aboutsummaryrefslogtreecommitdiff
path: root/drivers/dma/ioat/dma_v2.c
blob: 143cb1b3cb269908b15107bd26364bdac9875def (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
/*
 * Intel I/OAT DMA Linux driver
 * Copyright(c) 2004 - 2009 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.
 *
 * 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.,
 * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
 *
 * 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 (versions >= 2), which
 * does asynchronous data movement and checksumming 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/i7300_idle.h>
#include "dma.h"
#include "dma_v2.h"
#include "registers.h"
#include "hw.h"

int ioat_ring_alloc_order = 8;
module_param(ioat_ring_alloc_order, int, 0644);
MODULE_PARM_DESC(ioat_ring_alloc_order,
		 "ioat2+: allocate 2^n descriptors per channel"
		 " (default: 8 max: 16)");
static int ioat_ring_max_alloc_order = IOAT_MAX_ORDER;
module_param(ioat_ring_max_alloc_order, int, 0644);
MODULE_PARM_DESC(ioat_ring_max_alloc_order,
		 "ioat2+: upper limit for ring size (default: 16)");

void __ioat2_issue_pending(struct ioat2_dma_chan *ioat)
{
	struct ioat_chan_common *chan = &ioat->base;

	ioat->dmacount += ioat2_ring_pending(ioat);
	ioat->issued = ioat->head;
	writew(ioat->dmacount, chan->reg_base + IOAT_CHAN_DMACOUNT_OFFSET);
	dev_dbg(to_dev(chan),
		"%s: head: %#x tail: %#x issued: %#x count: %#x\n",
		__func__, ioat->head, ioat->tail, ioat->issued, ioat->dmacount);
}

void ioat2_issue_pending(struct dma_chan *c)
{
	struct ioat2_dma_chan *ioat = to_ioat2_chan(c);

	if (ioat2_ring_pending(ioat)) {
		spin_lock_bh(&ioat->prep_lock);
		__ioat2_issue_pending(ioat);
		spin_unlock_bh(&ioat->prep_lock);
	}
}

/**
 * ioat2_update_pending - log pending descriptors
 * @ioat: ioat2+ channel
 *
 * Check if the number of unsubmitted descriptors has exceeded the
 * watermark.  Called with prep_lock held
 */
static void ioat2_update_pending(struct ioat2_dma_chan *ioat)
{
	if (ioat2_ring_pending(ioat) > ioat_pending_level)
		__ioat2_issue_pending(ioat);
}

static void __ioat2_start_null_desc(struct ioat2_dma_chan *ioat)
{
	struct ioat_ring_ent *desc;
	struct ioat_dma_descriptor *hw;

	if (ioat2_ring_space(ioat) < 1) {
		dev_err(to_dev(&ioat->base),
			"Unable to start null desc - ring full\n");
		return;
	}

	dev_dbg(to_dev(&ioat->base), "%s: head: %#x tail: %#x issued: %#x\n",
		__func__, ioat->head, ioat->tail, ioat->issued);
	desc = ioat2_get_ring_ent(ioat, ioat->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);
	ioat2_set_chainaddr(ioat, desc->txd.phys);
	dump_desc_dbg(ioat, desc);
	wmb();
	ioat->head += 1;
	__ioat2_issue_pending(ioat);
}

static void ioat2_start_null_desc(struct ioat2_dma_chan *ioat)
{
	spin_lock_bh(&ioat->prep_lock);
	__ioat2_start_null_desc(ioat);
	spin_unlock_bh(&ioat->prep_lock);
}

static void __cleanup(struct ioat2_dma_chan *ioat, dma_addr_t phys_complete)
{
	struct ioat_chan_common *chan = &ioat->base;
	struct dma_async_tx_descriptor *tx;
	struct ioat_ring_ent *desc;
	bool seen_current = false;
	u16 active;
	int idx = ioat->tail, i;

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

	active = ioat2_ring_active(ioat);
	for (i = 0; i < active && !seen_current; i++) {
		smp_read_barrier_depends();
		prefetch(ioat2_get_ring_ent(ioat, idx + i + 1));
		desc = ioat2_get_ring_ent(ioat, idx + i);
		tx = &desc->txd;
		dump_desc_dbg(ioat, desc);
		if (tx->cookie) {
			ioat_dma_unmap(chan, tx->flags, desc->len, desc->hw);
			chan->completed_cookie = tx->cookie;
			tx->cookie = 0;
			if (tx->callback) {
				tx->callback(tx->callback_param);
				tx->callback = NULL;
			}
		}

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

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

/**
 * ioat2_cleanup - clean finished descriptors (advance tail pointer)
 * @chan: ioat channel to be cleaned up
 */
static void ioat2_cleanup(struct ioat2_dma_chan *ioat)
{
	struct ioat_chan_common *chan = &ioat->base;
	dma_addr_t phys_complete;

	spin_lock_bh(&chan->cleanup_lock);
	if (ioat_cleanup_preamble(chan, &phys_complete))
		__cleanup(ioat, phys_complete);
	spin_unlock_bh(&chan->cleanup_lock);
}

void ioat2_cleanup_event(unsigned long data)
{
	struct ioat2_dma_chan *ioat = to_ioat2_chan((void *) data);

	ioat2_cleanup(ioat);
	writew(IOAT_CHANCTRL_RUN, ioat->base.reg_base + IOAT_CHANCTRL_OFFSET);
}

void __ioat2_restart_chan(struct ioat2_dma_chan *ioat)
{
	struct ioat_chan_common *chan = &ioat->base;

	/* set the tail to be re-issued */
	ioat->issued = ioat->tail;
	ioat->dmacount = 0;
	set_bit(IOAT_COMPLETION_PENDING, &chan->state);
	mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT);

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

	if (ioat2_ring_pending(ioat)) {
		struct ioat_ring_ent *desc;

		desc = ioat2_get_ring_ent(ioat, ioat->tail);
		ioat2_set_chainaddr(ioat, desc->txd.phys);
		__ioat2_issue_pending(ioat);
	} else
		__ioat2_start_null_desc(ioat);
}

int ioat2_quiesce(struct ioat_chan_common *chan, unsigned long tmo)
{
	unsigned long end = jiffies + tmo;
	int err = 0;
	u32 status;

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

	return err;
}

int ioat2_reset_sync(struct ioat_chan_common *chan, unsigned long tmo)
{
	unsigned long end = jiffies + tmo;
	int err = 0;

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

	return err;
}

static void ioat2_restart_channel(struct ioat2_dma_chan *ioat)
{
	struct ioat_chan_common *chan = &ioat->base;
	dma_addr_t phys_complete;

	ioat2_quiesce(chan, 0);
	if (ioat_cleanup_preamble(chan, &phys_complete))
		__cleanup(ioat, phys_complete);

	__ioat2_restart_chan(ioat);
}

void ioat2_timer_event(unsigned long data)
{
	struct ioat2_dma_chan *ioat = to_ioat2_chan((void *) data);
	struct ioat_chan_common *chan = &ioat->base;

	if (test_bit(IOAT_COMPLETION_PENDING, &chan->state)) {
		dma_addr_t phys_complete;
		u64 status;

		status = ioat_chansts(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(chan->reg_base + IOAT_CHANERR_OFFSET);
			dev_err(to_dev(chan), "%s: Channel halted (%x)\n",
				__func__, chanerr);
			if (test_bit(IOAT_RUN, &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(&chan->cleanup_lock);
		if (ioat_cleanup_preamble(chan, &phys_complete)) {
			__cleanup(ioat, phys_complete);
		} else if (test_bit(IOAT_COMPLETION_ACK, &chan->state)) {
			spin_lock_bh(&ioat->prep_lock);
			ioat2_restart_channel(ioat);
			spin_unlock_bh(&ioat->prep_lock);
		} else {
			set_bit(IOAT_COMPLETION_ACK, &chan->state);
			mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT);
		}
		spin_unlock_bh(&chan->cleanup_lock);
	} else {
		u16 active;

		/* if the ring is idle, empty, and oversized try to step
		 * down the size
		 */
		spin_lock_bh(&chan->cleanup_lock);
		spin_lock_bh(&ioat->prep_lock);
		active = ioat2_ring_active(ioat);
		if (active == 0 && ioat->alloc_order > ioat_get_alloc_order())
			reshape_ring(ioat, ioat->alloc_order-1);
		spin_unlock_bh(&ioat->prep_lock);
		spin_unlock_bh(&chan->cleanup_lock);

		/* keep shrinking until we get back to our minimum
		 * default size
		 */
		if (ioat->alloc_order > ioat_get_alloc_order())
			mod_timer(&chan->timer, jiffies + IDLE_TIMEOUT);
	}
}

static int ioat2_reset_hw(struct ioat_chan_common *chan)
{
	/* throw away whatever the channel was doing and get it initialized */
	u32 chanerr;

	ioat2_quiesce(chan, msecs_to_jiffies(100));

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

	return ioat2_reset_sync(chan, msecs_to_jiffies(200));
}

/**
 * ioat2_enumerate_channels - find and initialize the device's channels
 * @device: the device to be enumerated
 */
int ioat2_enumerate_channels(struct ioatdma_device *device)
{
	struct ioat2_dma_chan *ioat;
	struct device *dev = &device->pdev->dev;
	struct dma_device *dma = &device->common;
	u8 xfercap_log;
	int i;

	INIT_LIST_HEAD(&dma->channels);
	dma->chancnt = readb(device->reg_base + IOAT_CHANCNT_OFFSET);
	dma->chancnt &= 0x1f; /* bits [4:0] valid */
	if (dma->chancnt > ARRAY_SIZE(device->idx)) {
		dev_warn(dev, "(%d) exceeds max supported channels (%zu)\n",
			 dma->chancnt, ARRAY_SIZE(device->idx));
		dma->chancnt = ARRAY_SIZE(device->idx);
	}
	xfercap_log = readb(device->reg_base + IOAT_XFERCAP_OFFSET);
	xfercap_log &= 0x1f; /* bits [4:0] valid */
	if (xfercap_log == 0)
		return 0;
	dev_dbg(dev, "%s: xfercap = %d\n", __func__, 1 << xfercap_log);

	/* FIXME which i/oat version is i7300? */
#ifdef CONFIG_I7300_IDLE_IOAT_CHANNEL
	if (i7300_idle_platform_probe(NULL, NULL, 1) == 0)
		dma->chancnt--;
#endif
	for (i = 0; i < dma->chancnt; i++) {
		ioat = devm_kzalloc(dev, sizeof(*ioat), GFP_KERNEL);
		if (!ioat)
			break;

		ioat_init_channel(device, &ioat->base, i);
		ioat->xfercap_log = xfercap_log;
		spin_lock_init(&ioat->prep_lock);
		if (device->reset_hw(&ioat->base)) {
			i = 0;
			break;
		}
	}
	dma->chancnt = i;
	return i;
}

static dma_cookie_t ioat2_tx_submit_unlock(struct dma_async_tx_descriptor *tx)
{
	struct dma_chan *c = tx->chan;
	struct ioat2_dma_chan *ioat = to_ioat2_chan(c);
	struct ioat_chan_common *chan = &ioat->base;
	dma_cookie_t cookie = c->cookie;

	cookie++;
	if (cookie < 0)
		cookie = 1;
	tx->cookie = cookie;
	c->cookie = cookie;
	dev_dbg(to_dev(&ioat->base), "%s: cookie: %d\n", __func__, cookie);

	if (!test_and_set_bit(IOAT_COMPLETION_PENDING, &chan->state))
		mod_timer(&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->head += ioat->produce;

	ioat2_update_pending(ioat);
	spin_unlock_bh(&ioat->prep_lock);

	return cookie;
}

static struct ioat_ring_ent *ioat2_alloc_ring_ent(struct dma_chan *chan, gfp_t flags)
{
	struct ioat_dma_descriptor *hw;
	struct ioat_ring_ent *desc;
	struct ioatdma_device *dma;
	dma_addr_t phys;

	dma = to_ioatdma_device(chan->device);
	hw = pci_pool_alloc(dma->dma_pool, flags, &phys);
	if (!hw)
		return NULL;
	memset(hw, 0, sizeof(*hw));

	desc = kmem_cache_alloc(ioat2_cache, flags);
	if (!desc) {
		pci_pool_free(dma->dma_pool, hw, phys);
		return NULL;
	}
	memset(desc, 0, sizeof(*desc));

	dma_async_tx_descriptor_init(&desc->txd, chan);
	desc->txd.tx_submit = ioat2_tx_submit_unlock;
	desc->hw = hw;
	desc->txd.phys = phys;
	return desc;
}

static void ioat2_free_ring_ent(struct ioat_ring_ent *desc, struct dma_chan *chan)
{
	struct ioatdma_device *dma;

	dma = to_ioatdma_device(chan->device);
	pci_pool_free(dma->dma_pool, desc->hw, desc->txd.phys);
	kmem_cache_free(ioat2_cache, desc);
}

static struct ioat_ring_ent **ioat2_alloc_ring(struct dma_chan *c, int order, gfp_t flags)
{
	struct ioat_ring_ent **ring;
	int descs = 1 << order;
	int i;

	if (order > ioat_get_max_alloc_order())
		return NULL;

	/* allocate the array to hold the software ring */
	ring = kcalloc(descs, sizeof(*ring), flags);
	if (!ring)
		return NULL;
	for (i = 0; i < descs; i++) {
		ring[i] = ioat2_alloc_ring_ent(c, flags);
		if (!ring[i]) {
			while (i--)
				ioat2_free_ring_ent(ring[i], c);
			kfree(ring);
			return NULL;
		}
		set_desc_id(ring[i], i);
	}

	/* link descs */
	for (i = 0; i < 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;
}

void ioat2_free_chan_resources(struct dma_chan *c);

/* ioat2_alloc_chan_resources - allocate/initialize ioat2 descriptor ring
 * @chan: channel to be initialized
 */
int ioat2_alloc_chan_resources(struct dma_chan *c)
{
	struct ioat2_dma_chan *ioat = to_ioat2_chan(c);
	struct ioat_chan_common *chan = &ioat->base;
	struct ioat_ring_ent **ring;
	u64 status;
	int order;
	int i = 0;

	/* have we already been set up? */
	if (ioat->ring)
		return 1 << ioat->alloc_order;

	/* Setup register to interrupt and write completion status on error */
	writew(IOAT_CHANCTRL_RUN, chan->reg_base + IOAT_CHANCTRL_OFFSET);

	/* allocate a completion writeback area */
	/* doing 2 32bit writes to mmio since 1 64b write doesn't work */
	chan->completion = pci_pool_alloc(chan->device->completion_pool,
					  GFP_KERNEL, &chan->completion_dma);
	if (!chan->completion)
		return -ENOMEM;

	memset(chan->completion, 0, sizeof(*chan->completion));
	writel(((u64) chan->completion_dma) & 0x00000000FFFFFFFF,
	       chan->reg_base + IOAT_CHANCMP_OFFSET_LOW);
	writel(((u64) chan->completion_dma) >> 32,
	       chan->reg_base + IOAT_CHANCMP_OFFSET_HIGH);

	order = ioat_get_alloc_order();
	ring = ioat2_alloc_ring(c, order, GFP_KERNEL);
	if (!ring)
		return -ENOMEM;

	spin_lock_bh(&chan->cleanup_lock);
	spin_lock_bh(&ioat->prep_lock);
	ioat->ring = ring;
	ioat->head = 0;
	ioat->issued = 0;
	ioat->tail = 0;
	ioat->alloc_order = order;
	spin_unlock_bh(&ioat->prep_lock);
	spin_unlock_bh(&chan->cleanup_lock);

	tasklet_enable(&chan->cleanup_task);
	ioat2_start_null_desc(ioat);

	/* check that we got off the ground */
	do {
		udelay(1);
		status = ioat_chansts(chan);
	} while (i++ < 20 && !is_ioat_active(status) && !is_ioat_idle(status));

	if (is_ioat_active(status) || is_ioat_idle(status)) {
		set_bit(IOAT_RUN, &chan->state);
		return 1 << ioat->alloc_order;
	} else {
		u32 chanerr = readl(chan->reg_base + IOAT_CHANERR_OFFSET);

		dev_WARN(to_dev(chan),
			"failed to start channel chanerr: %#x\n", chanerr);
		ioat2_free_chan_resources(c);
		return -EFAULT;
	}
}

bool reshape_ring(struct ioat2_dma_chan *ioat, int order)
{
	/* reshape differs from normal ring allocation in that we want
	 * to allocate a new software ring while only
	 * extending/truncating the hardware ring
	 */
	struct ioat_chan_common *chan = &ioat->base;
	struct dma_chan *c = &chan->common;
	const u32 curr_size = ioat2_ring_size(ioat);
	const u16 active = ioat2_ring_active(ioat);
	const u32 new_size = 1 << order;
	struct ioat_ring_ent **ring;
	u16 i;

	if (order > ioat_get_max_alloc_order())
		return false;

	/* double check that we have at least 1 free descriptor */
	if (active == curr_size)
		return false;

	/* when shrinking, verify that we can hold the current active
	 * set in the new ring
	 */
	if (active >= new_size)
		return false;

	/* allocate the array to hold the software ring */
	ring = kcalloc(new_size, sizeof(*ring), GFP_NOWAIT);
	if (!ring)
		return false;

	/* allocate/trim descriptors as needed */
	if (new_size > curr_size) {
		/* copy current descriptors to the new ring */
		for (i = 0; i < curr_size; i++) {
			u16 curr_idx = (ioat->tail+i) & (curr_size-1);
			u16 new_idx = (ioat->tail+i) & (new_size-1);

			ring[new_idx] = ioat->ring[curr_idx];
			set_desc_id(ring[new_idx], new_idx);
		}

		/* add new descriptors to the ring */
		for (i = curr_size; i < new_size; i++) {
			u16 new_idx = (ioat->tail+i) & (new_size-1);

			ring[new_idx] = ioat2_alloc_ring_ent(c, GFP_NOWAIT);
			if (!ring[new_idx]) {
				while (i--) {
					u16 new_idx = (ioat->tail+i) & (new_size-1);

					ioat2_free_ring_ent(ring[new_idx], c);
				}
				kfree(ring);
				return false;
			}
			set_desc_id(ring[new_idx], new_idx);
		}

		/* hw link new descriptors */
		for (i = curr_size-1; i < new_size; i++) {
			u16 new_idx = (ioat->tail+i) & (new_size-1);
			struct ioat_ring_ent *next = ring[(new_idx+1) & (new_size-1)];
			struct ioat_dma_descriptor *hw = ring[new_idx]->hw;

			hw->next = next->txd.phys;
		}
	} else {
		struct ioat_dma_descriptor *hw;
		struct ioat_ring_ent *next;

		/* copy current descriptors to the new ring, dropping the
		 * removed descriptors
		 */
		for (i = 0; i < new_size; i++) {
			u16 curr_idx = (ioat->tail+i) & (curr_size-1);
			u16 new_idx = (ioat->tail+i) & (new_size-1);

			ring[new_idx] = ioat->ring[curr_idx];
			set_desc_id(ring[new_idx], new_idx);
		}

		/* free deleted descriptors */
		for (i = new_size; i < curr_size; i++) {
			struct ioat_ring_ent *ent;

			ent = ioat2_get_ring_ent(ioat, ioat->tail+i);
			ioat2_free_ring_ent(ent, c);
		}

		/* fix up hardware ring */
		hw = ring[(ioat->tail+new_size-1) & (new_size-1)]->hw;
		next = ring[(ioat->tail+new_size) & (new_size-1)];
		hw->next = next->txd.phys;
	}

	dev_dbg(to_dev(chan), "%s: allocated %d descriptors\n",
		__func__, new_size);

	kfree(ioat->ring);
	ioat->ring = ring;
	ioat->alloc_order = order;

	return true;
}

/**
 * ioat2_check_space_lock - verify space and grab ring producer lock
 * @ioat: ioat2,3 channel (ring) to operate on
 * @num_descs: allocation length
 */
int ioat2_check_space_lock(struct ioat2_dma_chan *ioat, int num_descs)
{
	struct ioat_chan_common *chan = &ioat->base;
	bool retry;

 retry:
	spin_lock_bh(&ioat->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(ioat2_ring_space(ioat) > num_descs)) {
		dev_dbg(to_dev(chan), "%s: num_descs: %d (%x:%x:%x)\n",
			__func__, num_descs, ioat->head, ioat->tail, ioat->issued);
		ioat->produce = num_descs;
		return 0;  /* with ioat->prep_lock held */
	}
	retry = test_and_set_bit(IOAT_RESHAPE_PENDING, &chan->state);
	spin_unlock_bh(&ioat->prep_lock);

	/* is another cpu already trying to expand the ring? */
	if (retry)
		goto retry;

	spin_lock_bh(&chan->cleanup_lock);
	spin_lock_bh(&ioat->prep_lock);
	retry = reshape_ring(ioat, ioat->alloc_order + 1);
	clear_bit(IOAT_RESHAPE_PENDING, &chan->state);
	spin_unlock_bh(&ioat->prep_lock);
	spin_unlock_bh(&chan->cleanup_lock);

	/* if we were able to expand the ring retry the allocation */
	if (retry)
		goto retry;

	if (printk_ratelimit())
		dev_dbg(to_dev(chan), "%s: ring full! num_descs: %d (%x:%x:%x)\n",
			__func__, num_descs, ioat->head, ioat->tail, ioat->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 (jiffies > chan->timer.expires && timer_pending(&chan->timer)) {
		struct ioatdma_device *device = chan->device;

		mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT);
		device->timer_fn((unsigned long) &chan->common);
	}

	return -ENOMEM;
}

struct dma_async_tx_descriptor *
ioat2_dma_prep_memcpy_lock(struct dma_chan *c, dma_addr_t dma_dest,
			   dma_addr_t dma_src, size_t len, unsigned long flags)
{
	struct ioat2_dma_chan *ioat = to_ioat2_chan(c);
	struct ioat_dma_descriptor *hw;
	struct ioat_ring_ent *desc;
	dma_addr_t dst = dma_dest;
	dma_addr_t src = dma_src;
	size_t total_len = len;
	int num_descs, idx, i;

	num_descs = ioat2_xferlen_to_descs(ioat, len);
	if (likely(num_descs) && ioat2_check_space_lock(ioat, num_descs) == 0)
		idx = ioat->head;
	else
		return NULL;
	i = 0;
	do {
		size_t copy = min_t(size_t, len, 1 << ioat->xfercap_log);

		desc = ioat2_get_ring_ent(ioat, idx + i);
		hw = desc->hw;

		hw->size = copy;
		hw->ctl = 0;
		hw->src_addr = src;
		hw->dst_addr = dst;

		len -= copy;
		dst += copy;
		src += copy;
		dump_desc_dbg(ioat, desc);
	} while (++i < num_descs);

	desc->txd.flags = flags;
	desc->len = total_len;
	hw->ctl_f.int_en = !!(flags & DMA_PREP_INTERRUPT);
	hw->ctl_f.fence = !!(flags & DMA_PREP_FENCE);
	hw->ctl_f.compl_write = 1;
	dump_desc_dbg(ioat, desc);
	/* we leave the channel locked to ensure in order submission */

	return &desc->txd;
}

/**
 * ioat2_free_chan_resources - release all the descriptors
 * @chan: the channel to be cleaned
 */
void ioat2_free_chan_resources(struct dma_chan *c)
{
	struct ioat2_dma_chan *ioat = to_ioat2_chan(c);
	struct ioat_chan_common *chan = &ioat->base;
	struct ioatdma_device *device = chan->device;
	struct ioat_ring_ent *desc;
	const u16 total_descs = 1 << ioat->alloc_order;
	int descs;
	int i;

	/* Before freeing channel resources first check
	 * if they have been previously allocated for this channel.
	 */
	if (!ioat->ring)
		return;

	tasklet_disable(&chan->cleanup_task);
	del_timer_sync(&chan->timer);
	device->cleanup_fn((unsigned long) c);
	device->reset_hw(chan);
	clear_bit(IOAT_RUN, &chan->state);

	spin_lock_bh(&chan->cleanup_lock);
	spin_lock_bh(&ioat->prep_lock);
	descs = ioat2_ring_space(ioat);
	dev_dbg(to_dev(chan), "freeing %d idle descriptors\n", descs);
	for (i = 0; i < descs; i++) {
		desc = ioat2_get_ring_ent(ioat, ioat->head + i);
		ioat2_free_ring_ent(desc, c);
	}

	if (descs < total_descs)
		dev_err(to_dev(chan), "Freeing %d in use descriptors!\n",
			total_descs - descs);

	for (i = 0; i < total_descs - descs; i++) {
		desc = ioat2_get_ring_ent(ioat, ioat->tail + i);
		dump_desc_dbg(ioat, desc);
		ioat2_free_ring_ent(desc, c);
	}

	kfree(ioat->ring);
	ioat->ring = NULL;
	ioat->alloc_order = 0;
	pci_pool_free(device->completion_pool, chan->completion,
		      chan->completion_dma);
	spin_unlock_bh(&ioat->prep_lock);
	spin_unlock_bh(&chan->cleanup_lock);

	chan->last_completion = 0;
	chan->completion_dma = 0;
	ioat->dmacount = 0;
}

static ssize_t ring_size_show(struct dma_chan *c, char *page)
{
	struct ioat2_dma_chan *ioat = to_ioat2_chan(c);

	return sprintf(page, "%d\n", (1 << ioat->alloc_order) & ~1);
}
static struct ioat_sysfs_entry ring_size_attr = __ATTR_RO(ring_size);

static ssize_t ring_active_show(struct dma_chan *c, char *page)
{
	struct ioat2_dma_chan *ioat = to_ioat2_chan(c);

	/* ...taken outside the lock, no need to be precise */
	return sprintf(page, "%d\n", ioat2_ring_active(ioat));
}
static struct ioat_sysfs_entry ring_active_attr = __ATTR_RO(ring_active);

static struct attribute *ioat2_attrs[] = {
	&ring_size_attr.attr,
	&ring_active_attr.attr,
	&ioat_cap_attr.attr,
	&ioat_version_attr.attr,
	NULL,
};

struct kobj_type ioat2_ktype = {
	.sysfs_ops = &ioat_sysfs_ops,
	.default_attrs = ioat2_attrs,
};

int __devinit ioat2_dma_probe(struct ioatdma_device *device, int dca)
{
	struct pci_dev *pdev = device->pdev;
	struct dma_device *dma;
	struct dma_chan *c;
	struct ioat_chan_common *chan;
	int err;

	device->enumerate_channels = ioat2_enumerate_channels;
	device->reset_hw = ioat2_reset_hw;
	device->cleanup_fn = ioat2_cleanup_event;
	device->timer_fn = ioat2_timer_event;
	device->self_test = ioat_dma_self_test;
	dma = &device->common;
	dma->device_prep_dma_memcpy = ioat2_dma_prep_memcpy_lock;
	dma->device_issue_pending = ioat2_issue_pending;
	dma->device_alloc_chan_resources = ioat2_alloc_chan_resources;
	dma->device_free_chan_resources = ioat2_free_chan_resources;
	dma->device_tx_status = ioat_dma_tx_status;

	err = ioat_probe(device);
	if (err)
		return err;
	ioat_set_tcp_copy_break(2048);

	list_for_each_entry(c, &dma->channels, device_node) {
		chan = to_chan_common(c);
		writel(IOAT_DCACTRL_CMPL_WRITE_ENABLE | IOAT_DMA_DCA_ANY_CPU,
		       chan->reg_base + IOAT_DCACTRL_OFFSET);
	}

	err = ioat_register(device);
	if (err)
		return err;

	ioat_kobject_add(device, &ioat2_ktype);

	if (dca)
		device->dca = ioat2_dca_init(pdev, device->reg_base);

	return err;
}