summaryrefslogtreecommitdiff
path: root/drivers/dma/ioat/dma.c
blob: 5428746f03fbb90504369ac81cdd8374a90651eb (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
/*
 * Intel I/OAT DMA Linux driver
 * Copyright(c) 2004 - 2015 Intel Corporation.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms and conditions 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.
 *
 * The full GNU General Public License is included in this distribution in
 * the file called "COPYING".
 *
 */

/*
 * This driver supports an Intel I/OAT DMA engine, which does asynchronous
 * copy operations.
 */

#include <linux/init.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/pci.h>
#include <linux/interrupt.h>
#include <linux/dmaengine.h>
#include <linux/delay.h>
#include <linux/dma-mapping.h>
#include <linux/workqueue.h>
#include <linux/prefetch.h>
#include <linux/sizes.h>
#include "dma.h"
#include "registers.h"
#include "hw.h"

#include "../dmaengine.h"

static void ioat_eh(struct ioatdma_chan *ioat_chan);

/**
 * ioat_dma_do_interrupt - handler used for single vector interrupt mode
 * @irq: interrupt id
 * @data: interrupt data
 */
irqreturn_t ioat_dma_do_interrupt(int irq, void *data)
{
	struct ioatdma_device *instance = data;
	struct ioatdma_chan *ioat_chan;
	unsigned long attnstatus;
	int bit;
	u8 intrctrl;

	intrctrl = readb(instance->reg_base + IOAT_INTRCTRL_OFFSET);

	if (!(intrctrl & IOAT_INTRCTRL_MASTER_INT_EN))
		return IRQ_NONE;

	if (!(intrctrl & IOAT_INTRCTRL_INT_STATUS)) {
		writeb(intrctrl, instance->reg_base + IOAT_INTRCTRL_OFFSET);
		return IRQ_NONE;
	}

	attnstatus = readl(instance->reg_base + IOAT_ATTNSTATUS_OFFSET);
	for_each_set_bit(bit, &attnstatus, BITS_PER_LONG) {
		ioat_chan = ioat_chan_by_index(instance, bit);
		if (test_bit(IOAT_RUN, &ioat_chan->state))
			tasklet_schedule(&ioat_chan->cleanup_task);
	}

	writeb(intrctrl, instance->reg_base + IOAT_INTRCTRL_OFFSET);
	return IRQ_HANDLED;
}

/**
 * ioat_dma_do_interrupt_msix - handler used for vector-per-channel interrupt mode
 * @irq: interrupt id
 * @data: interrupt data
 */
irqreturn_t ioat_dma_do_interrupt_msix(int irq, void *data)
{
	struct ioatdma_chan *ioat_chan = data;

	if (test_bit(IOAT_RUN, &ioat_chan->state))
		tasklet_schedule(&ioat_chan->cleanup_task);

	return IRQ_HANDLED;
}

void ioat_stop(struct ioatdma_chan *ioat_chan)
{
	struct ioatdma_device *ioat_dma = ioat_chan->ioat_dma;
	struct pci_dev *pdev = ioat_dma->pdev;
	int chan_id = chan_num(ioat_chan);
	struct msix_entry *msix;

	/* 1/ stop irq from firing tasklets
	 * 2/ stop the tasklet from re-arming irqs
	 */
	clear_bit(IOAT_RUN, &ioat_chan->state);

	/* flush inflight interrupts */
	switch (ioat_dma->irq_mode) {
	case IOAT_MSIX:
		msix = &ioat_dma->msix_entries[chan_id];
		synchronize_irq(msix->vector);
		break;
	case IOAT_MSI:
	case IOAT_INTX:
		synchronize_irq(pdev->irq);
		break;
	default:
		break;
	}

	/* flush inflight timers */
	del_timer_sync(&ioat_chan->timer);

	/* flush inflight tasklet runs */
	tasklet_kill(&ioat_chan->cleanup_task);

	/* final cleanup now that everything is quiesced and can't re-arm */
	ioat_cleanup_event((unsigned long)&ioat_chan->dma_chan);
}

static void __ioat_issue_pending(struct ioatdma_chan *ioat_chan)
{
	ioat_chan->dmacount += ioat_ring_pending(ioat_chan);
	ioat_chan->issued = ioat_chan->head;
	writew(ioat_chan->dmacount,
	       ioat_chan->reg_base + IOAT_CHAN_DMACOUNT_OFFSET);
	dev_dbg(to_dev(ioat_chan),
		"%s: head: %#x tail: %#x issued: %#x count: %#x\n",
		__func__, ioat_chan->head, ioat_chan->tail,
		ioat_chan->issued, ioat_chan->dmacount);
}

void ioat_issue_pending(struct dma_chan *c)
{
	struct ioatdma_chan *ioat_chan = to_ioat_chan(c);

	if (ioat_ring_pending(ioat_chan)) {
		spin_lock_bh(&ioat_chan->prep_lock);
		__ioat_issue_pending(ioat_chan);
		spin_unlock_bh(&ioat_chan->prep_lock);
	}
}

/**
 * ioat_update_pending - log pending descriptors
 * @ioat: ioat+ channel
 *
 * Check if the number of unsubmitted descriptors has exceeded the
 * watermark.  Called with prep_lock held
 */
static void ioat_update_pending(struct ioatdma_chan *ioat_chan)
{
	if (ioat_ring_pending(ioat_chan) > ioat_pending_level)
		__ioat_issue_pending(ioat_chan);
}

static void __ioat_start_null_desc(struct ioatdma_chan *ioat_chan)
{
	struct ioat_ring_ent *desc;
	struct ioat_dma_descriptor *hw;

	if (ioat_ring_space(ioat_chan) < 1) {
		dev_err(to_dev(ioat_chan),
			"Unable to start null desc - ring full\n");
		return;
	}

	dev_dbg(to_dev(ioat_chan),
		"%s: head: %#x tail: %#x issued: %#x\n",
		__func__, ioat_chan->head, ioat_chan->tail, ioat_chan->issued);
	desc = ioat_get_ring_ent(ioat_chan, ioat_chan->head);

	hw = desc->hw;
	hw->ctl = 0;
	hw->ctl_f.null = 1;
	hw->ctl_f.int_en = 1;
	hw->ctl_f.compl_write = 1;
	/* set size to non-zero value (channel returns error when size is 0) */
	hw->size = NULL_DESC_BUFFER_SIZE;
	hw->src_addr = 0;
	hw->dst_addr = 0;
	async_tx_ack(&desc->txd);
	ioat_set_chainaddr(ioat_chan, desc->txd.phys);
	dump_desc_dbg(ioat_chan, desc);
	/* make sure descriptors are written before we submit */
	wmb();
	ioat_chan->head += 1;
	__ioat_issue_pending(ioat_chan);
}

void ioat_start_null_desc(struct ioatdma_chan *ioat_chan)
{
	spin_lock_bh(&ioat_chan->prep_lock);
	if (!test_bit(IOAT_CHAN_DOWN, &ioat_chan->state))
		__ioat_start_null_desc(ioat_chan);
	spin_unlock_bh(&ioat_chan->prep_lock);
}

static void __ioat_restart_chan(struct ioatdma_chan *ioat_chan)
{
	/* set the tail to be re-issued */
	ioat_chan->issued = ioat_chan->tail;
	ioat_chan->dmacount = 0;
	mod_timer(&ioat_chan->timer, jiffies + COMPLETION_TIMEOUT);

	dev_dbg(to_dev(ioat_chan),
		"%s: head: %#x tail: %#x issued: %#x count: %#x\n",
		__func__, ioat_chan->head, ioat_chan->tail,
		ioat_chan->issued, ioat_chan->dmacount);

	if (ioat_ring_pending(ioat_chan)) {
		struct ioat_ring_ent *desc;

		desc = ioat_get_ring_ent(ioat_chan, ioat_chan->tail);
		ioat_set_chainaddr(ioat_chan, desc->txd.phys);
		__ioat_issue_pending(ioat_chan);
	} else
		__ioat_start_null_desc(ioat_chan);
}

static int ioat_quiesce(struct ioatdma_chan *ioat_chan, unsigned long tmo)
{
	unsigned long end = jiffies + tmo;
	int err = 0;
	u32 status;

	status = ioat_chansts(ioat_chan);
	if (is_ioat_active(status) || is_ioat_idle(status))
		ioat_suspend(ioat_chan);
	while (is_ioat_active(status) || is_ioat_idle(status)) {
		if (tmo && time_after(jiffies, end)) {
			err = -ETIMEDOUT;
			break;
		}
		status = ioat_chansts(ioat_chan);
		cpu_relax();
	}

	return err;
}

static int ioat_reset_sync(struct ioatdma_chan *ioat_chan, unsigned long tmo)
{
	unsigned long end = jiffies + tmo;
	int err = 0;

	ioat_reset(ioat_chan);
	while (ioat_reset_pending(ioat_chan)) {
		if (end && time_after(jiffies, end)) {
			err = -ETIMEDOUT;
			break;
		}
		cpu_relax();
	}

	return err;
}

static dma_cookie_t ioat_tx_submit_unlock(struct dma_async_tx_descriptor *tx)
	__releases(&ioat_chan->prep_lock)
{
	struct dma_chan *c = tx->chan;
	struct ioatdma_chan *ioat_chan = to_ioat_chan(c);
	dma_cookie_t cookie;

	cookie = dma_cookie_assign(tx);
	dev_dbg(to_dev(ioat_chan), "%s: cookie: %d\n", __func__, cookie);

	if (!test_and_set_bit(IOAT_CHAN_ACTIVE, &ioat_chan->state))
		mod_timer(&ioat_chan->timer, jiffies + COMPLETION_TIMEOUT);

	/* make descriptor updates visible before advancing ioat->head,
	 * this is purposefully not smp_wmb() since we are also
	 * publishing the descriptor updates to a dma device
	 */
	wmb();

	ioat_chan->head += ioat_chan->produce;

	ioat_update_pending(ioat_chan);
	spin_unlock_bh(&ioat_chan->prep_lock);

	return cookie;
}

static struct ioat_ring_ent *
ioat_alloc_ring_ent(struct dma_chan *chan, int idx, gfp_t flags)
{
	struct ioat_dma_descriptor *hw;
	struct ioat_ring_ent *desc;
	struct ioatdma_device *ioat_dma;
	struct ioatdma_chan *ioat_chan = to_ioat_chan(chan);
	int chunk;
	dma_addr_t phys;
	u8 *pos;
	off_t offs;

	ioat_dma = to_ioatdma_device(chan->device);

	chunk = idx / IOAT_DESCS_PER_2M;
	idx &= (IOAT_DESCS_PER_2M - 1);
	offs = idx * IOAT_DESC_SZ;
	pos = (u8 *)ioat_chan->descs[chunk].virt + offs;
	phys = ioat_chan->descs[chunk].hw + offs;
	hw = (struct ioat_dma_descriptor *)pos;
	memset(hw, 0, sizeof(*hw));

	desc = kmem_cache_zalloc(ioat_cache, flags);
	if (!desc)
		return NULL;

	dma_async_tx_descriptor_init(&desc->txd, chan);
	desc->txd.tx_submit = ioat_tx_submit_unlock;
	desc->hw = hw;
	desc->txd.phys = phys;
	return desc;
}

void ioat_free_ring_ent(struct ioat_ring_ent *desc, struct dma_chan *chan)
{
	kmem_cache_free(ioat_cache, desc);
}

struct ioat_ring_ent **
ioat_alloc_ring(struct dma_chan *c, int order, gfp_t flags)
{
	struct ioatdma_chan *ioat_chan = to_ioat_chan(c);
	struct ioat_ring_ent **ring;
	int total_descs = 1 << order;
	int i, chunks;

	/* allocate the array to hold the software ring */
	ring = kcalloc(total_descs, sizeof(*ring), flags);
	if (!ring)
		return NULL;

	ioat_chan->desc_chunks = chunks = (total_descs * IOAT_DESC_SZ) / SZ_2M;

	for (i = 0; i < chunks; i++) {
		struct ioat_descs *descs = &ioat_chan->descs[i];

		descs->virt = dma_alloc_coherent(to_dev(ioat_chan),
						 SZ_2M, &descs->hw, flags);
		if (!descs->virt && (i > 0)) {
			int idx;

			for (idx = 0; idx < i; idx++) {
				dma_free_coherent(to_dev(ioat_chan), SZ_2M,
						  descs->virt, descs->hw);
				descs->virt = NULL;
				descs->hw = 0;
			}

			ioat_chan->desc_chunks = 0;
			kfree(ring);
			return NULL;
		}
	}

	for (i = 0; i < total_descs; i++) {
		ring[i] = ioat_alloc_ring_ent(c, i, flags);
		if (!ring[i]) {
			int idx;

			while (i--)
				ioat_free_ring_ent(ring[i], c);

			for (idx = 0; idx < ioat_chan->desc_chunks; idx++) {
				dma_free_coherent(to_dev(ioat_chan),
						  SZ_2M,
						  ioat_chan->descs[idx].virt,
						  ioat_chan->descs[idx].hw);
				ioat_chan->descs[idx].virt = NULL;
				ioat_chan->descs[idx].hw = 0;
			}

			ioat_chan->desc_chunks = 0;
			kfree(ring);
			return NULL;
		}
		set_desc_id(ring[i], i);
	}

	/* link descs */
	for (i = 0; i < total_descs-1; i++) {
		struct ioat_ring_ent *next = ring[i+1];
		struct ioat_dma_descriptor *hw = ring[i]->hw;

		hw->next = next->txd.phys;
	}
	ring[i]->hw->next = ring[0]->txd.phys;

	return ring;
}

/**
 * ioat_check_space_lock - verify space and grab ring producer lock
 * @ioat: ioat,3 channel (ring) to operate on
 * @num_descs: allocation length
 */
int ioat_check_space_lock(struct ioatdma_chan *ioat_chan, int num_descs)
	__acquires(&ioat_chan->prep_lock)
{
	spin_lock_bh(&ioat_chan->prep_lock);
	/* never allow the last descriptor to be consumed, we need at
	 * least one free at all times to allow for on-the-fly ring
	 * resizing.
	 */
	if (likely(ioat_ring_space(ioat_chan) > num_descs)) {
		dev_dbg(to_dev(ioat_chan), "%s: num_descs: %d (%x:%x:%x)\n",
			__func__, num_descs, ioat_chan->head,
			ioat_chan->tail, ioat_chan->issued);
		ioat_chan->produce = num_descs;
		return 0;  /* with ioat->prep_lock held */
	}
	spin_unlock_bh(&ioat_chan->prep_lock);

	dev_dbg_ratelimited(to_dev(ioat_chan),
			    "%s: ring full! num_descs: %d (%x:%x:%x)\n",
			    __func__, num_descs, ioat_chan->head,
			    ioat_chan->tail, ioat_chan->issued);

	/* progress reclaim in the allocation failure case we may be
	 * called under bh_disabled so we need to trigger the timer
	 * event directly
	 */
	if (time_is_before_jiffies(ioat_chan->timer.expires)
	    && timer_pending(&ioat_chan->timer)) {
		mod_timer(&ioat_chan->timer, jiffies + COMPLETION_TIMEOUT);
		ioat_timer_event((unsigned long)ioat_chan);
	}

	return -ENOMEM;
}

static bool desc_has_ext(struct ioat_ring_ent *desc)
{
	struct ioat_dma_descriptor *hw = desc->hw;

	if (hw->ctl_f.op == IOAT_OP_XOR ||
	    hw->ctl_f.op == IOAT_OP_XOR_VAL) {
		struct ioat_xor_descriptor *xor = desc->xor;

		if (src_cnt_to_sw(xor->ctl_f.src_cnt) > 5)
			return true;
	} else if (hw->ctl_f.op == IOAT_OP_PQ ||
		   hw->ctl_f.op == IOAT_OP_PQ_VAL) {
		struct ioat_pq_descriptor *pq = desc->pq;

		if (src_cnt_to_sw(pq->ctl_f.src_cnt) > 3)
			return true;
	}

	return false;
}

static void
ioat_free_sed(struct ioatdma_device *ioat_dma, struct ioat_sed_ent *sed)
{
	if (!sed)
		return;

	dma_pool_free(ioat_dma->sed_hw_pool[sed->hw_pool], sed->hw, sed->dma);
	kmem_cache_free(ioat_sed_cache, sed);
}

static u64 ioat_get_current_completion(struct ioatdma_chan *ioat_chan)
{
	u64 phys_complete;
	u64 completion;

	completion = *ioat_chan->completion;
	phys_complete = ioat_chansts_to_addr(completion);

	dev_dbg(to_dev(ioat_chan), "%s: phys_complete: %#llx\n", __func__,
		(unsigned long long) phys_complete);

	return phys_complete;
}

static bool ioat_cleanup_preamble(struct ioatdma_chan *ioat_chan,
				   u64 *phys_complete)
{
	*phys_complete = ioat_get_current_completion(ioat_chan);
	if (*phys_complete == ioat_chan->last_completion)
		return false;

	clear_bit(IOAT_COMPLETION_ACK, &ioat_chan->state);
	mod_timer(&ioat_chan->timer, jiffies + COMPLETION_TIMEOUT);

	return true;
}

static void
desc_get_errstat(struct ioatdma_chan *ioat_chan, struct ioat_ring_ent *desc)
{
	struct ioat_dma_descriptor *hw = desc->hw;

	switch (hw->ctl_f.op) {
	case IOAT_OP_PQ_VAL:
	case IOAT_OP_PQ_VAL_16S:
	{
		struct ioat_pq_descriptor *pq = desc->pq;

		/* check if there's error written */
		if (!pq->dwbes_f.wbes)
			return;

		/* need to set a chanerr var for checking to clear later */

		if (pq->dwbes_f.p_val_err)
			*desc->result |= SUM_CHECK_P_RESULT;

		if (pq->dwbes_f.q_val_err)
			*desc->result |= SUM_CHECK_Q_RESULT;

		return;
	}
	default:
		return;
	}
}

/**
 * __cleanup - reclaim used descriptors
 * @ioat: channel (ring) to clean
 */
static void __cleanup(struct ioatdma_chan *ioat_chan, dma_addr_t phys_complete)
{
	struct ioatdma_device *ioat_dma = ioat_chan->ioat_dma;
	struct ioat_ring_ent *desc;
	bool seen_current = false;
	int idx = ioat_chan->tail, i;
	u16 active;

	dev_dbg(to_dev(ioat_chan), "%s: head: %#x tail: %#x issued: %#x\n",
		__func__, ioat_chan->head, ioat_chan->tail, ioat_chan->issued);

	/*
	 * At restart of the channel, the completion address and the
	 * channel status will be 0 due to starting a new chain. Since
	 * it's new chain and the first descriptor "fails", there is
	 * nothing to clean up. We do not want to reap the entire submitted
	 * chain due to this 0 address value and then BUG.
	 */
	if (!phys_complete)
		return;

	active = ioat_ring_active(ioat_chan);
	for (i = 0; i < active && !seen_current; i++) {
		struct dma_async_tx_descriptor *tx;

		smp_read_barrier_depends();
		prefetch(ioat_get_ring_ent(ioat_chan, idx + i + 1));
		desc = ioat_get_ring_ent(ioat_chan, idx + i);
		dump_desc_dbg(ioat_chan, desc);

		/* set err stat if we are using dwbes */
		if (ioat_dma->cap & IOAT_CAP_DWBES)
			desc_get_errstat(ioat_chan, desc);

		tx = &desc->txd;
		if (tx->cookie) {
			dma_cookie_complete(tx);
			dma_descriptor_unmap(tx);
			if (tx->callback) {
				tx->callback(tx->callback_param);
				tx->callback = NULL;
			}
		}

		if (tx->phys == phys_complete)
			seen_current = true;

		/* skip extended descriptors */
		if (desc_has_ext(desc)) {
			BUG_ON(i + 1 >= active);
			i++;
		}

		/* cleanup super extended descriptors */
		if (desc->sed) {
			ioat_free_sed(ioat_dma, desc->sed);
			desc->sed = NULL;
		}
	}

	/* finish all descriptor reads before incrementing tail */
	smp_mb();
	ioat_chan->tail = idx + i;
	/* no active descs have written a completion? */
	BUG_ON(active && !seen_current);
	ioat_chan->last_completion = phys_complete;

	if (active - i == 0) {
		dev_dbg(to_dev(ioat_chan), "%s: cancel completion timeout\n",
			__func__);
		mod_timer(&ioat_chan->timer, jiffies + IDLE_TIMEOUT);
	}

	/* 5 microsecond delay per pending descriptor */
	writew(min((5 * (active - i)), IOAT_INTRDELAY_MASK),
	       ioat_chan->ioat_dma->reg_base + IOAT_INTRDELAY_OFFSET);
}

static void ioat_cleanup(struct ioatdma_chan *ioat_chan)
{
	u64 phys_complete;

	spin_lock_bh(&ioat_chan->cleanup_lock);

	if (ioat_cleanup_preamble(ioat_chan, &phys_complete))
		__cleanup(ioat_chan, phys_complete);

	if (is_ioat_halted(*ioat_chan->completion)) {
		u32 chanerr = readl(ioat_chan->reg_base + IOAT_CHANERR_OFFSET);

		if (chanerr & IOAT_CHANERR_HANDLE_MASK) {
			mod_timer(&ioat_chan->timer, jiffies + IDLE_TIMEOUT);
			ioat_eh(ioat_chan);
		}
	}

	spin_unlock_bh(&ioat_chan->cleanup_lock);
}

void ioat_cleanup_event(unsigned long data)
{
	struct ioatdma_chan *ioat_chan = to_ioat_chan((void *)data);

	ioat_cleanup(ioat_chan);
	if (!test_bit(IOAT_RUN, &ioat_chan->state))
		return;
	writew(IOAT_CHANCTRL_RUN, ioat_chan->reg_base + IOAT_CHANCTRL_OFFSET);
}

static void ioat_restart_channel(struct ioatdma_chan *ioat_chan)
{
	u64 phys_complete;

	ioat_quiesce(ioat_chan, 0);
	if (ioat_cleanup_preamble(ioat_chan, &phys_complete))
		__cleanup(ioat_chan, phys_complete);

	__ioat_restart_chan(ioat_chan);
}

static void ioat_eh(struct ioatdma_chan *ioat_chan)
{
	struct pci_dev *pdev = to_pdev(ioat_chan);
	struct ioat_dma_descriptor *hw;
	struct dma_async_tx_descriptor *tx;
	u64 phys_complete;
	struct ioat_ring_ent *desc;
	u32 err_handled = 0;
	u32 chanerr_int;
	u32 chanerr;

	/* cleanup so tail points to descriptor that caused the error */
	if (ioat_cleanup_preamble(ioat_chan, &phys_complete))
		__cleanup(ioat_chan, phys_complete);

	chanerr = readl(ioat_chan->reg_base + IOAT_CHANERR_OFFSET);
	pci_read_config_dword(pdev, IOAT_PCI_CHANERR_INT_OFFSET, &chanerr_int);

	dev_dbg(to_dev(ioat_chan), "%s: error = %x:%x\n",
		__func__, chanerr, chanerr_int);

	desc = ioat_get_ring_ent(ioat_chan, ioat_chan->tail);
	hw = desc->hw;
	dump_desc_dbg(ioat_chan, desc);

	switch (hw->ctl_f.op) {
	case IOAT_OP_XOR_VAL:
		if (chanerr & IOAT_CHANERR_XOR_P_OR_CRC_ERR) {
			*desc->result |= SUM_CHECK_P_RESULT;
			err_handled |= IOAT_CHANERR_XOR_P_OR_CRC_ERR;
		}
		break;
	case IOAT_OP_PQ_VAL:
	case IOAT_OP_PQ_VAL_16S:
		if (chanerr & IOAT_CHANERR_XOR_P_OR_CRC_ERR) {
			*desc->result |= SUM_CHECK_P_RESULT;
			err_handled |= IOAT_CHANERR_XOR_P_OR_CRC_ERR;
		}
		if (chanerr & IOAT_CHANERR_XOR_Q_ERR) {
			*desc->result |= SUM_CHECK_Q_RESULT;
			err_handled |= IOAT_CHANERR_XOR_Q_ERR;
		}
		break;
	}

	/* fault on unhandled error or spurious halt */
	if (chanerr ^ err_handled || chanerr == 0) {
		dev_err(to_dev(ioat_chan), "%s: fatal error (%x:%x)\n",
			__func__, chanerr, err_handled);
		BUG();
	} else { /* cleanup the faulty descriptor */
		tx = &desc->txd;
		if (tx->cookie) {
			dma_cookie_complete(tx);
			dma_descriptor_unmap(tx);
			if (tx->callback) {
				tx->callback(tx->callback_param);
				tx->callback = NULL;
			}
		}
	}

	writel(chanerr, ioat_chan->reg_base + IOAT_CHANERR_OFFSET);
	pci_write_config_dword(pdev, IOAT_PCI_CHANERR_INT_OFFSET, chanerr_int);

	/* mark faulting descriptor as complete */
	*ioat_chan->completion = desc->txd.phys;

	spin_lock_bh(&ioat_chan->prep_lock);
	ioat_restart_channel(ioat_chan);
	spin_unlock_bh(&ioat_chan->prep_lock);
}

static void check_active(struct ioatdma_chan *ioat_chan)
{
	if (ioat_ring_active(ioat_chan)) {
		mod_timer(&ioat_chan->timer, jiffies + COMPLETION_TIMEOUT);
		return;
	}

	if (test_and_clear_bit(IOAT_CHAN_ACTIVE, &ioat_chan->state))
		mod_timer(&ioat_chan->timer, jiffies + IDLE_TIMEOUT);
}

void ioat_timer_event(unsigned long data)
{
	struct ioatdma_chan *ioat_chan = to_ioat_chan((void *)data);
	dma_addr_t phys_complete;
	u64 status;

	status = ioat_chansts(ioat_chan);

	/* when halted due to errors check for channel
	 * programming errors before advancing the completion state
	 */
	if (is_ioat_halted(status)) {
		u32 chanerr;

		chanerr = readl(ioat_chan->reg_base + IOAT_CHANERR_OFFSET);
		dev_err(to_dev(ioat_chan), "%s: Channel halted (%x)\n",
			__func__, chanerr);
		if (test_bit(IOAT_RUN, &ioat_chan->state))
			BUG_ON(is_ioat_bug(chanerr));
		else /* we never got off the ground */
			return;
	}

	/* if we haven't made progress and we have already
	 * acknowledged a pending completion once, then be more
	 * forceful with a restart
	 */
	spin_lock_bh(&ioat_chan->cleanup_lock);
	if (ioat_cleanup_preamble(ioat_chan, &phys_complete))
		__cleanup(ioat_chan, phys_complete);
	else if (test_bit(IOAT_COMPLETION_ACK, &ioat_chan->state)) {
		spin_lock_bh(&ioat_chan->prep_lock);
		ioat_restart_channel(ioat_chan);
		spin_unlock_bh(&ioat_chan->prep_lock);
		spin_unlock_bh(&ioat_chan->cleanup_lock);
		return;
	} else {
		set_bit(IOAT_COMPLETION_ACK, &ioat_chan->state);
		mod_timer(&ioat_chan->timer, jiffies + COMPLETION_TIMEOUT);
	}


	if (ioat_ring_active(ioat_chan))
		mod_timer(&ioat_chan->timer, jiffies + COMPLETION_TIMEOUT);
	else {
		spin_lock_bh(&ioat_chan->prep_lock);
		check_active(ioat_chan);
		spin_unlock_bh(&ioat_chan->prep_lock);
	}
	spin_unlock_bh(&ioat_chan->cleanup_lock);
}

enum dma_status
ioat_tx_status(struct dma_chan *c, dma_cookie_t cookie,
		struct dma_tx_state *txstate)
{
	struct ioatdma_chan *ioat_chan = to_ioat_chan(c);
	enum dma_status ret;

	ret = dma_cookie_status(c, cookie, txstate);
	if (ret == DMA_COMPLETE)
		return ret;

	ioat_cleanup(ioat_chan);

	return dma_cookie_status(c, cookie, txstate);
}

int ioat_reset_hw(struct ioatdma_chan *ioat_chan)
{
	/* throw away whatever the channel was doing and get it
	 * initialized, with ioat3 specific workarounds
	 */
	struct ioatdma_device *ioat_dma = ioat_chan->ioat_dma;
	struct pci_dev *pdev = ioat_dma->pdev;
	u32 chanerr;
	u16 dev_id;
	int err;

	ioat_quiesce(ioat_chan, msecs_to_jiffies(100));

	chanerr = readl(ioat_chan->reg_base + IOAT_CHANERR_OFFSET);
	writel(chanerr, ioat_chan->reg_base + IOAT_CHANERR_OFFSET);

	if (ioat_dma->version < IOAT_VER_3_3) {
		/* clear any pending errors */
		err = pci_read_config_dword(pdev,
				IOAT_PCI_CHANERR_INT_OFFSET, &chanerr);
		if (err) {
			dev_err(&pdev->dev,
				"channel error register unreachable\n");
			return err;
		}
		pci_write_config_dword(pdev,
				IOAT_PCI_CHANERR_INT_OFFSET, chanerr);

		/* Clear DMAUNCERRSTS Cfg-Reg Parity Error status bit
		 * (workaround for spurious config parity error after restart)
		 */
		pci_read_config_word(pdev, IOAT_PCI_DEVICE_ID_OFFSET, &dev_id);
		if (dev_id == PCI_DEVICE_ID_INTEL_IOAT_TBG0) {
			pci_write_config_dword(pdev,
					       IOAT_PCI_DMAUNCERRSTS_OFFSET,
					       0x10);
		}
	}

	if (is_bwd_ioat(pdev) && (ioat_dma->irq_mode == IOAT_MSIX)) {
		ioat_dma->msixtba0 = readq(ioat_dma->reg_base + 0x1000);
		ioat_dma->msixdata0 = readq(ioat_dma->reg_base + 0x1008);
		ioat_dma->msixpba = readq(ioat_dma->reg_base + 0x1800);
	}


	err = ioat_reset_sync(ioat_chan, msecs_to_jiffies(200));
	if (!err) {
		if (is_bwd_ioat(pdev) && (ioat_dma->irq_mode == IOAT_MSIX)) {
			writeq(ioat_dma->msixtba0, ioat_dma->reg_base + 0x1000);
			writeq(ioat_dma->msixdata0, ioat_dma->reg_base + 0x1008);
			writeq(ioat_dma->msixpba, ioat_dma->reg_base + 0x1800);
		}
	}

	if (err)
		dev_err(&pdev->dev, "Failed to reset: %d\n", err);

	return err;
}