsigned long polarity, unsigned long trigger)
{
	int irq, mask = 1, err;
	unsigned int dest;
	unsigned long flags;
	struct iosapic_rte_info *rte;
	u32 low32;
	unsigned char dmode;
	struct irq_desc *desc;

	/*
	 * If this GSI has already been registered (i.e., it's a
	 * shared interrupt, or we lost a race to register it),
	 * don't touch the RTE.
	 */
	spin_lock_irqsave(&iosapic_lock, flags);
	irq = __gsi_to_irq(gsi);
	if (irq > 0) {
		rte = find_rte(irq, gsi);
		if(iosapic_intr_info[irq].count == 0) {
			assign_irq_vector(irq);
			dynamic_irq_init(irq);
		} else if (rte->refcnt != NO_REF_RTE) {
			rte->refcnt++;
			goto unlock_iosapic_lock;
		}
	} else
		irq = create_irq();

	/* If vector is running out, we try to find a sharable vector */
	if (irq < 0) {
		irq = iosapic_find_sharable_irq(trigger, polarity);
		if (irq < 0)
			goto unlock_iosapic_lock;
	}

	desc = irq_to_desc(irq);
	raw_spin_lock(&desc->lock);
	dest = get_target_cpu(gsi, irq);
	dmode = choose_dmode();
	err = register_intr(gsi, irq, dmode, polarity, trigger);
	if (err < 0) {
		raw_spin_unlock(&desc->lock);
		irq = err;
		goto unlock_iosapic_lock;
	}

	/*
	 * If the vector is shared and already unmasked for other
	 * interrupt sources, don't mask it.
	 */
	low32 = iosapic_intr_info[irq].low32;
	if (irq_is_shared(irq) && !(low32 & IOSAPIC_MASK))
		mask = 0;
	set_rte(gsi, irq, dest, mask);

	printk(KERN_INFO "GSI %u (%s, %s) -> CPU %d (0x%04x) vector %d\n",
	       gsi, (trigger == IOSAPIC_EDGE ? "edge" : "level"),
	       (polarity == IOSAPIC_POL_HIGH ? "high" : "low"),
	       cpu_logical_id(dest), dest, irq_to_vector(irq));

	raw_spin_unlock(&desc->lock);
 unlock_iosapic_lock:
	spin_unlock_irqrestore(&iosapic_lock, flags);
	return irq;
}

void
iosapic_unregister_intr (unsigned int gsi)
{
	unsigned long flags;
	int irq, index;
	u32 low32;
	unsigned long trigger, polarity;
	unsigned int dest;
	struct iosapic_rte_info *rte;

	/*
	 * If the irq associated with the gsi is not found,
	 * iosapic_unregister_intr() is unbalanced. We need to check
	 * this again after getting locks.
	 */
	irq = gsi_to_irq(gsi);
	if (irq < 0) {
		printk(KERN_ERR "iosapic_unregister_intr(%u) unbalanced\n",
		       gsi);
		WARN_ON(1);
		return;
	}

	spin_lock_irqsave(&iosapic_lock, flags);
	if ((rte = find_rte(irq, gsi)) == NULL) {
		printk(KERN_ERR "iosapic_unregister_intr(%u) unbalanced\n",
		       gsi);
		WARN_ON(1);
		goto out;
	}

	if (--rte->refcnt > 0)
		goto out;

	rte->refcnt = NO_REF_RTE;

	/* Mask the interrupt */
	low32 = iosapic_intr_info[irq].low32 | IOSAPIC_MASK;
	iosapic_write(rte->iosapic, IOSAPIC_RTE_LOW(rte->rte_index), low32);

	iosapic_intr_info[irq].count--;
	index = find_iosapic(gsi);
	iosapic_lists[index].rtes_inuse--;
	WARN_ON(iosapic_lists[index].rtes_inuse < 0);

	trigger  = iosapic_intr_info[irq].trigger;
	polarity = iosapic_intr_info[irq].polarity;
	dest     = iosapic_intr_info[irq].dest;
	printk(KERN_INFO
	       "GSI %u (%s, %s) -> CPU %d (0x%04x) vector %d unregistered\n",
	       gsi, (trigger == IOSAPIC_EDGE ? "edge" : "level"),
	       (polarity == IOSAPIC_POL_HIGH ? "high" : "low"),
	       cpu_logical_id(dest), dest, irq_to_vector(irq));

	if (iosapic_intr_info[irq].count == 0) {
#ifdef CONFIG_SMP
		/* Clear affinity */
		cpumask_setall(irq_get_irq_data(irq)->affinity);
#endif
		/* Clear the interrupt information */
		iosapic_intr_info[irq].dest = 0;
		iosapic_intr_info[irq].dmode = 0;
		iosapic_intr_info[irq].polarity = 0;
		iosapic_intr_info[irq].trigger = 0;
		iosapic_intr_info[irq].low32 |= IOSAPIC_MASK;

		/* Destroy and reserve IRQ */
		destroy_and_reserve_irq(irq);
	}
 out:
	spin_unlock_irqrestore(&iosapic_lock, flags);
}

/*
 * ACPI calls this when it finds an entry for a platform interrupt.
 */
int __init
iosapic_register_platform_intr (u32 int_type, unsigned int gsi,
				int iosapic_vector, u16 eid, u16 id,
				unsigned long polarity, unsigned long trigger)
{
	static const char * const name[] = {"unknown", "PMI", "INIT", "CPEI"};
	unsigned char delivery;
	int irq, vector, mask = 0;
	unsigned int dest = ((id << 8) | eid) & 0xffff;

	switch (int_type) {
	      case ACPI_INTERRUPT_PMI:
		irq = vector = iosapic_vector;
		bind_irq_vector(irq, vector, CPU_MASK_ALL);
		/*
		 * since PMI vector is alloc'd by FW(ACPI) not by kernel,
		 * we need to make sure the vector is available
		 */
		iosapic_reassign_vector(irq);
		delivery = IOSAPIC_PMI;
		break;
	      case ACPI_INTERRUPT_INIT:
		irq = create_irq();
		if (irq < 0)
			panic("%s: out of interrupt vectors!\n", __func__);
		vector = irq_to_vector(irq);
		delivery = IOSAPIC_INIT;
		break;
	      case ACPI_INTERRUPT_CPEI:
		irq = vector = IA64_CPE_VECTOR;
		BUG_ON(bind_irq_vector(irq, vector, CPU_MASK_ALL));
		delivery = IOSAPIC_FIXED;
		mask = 1;
		break;
	      default:
		printk(KERN_ERR "%s: invalid int type 0x%x\n", __func__,
		       int_type);
		return -1;
	}

	register_intr(gsi, irq, delivery, polarity, trigger);

	printk(KERN_INFO
	       "PLATFORM int %s (0x%x): GSI %u (%s, %s) -> CPU %d (0x%04x)"
	       " vector %d\n",
	       int_type < ARRAY_SIZE(name) ? name[int_type] : "unknown",
	       int_type, gsi, (trigger == IOSAPIC_EDGE ? "edge" : "level"),
	       (polarity == IOSAPIC_POL_HIGH ? "high" : "low"),
	       cpu_logical_id(dest), dest, vector);

	set_rte(gsi, irq, dest, mask);
	return vector;
}

/*
 * ACPI calls this when it finds an entry for a legacy ISA IRQ override.
 */
void __devinit
iosapic_override_isa_irq (unsigned int isa_irq, unsigned int gsi,
			  unsigned long polarity,
			  unsigned long trigger)
{
	int vector, irq;
	unsigned int dest = cpu_physical_id(smp_processor_id());
	unsigned char dmode;

	irq = vector = isa_irq_to_vector(isa_irq);
	BUG_ON(bind_irq_vector(irq, vector, CPU_MASK_ALL));
	dmode = choose_dmode();
	register_intr(gsi, irq, dmode, polarity, trigger);

	DBG("ISA: IRQ %u -> GSI %u (%s,%s) -> CPU %d (0x%04x) vector %d\n",
	    isa_irq, gsi, trigger == IOSAPIC_EDGE ? "edge" : "level",
	    polarity == IOSAPIC_POL_HIGH ? "high" : "low",
	    cpu_logical_id(dest), dest, vector);

	set_rte(gsi, irq, dest, 1);
}

void __init
ia64_native_iosapic_pcat_compat_init(void)
{
	if (pcat_compat) {
		/*
		 * Disable the compatibility mode interrupts (8259 style),
		 * needs IN/OUT support enabled.
		 */
		printk(KERN_INFO
		       "%s: Disabling PC-AT compatible 8259 interrupts\n",
		       __func__);
		outb(0xff, 0xA1);
		outb(0xff, 0x21);
	}
}

void __init
iosapic_system_init (int system_pcat_compat)
{
	int irq;

	for (irq = 0; irq < NR_IRQS; ++irq) {
		iosapic_intr_info[irq].low32 = IOSAPIC_MASK;
		/* mark as unused */
		INIT_LIST_HEAD(&iosapic_intr_info[irq].rtes);

		iosapic_intr_info[irq].count = 0;
	}

	pcat_compat = system_pcat_compat;
	if (pcat_compat)
		iosapic_pcat_compat_init();
}

static inline int
iosapic_alloc (void)
{
	int index;

	for (index = 0; index < NR_IOSAPICS; index++)
		if (!iosapic_lists[index].addr)
			return index;

	printk(KERN_WARNING "%s: failed to allocate iosapic\n", __func__);
	return -1;
}

static inline void
iosapic_free (int index)
{
	memset(&iosapic_lists[index], 0, sizeof(iosapic_lists[0]));
}

static inline int
iosapic_check_gsi_range (unsigned int gsi_base, unsigned int ver)
{
	int index;
	unsigned int gsi_end, base, end;

	/* check gsi range */
	gsi_end = gsi_base + ((ver >> 16) & 0xff);
	for (index = 0; index < NR_IOSAPICS; index++) {
		if (!iosapic_lists[index].addr)
			continue;

		base = iosapic_lists[index].gsi_base;
		end  = base + iosapic_lists[index].num_rte - 1;

		if (gsi_end < base || end < gsi_base)
			continue; /* OK */

		return -EBUSY;
	}
	return 0;
}

int __devinit
iosapic_init (unsigned long phys_addr, unsigned int gsi_base)
{
	int num_rte, err, index;
	unsigned int isa_irq, ver;
	char __iomem *addr;
	unsigned long flags;

	spin_lock_irqsave(&iosapic_lock, flags);
	index = find_iosapic(gsi_base);
	if (index >= 0) {
		spin_unlock_irqrestore(&iosapic_lock, flags);
		return -EBUSY;
	}

	addr = ioremap(phys_addr, 0);
	if (addr == NULL) {
		spin_unlock_irqrestore(&iosapic_lock, flags);
		return -ENOMEM;
	}
	ver = iosapic_version(addr);
	if ((err = iosapic_check_gsi_range(gsi_base, ver))) {
		iounmap(addr);
		spin_unlock_irqrestore(&iosapic_lock, flags);
		return err;
	}

	/*
	 * The MAX_REDIR register holds the highest input pin number
	 * (starting from 0).  We add 1 so that we can use it for
	 * number of pins (= RTEs)
	 */
	num_rte = ((ver >> 16) & 0xff) + 1;

	index = iosapic_alloc();
	iosapic_lists[index].addr = addr;
	iosapic_lists[index].gsi_base = gsi_base;
	iosapic_lists[index].num_rte = num_rte;
#ifdef CONFIG_NUMA
	iosapic_lists[index].node = MAX_NUMNODES;
#endif
	spin_lock_init(&iosapic_lists[index].lock);
	spin_unlock_irqrestore(&iosapic_lock, flags);

	if ((gsi_base == 0) && pcat_compat) {
		/*
		 * Map the legacy ISA devices into the IOSAPIC data.  Some of
		 * these may get reprogrammed later on with data from the ACPI
		 * Interrupt Source Override table.
		 */
		for (isa_irq = 0; isa_irq < 16; ++isa_irq)
			iosapic_override_isa_irq(isa_irq, isa_irq,
						 IOSAPIC_POL_HIGH,
						 IOSAPIC_EDGE);
	}
	return 0;
}

#ifdef CONFIG_HOTPLUG
int
iosapic_remove (unsigned int gsi_base)
{
	int index, err = 0;
	unsigned long flags;

	spin_lock_irqsave(&iosapic_lock, flags);
	index = find_iosapic(gsi_base);
	if (index < 0) {
		printk(KERN_WARNING "%s: No IOSAPIC for GSI base %u\n",
		       __func__, gsi_base);
		goto out;
	}

	if (iosapic_lists[index].rtes_inuse) {
		err = -EBUSY;
		printk(KERN_WARNING "%s: IOSAPIC for GSI base %u is busy\n",
		       __func__, gsi_base);
		goto out;
	}

	iounmap(iosapic_lists[index].addr);
	iosapic_free(index);
 out:
	spin_unlock_irqrestore(&iosapic_lock, flags);
	return err;
}
#endif /* CONFIG_HOTPLUG */

#ifdef CONFIG_NUMA
void __devinit
map_iosapic_to_node(unsigned int gsi_base, int node)
{
	int index;

	index = find_iosapic(gsi_base);
	if (index < 0) {
		printk(KERN_WARNING "%s: No IOSAPIC for GSI %u\n",
		       __func__, gsi_base);
		return;
	}
	iosapic_lists[index].node = node;
	return;
}
#endif