summaryrefslogtreecommitdiffstats
path: root/drivers/iommu/amd_iommu_v2.c
blob: f7b875bb70d42138027f49ebde8150d27ce14cd2 (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
/*
 * Copyright (C) 2010-2012 Advanced Micro Devices, Inc.
 * Author: Joerg Roedel <jroedel@suse.de>
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License version 2 as published
 * by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
 */

#include <linux/mmu_notifier.h>
#include <linux/amd-iommu.h>
#include <linux/mm_types.h>
#include <linux/profile.h>
#include <linux/module.h>
#include <linux/sched.h>
#include <linux/iommu.h>
#include <linux/wait.h>
#include <linux/pci.h>
#include <linux/gfp.h>

#include "amd_iommu_types.h"
#include "amd_iommu_proto.h"

MODULE_LICENSE("GPL v2");
MODULE_AUTHOR("Joerg Roedel <jroedel@suse.de>");

#define MAX_DEVICES		0x10000
#define PRI_QUEUE_SIZE		512

struct pri_queue {
	atomic_t inflight;
	bool finish;
	int status;
};

struct pasid_state {
	struct list_head list;			/* For global state-list */
	atomic_t count;				/* Reference count */
	unsigned mmu_notifier_count;		/* Counting nested mmu_notifier
						   calls */
	struct mm_struct *mm;			/* mm_struct for the faults */
	struct mmu_notifier mn;                 /* mmu_notifier handle */
	struct pri_queue pri[PRI_QUEUE_SIZE];	/* PRI tag states */
	struct device_state *device_state;	/* Link to our device_state */
	int pasid;				/* PASID index */
	bool invalid;				/* Used during setup and
						   teardown of the pasid */
	spinlock_t lock;			/* Protect pri_queues and
						   mmu_notifer_count */
	wait_queue_head_t wq;			/* To wait for count == 0 */
};

struct device_state {
	struct list_head list;
	u16 devid;
	atomic_t count;
	struct pci_dev *pdev;
	struct pasid_state **states;
	struct iommu_domain *domain;
	int pasid_levels;
	int max_pasids;
	amd_iommu_invalid_ppr_cb inv_ppr_cb;
	amd_iommu_invalidate_ctx inv_ctx_cb;
	spinlock_t lock;
	wait_queue_head_t wq;
};

struct fault {
	struct work_struct work;
	struct device_state *dev_state;
	struct pasid_state *state;
	struct mm_struct *mm;
	u64 address;
	u16 devid;
	u16 pasid;
	u16 tag;
	u16 finish;
	u16 flags;
};

static LIST_HEAD(state_list);
static spinlock_t state_lock;

static struct workqueue_struct *iommu_wq;

static void free_pasid_states(struct device_state *dev_state);

static u16 device_id(struct pci_dev *pdev)
{
	u16 devid;

	devid = pdev->bus->number;
	devid = (devid << 8) | pdev->devfn;

	return devid;
}

static struct device_state *__get_device_state(u16 devid)
{
	struct device_state *dev_state;

	list_for_each_entry(dev_state, &state_list, list) {
		if (dev_state->devid == devid)
			return dev_state;
	}

	return NULL;
}

static struct device_state *get_device_state(u16 devid)
{
	struct device_state *dev_state;
	unsigned long flags;

	spin_lock_irqsave(&state_lock, flags);
	dev_state = __get_device_state(devid);
	if (dev_state != NULL)
		atomic_inc(&dev_state->count);
	spin_unlock_irqrestore(&state_lock, flags);

	return dev_state;
}

static void free_device_state(struct device_state *dev_state)
{
	struct iommu_group *group;

	/*
	 * First detach device from domain - No more PRI requests will arrive
	 * from that device after it is unbound from the IOMMUv2 domain.
	 */
	group = iommu_group_get(&dev_state->pdev->dev);
	if (WARN_ON(!group))
		return;

	iommu_detach_group(dev_state->domain, group);

	iommu_group_put(group);

	/* Everything is down now, free the IOMMUv2 domain */
	iommu_domain_free(dev_state->domain);

	/* Finally get rid of the device-state */
	kfree(dev_state);
}

static void put_device_state(struct device_state *dev_state)
{
	if (atomic_dec_and_test(&dev_state->count))
		wake_up(&dev_state->wq);
}

/* Must be called under dev_state->lock */
static struct pasid_state **__get_pasid_state_ptr(struct device_state *dev_state,
						  int pasid, bool alloc)
{
	struct pasid_state **root, **ptr;
	int level, index;

	level = dev_state->pasid_levels;
	root  = dev_state->states;

	while (true) {

		index = (pasid >> (9 * level)) & 0x1ff;
		ptr   = &root[index];

		if (level == 0)
			break;

		if (*ptr == NULL) {
			if (!alloc)
				return NULL;

			*ptr = (void *)get_zeroed_page(GFP_ATOMIC);
			if (*ptr == NULL)
				return NULL;
		}

		root   = (struct pasid_state **)*ptr;
		level -= 1;
	}

	return ptr;
}

static int set_pasid_state(struct device_state *dev_state,
			   struct pasid_state *pasid_state,
			   int pasid)
{
	struct pasid_state **ptr;
	unsigned long flags;
	int ret;

	spin_lock_irqsave(&dev_state->lock, flags);
	ptr = __get_pasid_state_ptr(dev_state, pasid, true);

	ret = -ENOMEM;
	if (ptr == NULL)
		goto out_unlock;

	ret = -ENOMEM;
	if (*ptr != NULL)
		goto out_unlock;

	*ptr = pasid_state;

	ret = 0;

out_unlock:
	spin_unlock_irqrestore(&dev_state->lock, flags);

	return ret;
}

static void clear_pasid_state(struct device_state *dev_state, int pasid)
{
	struct pasid_state **ptr;
	unsigned long flags;

	spin_lock_irqsave(&dev_state->lock, flags);
	ptr = __get_pasid_state_ptr(dev_state, pasid, true);

	if (ptr == NULL)
		goto out_unlock;

	*ptr = NULL;

out_unlock:
	spin_unlock_irqrestore(&dev_state->lock, flags);
}

static struct pasid_state *get_pasid_state(struct device_state *dev_state,
					   int pasid)
{
	struct pasid_state **ptr, *ret = NULL;
	unsigned long flags;

	spin_lock_irqsave(&dev_state->lock, flags);
	ptr = __get_pasid_state_ptr(dev_state, pasid, false);

	if (ptr == NULL)
		goto out_unlock;

	ret = *ptr;
	if (ret)
		atomic_inc(&ret->count);

out_unlock:
	spin_unlock_irqrestore(&dev_state->lock, flags);

	return ret;
}

static void free_pasid_state(struct pasid_state *pasid_state)
{
	kfree(pasid_state);
}

static void put_pasid_state(struct pasid_state *pasid_state)
{
	if (atomic_dec_and_test(&pasid_state->count))
		wake_up(&pasid_state->wq);
}

static void put_pasid_state_wait(struct pasid_state *pasid_state)
{
	atomic_dec(&pasid_state->count);
	wait_event(pasid_state->wq, !atomic_read(&pasid_state->count));
	free_pasid_state(pasid_state);
}

static void unbind_pasid(struct pasid_state *pasid_state)
{
	struct iommu_domain *domain;

	domain = pasid_state->device_state->domain;

	/*
	 * Mark pasid_state as invalid, no more faults will we added to the
	 * work queue after this is visible everywhere.
	 */
	pasid_state->invalid = true;

	/* Make sure this is visible */
	smp_wmb();

	/* After this the device/pasid can't access the mm anymore */
	amd_iommu_domain_clear_gcr3(domain, pasid_state->pasid);

	/* Make sure no more pending faults are in the queue */
	flush_workqueue(iommu_wq);
}

static void free_pasid_states_level1(struct pasid_state **tbl)
{
	int i;

	for (i = 0; i < 512; ++i) {
		if (tbl[i] == NULL)
			continue;

		free_page((unsigned long)tbl[i]);
	}
}

static void free_pasid_states_level2(struct pasid_state **tbl)
{
	struct pasid_state **ptr;
	int i;

	for (i = 0; i < 512; ++i) {
		if (tbl[i] == NULL)
			continue;

		ptr = (struct pasid_state **)tbl[i];
		free_pasid_states_level1(ptr);
	}
}

static void free_pasid_states(struct device_state *dev_state)
{
	struct pasid_state *pasid_state;
	int i;

	for (i = 0; i < dev_state->max_pasids; ++i) {
		pasid_state = get_pasid_state(dev_state, i);
		if (pasid_state == NULL)
			continue;

		put_pasid_state(pasid_state);

		/*
		 * This will call the mn_release function and
		 * unbind the PASID
		 */
		mmu_notifier_unregister(&pasid_state->mn, pasid_state->mm);

		put_pasid_state_wait(pasid_state); /* Reference taken in
						      amd_iommu_bind_pasid */

		/* Drop reference taken in amd_iommu_bind_pasid */
		put_device_state(dev_state);
	}

	if (dev_state->pasid_levels == 2)
		free_pasid_states_level2(dev_state->states);
	else if (dev_state->pasid_levels == 1)
		free_pasid_states_level1(dev_state->states);
	else if (dev_state->pasid_levels != 0)
		BUG();

	free_page((unsigned long)dev_state->states);
}

static struct pasid_state *mn_to_state(struct mmu_notifier *mn)
{
	return container_of(mn, struct pasid_state, mn);
}

static void __mn_flush_page(struct mmu_notifier *mn,
			    unsigned long address)
{
	struct pasid_state *pasid_state;
	struct device_state *dev_state;

	pasid_state = mn_to_state(mn);
	dev_state   = pasid_state->device_state;

	amd_iommu_flush_page(dev_state->domain, pasid_state->pasid, address);
}

static int mn_clear_flush_young(struct mmu_notifier *mn,
				struct mm_struct *mm,
				unsigned long start,
				unsigned long end)
{
	for (; start < end; start += PAGE_SIZE)
		__mn_flush_page(mn, start);

	return 0;
}

static void mn_invalidate_page(struct mmu_notifier *mn,
			       struct mm_struct *mm,
			       unsigned long address)
{
	__mn_flush_page(mn, address);
}

static void mn_invalidate_range(struct mmu_notifier *mn,
				struct mm_struct *mm,
				unsigned long start, unsigned long end)
{
	struct pasid_state *pasid_state;
	struct device_state *dev_state;

	pasid_state = mn_to_state(mn);
	dev_state   = pasid_state->device_state;

	if ((start ^ (end - 1)) < PAGE_SIZE)
		amd_iommu_flush_page(dev_state->domain, pasid_state->pasid,
				     start);
	else
		amd_iommu_flush_tlb(dev_state->domain, pasid_state->pasid);
}

static void mn_release(struct mmu_notifier *mn, struct mm_struct *mm)
{
	struct pasid_state *pasid_state;
	struct device_state *dev_state;
	bool run_inv_ctx_cb;

	might_sleep();

	pasid_state    = mn_to_state(mn);
	dev_state      = pasid_state->device_state;
	run_inv_ctx_cb = !pasid_state->invalid;

	if (run_inv_ctx_cb && dev_state->inv_ctx_cb)
		dev_state->inv_ctx_cb(dev_state->pdev, pasid_state->pasid);

	unbind_pasid(pasid_state);
}

static struct mmu_notifier_ops iommu_mn = {
	.release		= mn_release,
	.clear_flush_young      = mn_clear_flush_young,
	.invalidate_page        = mn_invalidate_page,
	.invalidate_range       = mn_invalidate_range,
};

static void set_pri_tag_status(struct pasid_state *pasid_state,
			       u16 tag, int status)
{
	unsigned long flags;

	spin_lock_irqsave(&pasid_state->lock, flags);
	pasid_state->pri[tag].status = status;
	spin_unlock_irqrestore(&pasid_state->lock, flags);
}

static void finish_pri_tag(struct device_state *dev_state,
			   struct pasid_state *pasid_state,
			   u16 tag)
{
	unsigned long flags;

	spin_lock_irqsave(&pasid_state->lock, flags);
	if (atomic_dec_and_test(&pasid_state->pri[tag].inflight) &&
	    pasid_state->pri[tag].finish) {
		amd_iommu_complete_ppr(dev_state->pdev, pasid_state->pasid,
				       pasid_state->pri[tag].status, tag);
		pasid_state->pri[tag].finish = false;
		pasid_state->pri[tag].status = PPR_SUCCESS;
	}
	spin_unlock_irqrestore(&pasid_state->lock, flags);
}

static void handle_fault_error(struct fault *fault)
{
	int status;

	if (!fault->dev_state->inv_ppr_cb) {
		set_pri_tag_status(fault->state, fault->tag, PPR_INVALID);
		return;
	}

	status = fault->dev_state->inv_ppr_cb(fault->dev_state->pdev,
					      fault->pasid,
					      fault->address,
					      fault->flags);
	switch (status) {
	case AMD_IOMMU_INV_PRI_RSP_SUCCESS:
		set_pri_tag_status(fault->state, fault->tag, PPR_SUCCESS);
		break;
	case AMD_IOMMU_INV_PRI_RSP_INVALID:
		set_pri_tag_status(fault->state, fault->tag, PPR_INVALID);
		break;
	case AMD_IOMMU_INV_PRI_RSP_FAIL:
		set_pri_tag_status(fault->state, fault->tag, PPR_FAILURE);
		break;
	default:
		BUG();
	}
}

static void do_fault(struct work_struct *work)
{
	struct fault *fault = container_of(work, struct fault, work);
	struct mm_struct *mm;
	struct vm_area_struct *vma;
	u64 address;
	int ret, write;

	write = !!(fault->flags & PPR_FAULT_WRITE);

	mm = fault->state->mm;
	address = fault->address;

	down_read(&mm->mmap_sem);
	vma = find_extend_vma(mm, address);
	if (!vma || address < vma->vm_start) {
		/* failed to get a vma in the right range */
		up_read(&mm->mmap_sem);
		handle_fault_error(fault);
		goto out;
	}

	ret = handle_mm_fault(mm, vma, address, write);
	if (ret & VM_FAULT_ERROR) {
		/* failed to service fault */
		up_read(&mm->mmap_sem);
		handle_fault_error(fault);
		goto out;
	}

	up_read(&mm->mmap_sem);

out:
	finish_pri_tag(fault->dev_state, fault->state, fault->tag);

	put_pasid_state(fault->state);

	kfree(fault);
}

static int ppr_notifier(struct notifier_block *nb, unsigned long e, void *data)
{
	struct amd_iommu_fault *iommu_fault;
	struct pasid_state *pasid_state;
	struct device_state *dev_state;
	unsigned long flags;
	struct fault *fault;
	bool finish;
	u16 tag;
	int ret;

	iommu_fault = data;
	tag         = iommu_fault->tag & 0x1ff;
	finish      = (iommu_fault->tag >> 9) & 1;

	ret = NOTIFY_DONE;
	dev_state = get_device_state(iommu_fault->device_id);
	if (dev_state == NULL)
		goto out;

	pasid_state = get_pasid_state(dev_state, iommu_fault->pasid);
	if (pasid_state == NULL || pasid_state->invalid) {
		/* We know the device but not the PASID -> send INVALID */
		amd_iommu_complete_ppr(dev_state->pdev, iommu_fault->pasid,
				       PPR_INVALID, tag);
		goto out_drop_state;
	}

	spin_lock_irqsave(&pasid_state->lock, flags);
	atomic_inc(&pasid_state->pri[tag].inflight);
	if (finish)
		pasid_state->pri[tag].finish = true;
	spin_unlock_irqrestore(&pasid_state->lock, flags);

	fault = kzalloc(sizeof(*fault), GFP_ATOMIC);
	if (fault == NULL) {
		/* We are OOM - send success and let the device re-fault */
		finish_pri_tag(dev_state, pasid_state, tag);
		goto out_drop_state;
	}

	fault->dev_state = dev_state;
	fault->address   = iommu_fault->address;
	fault->state     = pasid_state;
	fault->tag       = tag;
	fault->finish    = finish;
	fault->pasid     = iommu_fault->pasid;
	fault->flags     = iommu_fault->flags;
	INIT_WORK(&fault->work, do_fault);

	queue_work(iommu_wq, &fault->work);

	ret = NOTIFY_OK;

out_drop_state:

	if (ret != NOTIFY_OK && pasid_state)
		put_pasid_state(pasid_state);

	put_device_state(dev_state);

out:
	return ret;
}

static struct notifier_block ppr_nb = {
	.notifier_call = ppr_notifier,
};

int amd_iommu_bind_pasid(struct pci_dev *pdev, int pasid,
			 struct task_struct *task)
{
	struct pasid_state *pasid_state;
	struct device_state *dev_state;
	struct mm_struct *mm;
	u16 devid;
	int ret;

	might_sleep();

	if (!amd_iommu_v2_supported())
		return -ENODEV;

	devid     = device_id(pdev);
	dev_state = get_device_state(devid);

	if (dev_state == NULL)
		return -EINVAL;

	ret = -EINVAL;
	if (pasid < 0 || pasid >= dev_state->max_pasids)
		goto out;

	ret = -ENOMEM;
	pasid_state = kzalloc(sizeof(*pasid_state), GFP_KERNEL);
	if (pasid_state == NULL)
		goto out;


	atomic_set(&pasid_state->count, 1);
	init_waitqueue_head(&pasid_state->wq);
	spin_lock_init(&pasid_state->lock);

	mm                        = get_task_mm(task);
	pasid_state->mm           = mm;
	pasid_state->device_state = dev_state;
	pasid_state->pasid        = pasid;
	pasid_state->invalid      = true; /* Mark as valid only if we are
					     done with setting up the pasid */
	pasid_state->mn.ops       = &iommu_mn;

	if (pasid_state->mm == NULL)
		goto out_free;

	mmu_notifier_register(&pasid_state->mn, mm);

	ret = set_pasid_state(dev_state, pasid_state, pasid);
	if (ret)
		goto out_unregister;

	ret = amd_iommu_domain_set_gcr3(dev_state->domain, pasid,
					__pa(pasid_state->mm->pgd));
	if (ret)
		goto out_clear_state;

	/* Now we are ready to handle faults */
	pasid_state->invalid = false;

	/*
	 * Drop the reference to the mm_struct here. We rely on the
	 * mmu_notifier release call-back to inform us when the mm
	 * is going away.
	 */
	mmput(mm);

	return 0;

out_clear_state:
	clear_pasid_state(dev_state, pasid);

out_unregister:
	mmu_notifier_unregister(&pasid_state->mn, mm);

out_free:
	mmput(mm);
	free_pasid_state(pasid_state);

out:
	put_device_state(dev_state);

	return ret;
}
EXPORT_SYMBOL(amd_iommu_bind_pasid);

void amd_iommu_unbind_pasid(struct pci_dev *pdev, int pasid)
{
	struct pasid_state *pasid_state;
	struct device_state *dev_state;
	u16 devid;

	might_sleep();

	if (!amd_iommu_v2_supported())
		return;

	devid = device_id(pdev);
	dev_state = get_device_state(devid);
	if (dev_state == NULL)
		return;

	if (pasid < 0 || pasid >= dev_state->max_pasids)
		goto out;

	pasid_state = get_pasid_state(dev_state, pasid);
	if (pasid_state == NULL)
		goto out;
	/*
	 * Drop reference taken here. We are safe because we still hold
	 * the reference taken in the amd_iommu_bind_pasid function.
	 */
	put_pasid_state(pasid_state);

	/* Clear the pasid state so that the pasid can be re-used */
	clear_pasid_state(dev_state, pasid_state->pasid);

	/*
	 * Call mmu_notifier_unregister to drop our reference
	 * to pasid_state->mm
	 */
	mmu_notifier_unregister(&pasid_state->mn, pasid_state->mm);

	put_pasid_state_wait(pasid_state); /* Reference taken in
					      amd_iommu_bind_pasid */
out:
	/* Drop reference taken in this function */
	put_device_state(dev_state);

	/* Drop reference taken in amd_iommu_bind_pasid */
	put_device_state(dev_state);
}
EXPORT_SYMBOL(amd_iommu_unbind_pasid);

int amd_iommu_init_device(struct pci_dev *pdev, int pasids)
{
	struct device_state *dev_state;
	struct iommu_group *group;
	unsigned long flags;
	int ret, tmp;
	u16 devid;

	might_sleep();

	if (!amd_iommu_v2_supported())
		return -ENODEV;

	if (pasids <= 0 || pasids > (PASID_MASK + 1))
		return -EINVAL;

	devid = device_id(pdev);

	dev_state = kzalloc(sizeof(*dev_state), GFP_KERNEL);
	if (dev_state == NULL)
		return -ENOMEM;

	spin_lock_init(&dev_state->lock);
	init_waitqueue_head(&dev_state->wq);
	dev_state->pdev  = pdev;
	dev_state->devid = devid;

	tmp = pasids;
	for (dev_state->pasid_levels = 0; (tmp - 1) & ~0x1ff; tmp >>= 9)
		dev_state->pasid_levels += 1;

	atomic_set(&dev_state->count, 1);
	dev_state->max_pasids = pasids;

	ret = -ENOMEM;
	dev_state->states = (void *)get_zeroed_page(GFP_KERNEL);
	if (dev_state->states == NULL)
		goto out_free_dev_state;

	dev_state->domain = iommu_domain_alloc(&pci_bus_type);
	if (dev_state->domain == NULL)
		goto out_free_states;

	amd_iommu_domain_direct_map(dev_state->domain);

	ret = amd_iommu_domain_enable_v2(dev_state->domain, pasids);
	if (ret)
		goto out_free_domain;

	group = iommu_group_get(&pdev->dev);
	if (!group)
		goto out_free_domain;

	ret = iommu_attach_group(dev_state->domain, group);
	if (ret != 0)
		goto out_drop_group;

	iommu_group_put(group);

	spin_lock_irqsave(&state_lock, flags);

	if (__get_device_state(devid) != NULL) {
		spin_unlock_irqrestore(&state_lock, flags);
		ret = -EBUSY;
		goto out_free_domain;
	}

	list_add_tail(&dev_state->list, &state_list);

	spin_unlock_irqrestore(&state_lock, flags);

	return 0;

out_drop_group:
	iommu_group_put(group);

out_free_domain:
	iommu_domain_free(dev_state->domain);

out_free_states:
	free_page((unsigned long)dev_state->states);

out_free_dev_state:
	kfree(dev_state);

	return ret;
}
EXPORT_SYMBOL(amd_iommu_init_device);

void amd_iommu_free_device(struct pci_dev *pdev)
{
	struct device_state *dev_state;
	unsigned long flags;
	u16 devid;

	if (!amd_iommu_v2_supported())
		return;

	devid = device_id(pdev);

	spin_lock_irqsave(&state_lock, flags);

	dev_state = __get_device_state(devid);
	if (dev_state == NULL) {
		spin_unlock_irqrestore(&state_lock, flags);
		return;
	}

	list_del(&dev_state->list);

	spin_unlock_irqrestore(&state_lock, flags);

	/* Get rid of any remaining pasid states */
	free_pasid_states(dev_state);

	put_device_state(dev_state);
	/*
	 * Wait until the last reference is dropped before freeing
	 * the device state.
	 */
	wait_event(dev_state->wq, !atomic_read(&dev_state->count));
	free_device_state(dev_state);
}
EXPORT_SYMBOL(amd_iommu_free_device);

int amd_iommu_set_invalid_ppr_cb(struct pci_dev *pdev,
				 amd_iommu_invalid_ppr_cb cb)
{
	struct device_state *dev_state;
	unsigned long flags;
	u16 devid;
	int ret;

	if (!amd_iommu_v2_supported())
		return -ENODEV;

	devid = device_id(pdev);

	spin_lock_irqsave(&state_lock, flags);

	ret = -EINVAL;
	dev_state = __get_device_state(devid);
	if (dev_state == NULL)
		goto out_unlock;

	dev_state->inv_ppr_cb = cb;

	ret = 0;

out_unlock:
	spin_unlock_irqrestore(&state_lock, flags);

	return ret;
}
EXPORT_SYMBOL(amd_iommu_set_invalid_ppr_cb);

int amd_iommu_set_invalidate_ctx_cb(struct pci_dev *pdev,
				    amd_iommu_invalidate_ctx cb)
{
	struct device_state *dev_state;
	unsigned long flags;
	u16 devid;
	int ret;

	if (!amd_iommu_v2_supported())
		return -ENODEV;

	devid = device_id(pdev);

	spin_lock_irqsave(&state_lock, flags);

	ret = -EINVAL;
	dev_state = __get_device_state(devid);
	if (dev_state == NULL)
		goto out_unlock;

	dev_state->inv_ctx_cb = cb;

	ret = 0;

out_unlock:
	spin_unlock_irqrestore(&state_lock, flags);

	return ret;
}
EXPORT_SYMBOL(amd_iommu_set_invalidate_ctx_cb);

static int __init amd_iommu_v2_init(void)
{
	int ret;

	pr_info("AMD IOMMUv2 driver by Joerg Roedel <jroedel@suse.de>\n");

	if (!amd_iommu_v2_supported()) {
		pr_info("AMD IOMMUv2 functionality not available on this system\n");
		/*
		 * Load anyway to provide the symbols to other modules
		 * which may use AMD IOMMUv2 optionally.
		 */
		return 0;
	}

	spin_lock_init(&state_lock);

	ret = -ENOMEM;
	iommu_wq = create_workqueue("amd_iommu_v2");
	if (iommu_wq == NULL)
		goto out;

	amd_iommu_register_ppr_notifier(&ppr_nb);

	return 0;

out:
	return ret;
}

static void __exit amd_iommu_v2_exit(void)
{
	struct device_state *dev_state;
	int i;

	if (!amd_iommu_v2_supported())
		return;

	amd_iommu_unregister_ppr_notifier(&ppr_nb);

	flush_workqueue(iommu_wq);

	/*
	 * The loop below might call flush_workqueue(), so call
	 * destroy_workqueue() after it
	 */
	for (i = 0; i < MAX_DEVICES; ++i) {
		dev_state = get_device_state(i);

		if (dev_state == NULL)
			continue;

		WARN_ON_ONCE(1);

		put_device_state(dev_state);
		amd_iommu_free_device(dev_state->pdev);
	}

	destroy_workqueue(iommu_wq);
}

module_init(amd_iommu_v2_init);
module_exit(amd_iommu_v2_exit);