-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathioat_dma.c
1741 lines (1514 loc) · 48 KB
/
ioat_dma.c
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
/*
* 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, which does asynchronous
* copy operations.
*/
#include <linux/init.h>
#include <linux/module.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/i7300_idle.h>
#include "ioatdma.h"
#include "ioatdma_registers.h"
#include "ioatdma_hw.h"
#define to_ioat_chan(chan) container_of(chan, struct ioat_dma_chan, common)
#define to_ioatdma_device(dev) container_of(dev, struct ioatdma_device, common)
#define to_ioat_desc(lh) container_of(lh, struct ioat_desc_sw, node)
#define tx_to_ioat_desc(tx) container_of(tx, struct ioat_desc_sw, async_tx)
#define chan_num(ch) ((int)((ch)->reg_base - (ch)->device->reg_base) / 0x80)
static int ioat_pending_level = 4;
module_param(ioat_pending_level, int, 0644);
MODULE_PARM_DESC(ioat_pending_level,
"high-water mark for pushing ioat descriptors (default: 4)");
#define RESET_DELAY msecs_to_jiffies(100)
#define WATCHDOG_DELAY round_jiffies(msecs_to_jiffies(2000))
static void ioat_dma_chan_reset_part2(struct work_struct *work);
static void ioat_dma_chan_watchdog(struct work_struct *work);
/*
* workaround for IOAT ver.3.0 null descriptor issue
* (channel returns error when size is 0)
*/
#define NULL_DESC_BUFFER_SIZE 1
/* internal functions */
static void ioat_dma_start_null_desc(struct ioat_dma_chan *ioat_chan);
static void ioat_dma_memcpy_cleanup(struct ioat_dma_chan *ioat_chan);
static struct ioat_desc_sw *
ioat1_dma_get_next_descriptor(struct ioat_dma_chan *ioat_chan);
static struct ioat_desc_sw *
ioat2_dma_get_next_descriptor(struct ioat_dma_chan *ioat_chan);
static inline struct ioat_dma_chan *ioat_lookup_chan_by_index(
struct ioatdma_device *device,
int index)
{
return device->idx[index];
}
/**
* ioat_dma_do_interrupt - handler used for single vector interrupt mode
* @irq: interrupt id
* @data: interrupt data
*/
static irqreturn_t ioat_dma_do_interrupt(int irq, void *data)
{
struct ioatdma_device *instance = data;
struct ioat_dma_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_bit(bit, &attnstatus, BITS_PER_LONG) {
ioat_chan = ioat_lookup_chan_by_index(instance, bit);
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
*/
static irqreturn_t ioat_dma_do_interrupt_msix(int irq, void *data)
{
struct ioat_dma_chan *ioat_chan = data;
tasklet_schedule(&ioat_chan->cleanup_task);
return IRQ_HANDLED;
}
static void ioat_dma_cleanup_tasklet(unsigned long data);
/**
* ioat_dma_enumerate_channels - find and initialize the device's channels
* @device: the device to be enumerated
*/
static int ioat_dma_enumerate_channels(struct ioatdma_device *device)
{
u8 xfercap_scale;
u32 xfercap;
int i;
struct ioat_dma_chan *ioat_chan;
/*
* IOAT ver.3 workarounds
*/
if (device->version == IOAT_VER_3_0) {
u32 chan_err_mask;
u16 dev_id;
u32 dmauncerrsts;
/*
* Write CHANERRMSK_INT with 3E07h to mask out the errors
* that can cause stability issues for IOAT ver.3
*/
chan_err_mask = 0x3E07;
pci_write_config_dword(device->pdev,
IOAT_PCI_CHANERRMASK_INT_OFFSET,
chan_err_mask);
/*
* Clear DMAUNCERRSTS Cfg-Reg Parity Error status bit
* (workaround for spurious config parity error after restart)
*/
pci_read_config_word(device->pdev,
IOAT_PCI_DEVICE_ID_OFFSET,
&dev_id);
if (dev_id == PCI_DEVICE_ID_INTEL_IOAT_TBG0) {
dmauncerrsts = 0x10;
pci_write_config_dword(device->pdev,
IOAT_PCI_DMAUNCERRSTS_OFFSET,
dmauncerrsts);
}
}
device->common.chancnt = readb(device->reg_base + IOAT_CHANCNT_OFFSET);
xfercap_scale = readb(device->reg_base + IOAT_XFERCAP_OFFSET);
xfercap = (xfercap_scale == 0 ? -1 : (1UL << xfercap_scale));
#ifdef CONFIG_I7300_IDLE_IOAT_CHANNEL
if (i7300_idle_platform_probe(NULL, NULL, 1) == 0) {
device->common.chancnt--;
}
#endif
for (i = 0; i < device->common.chancnt; i++) {
ioat_chan = kzalloc(sizeof(*ioat_chan), GFP_KERNEL);
if (!ioat_chan) {
device->common.chancnt = i;
break;
}
ioat_chan->device = device;
ioat_chan->reg_base = device->reg_base + (0x80 * (i + 1));
ioat_chan->xfercap = xfercap;
ioat_chan->desccount = 0;
INIT_DELAYED_WORK(&ioat_chan->work, ioat_dma_chan_reset_part2);
if (ioat_chan->device->version == IOAT_VER_2_0)
writel(IOAT_DCACTRL_CMPL_WRITE_ENABLE |
IOAT_DMA_DCA_ANY_CPU,
ioat_chan->reg_base + IOAT_DCACTRL_OFFSET);
else if (ioat_chan->device->version == IOAT_VER_3_0)
writel(IOAT_DMA_DCA_ANY_CPU,
ioat_chan->reg_base + IOAT_DCACTRL_OFFSET);
spin_lock_init(&ioat_chan->cleanup_lock);
spin_lock_init(&ioat_chan->desc_lock);
INIT_LIST_HEAD(&ioat_chan->free_desc);
INIT_LIST_HEAD(&ioat_chan->used_desc);
/* This should be made common somewhere in dmaengine.c */
ioat_chan->common.device = &device->common;
list_add_tail(&ioat_chan->common.device_node,
&device->common.channels);
device->idx[i] = ioat_chan;
tasklet_init(&ioat_chan->cleanup_task,
ioat_dma_cleanup_tasklet,
(unsigned long) ioat_chan);
tasklet_disable(&ioat_chan->cleanup_task);
}
return device->common.chancnt;
}
/**
* ioat_dma_memcpy_issue_pending - push potentially unrecognized appended
* descriptors to hw
* @chan: DMA channel handle
*/
static inline void __ioat1_dma_memcpy_issue_pending(
struct ioat_dma_chan *ioat_chan)
{
ioat_chan->pending = 0;
writeb(IOAT_CHANCMD_APPEND, ioat_chan->reg_base + IOAT1_CHANCMD_OFFSET);
}
static void ioat1_dma_memcpy_issue_pending(struct dma_chan *chan)
{
struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);
if (ioat_chan->pending > 0) {
spin_lock_bh(&ioat_chan->desc_lock);
__ioat1_dma_memcpy_issue_pending(ioat_chan);
spin_unlock_bh(&ioat_chan->desc_lock);
}
}
static inline void __ioat2_dma_memcpy_issue_pending(
struct ioat_dma_chan *ioat_chan)
{
ioat_chan->pending = 0;
writew(ioat_chan->dmacount,
ioat_chan->reg_base + IOAT_CHAN_DMACOUNT_OFFSET);
}
static void ioat2_dma_memcpy_issue_pending(struct dma_chan *chan)
{
struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);
if (ioat_chan->pending > 0) {
spin_lock_bh(&ioat_chan->desc_lock);
__ioat2_dma_memcpy_issue_pending(ioat_chan);
spin_unlock_bh(&ioat_chan->desc_lock);
}
}
/**
* ioat_dma_chan_reset_part2 - reinit the channel after a reset
*/
static void ioat_dma_chan_reset_part2(struct work_struct *work)
{
struct ioat_dma_chan *ioat_chan =
container_of(work, struct ioat_dma_chan, work.work);
struct ioat_desc_sw *desc;
spin_lock_bh(&ioat_chan->cleanup_lock);
spin_lock_bh(&ioat_chan->desc_lock);
ioat_chan->completion_virt->low = 0;
ioat_chan->completion_virt->high = 0;
ioat_chan->pending = 0;
/*
* count the descriptors waiting, and be sure to do it
* right for both the CB1 line and the CB2 ring
*/
ioat_chan->dmacount = 0;
if (ioat_chan->used_desc.prev) {
desc = to_ioat_desc(ioat_chan->used_desc.prev);
do {
ioat_chan->dmacount++;
desc = to_ioat_desc(desc->node.next);
} while (&desc->node != ioat_chan->used_desc.next);
}
/*
* write the new starting descriptor address
* this puts channel engine into ARMED state
*/
desc = to_ioat_desc(ioat_chan->used_desc.prev);
switch (ioat_chan->device->version) {
case IOAT_VER_1_2:
writel(((u64) desc->async_tx.phys) & 0x00000000FFFFFFFF,
ioat_chan->reg_base + IOAT1_CHAINADDR_OFFSET_LOW);
writel(((u64) desc->async_tx.phys) >> 32,
ioat_chan->reg_base + IOAT1_CHAINADDR_OFFSET_HIGH);
writeb(IOAT_CHANCMD_START, ioat_chan->reg_base
+ IOAT_CHANCMD_OFFSET(ioat_chan->device->version));
break;
case IOAT_VER_2_0:
writel(((u64) desc->async_tx.phys) & 0x00000000FFFFFFFF,
ioat_chan->reg_base + IOAT2_CHAINADDR_OFFSET_LOW);
writel(((u64) desc->async_tx.phys) >> 32,
ioat_chan->reg_base + IOAT2_CHAINADDR_OFFSET_HIGH);
/* tell the engine to go with what's left to be done */
writew(ioat_chan->dmacount,
ioat_chan->reg_base + IOAT_CHAN_DMACOUNT_OFFSET);
break;
}
dev_err(&ioat_chan->device->pdev->dev,
"chan%d reset - %d descs waiting, %d total desc\n",
chan_num(ioat_chan), ioat_chan->dmacount, ioat_chan->desccount);
spin_unlock_bh(&ioat_chan->desc_lock);
spin_unlock_bh(&ioat_chan->cleanup_lock);
}
/**
* ioat_dma_reset_channel - restart a channel
* @ioat_chan: IOAT DMA channel handle
*/
static void ioat_dma_reset_channel(struct ioat_dma_chan *ioat_chan)
{
u32 chansts, chanerr;
if (!ioat_chan->used_desc.prev)
return;
chanerr = readl(ioat_chan->reg_base + IOAT_CHANERR_OFFSET);
chansts = (ioat_chan->completion_virt->low
& IOAT_CHANSTS_DMA_TRANSFER_STATUS);
if (chanerr) {
dev_err(&ioat_chan->device->pdev->dev,
"chan%d, CHANSTS = 0x%08x CHANERR = 0x%04x, clearing\n",
chan_num(ioat_chan), chansts, chanerr);
writel(chanerr, ioat_chan->reg_base + IOAT_CHANERR_OFFSET);
}
/*
* whack it upside the head with a reset
* and wait for things to settle out.
* force the pending count to a really big negative
* to make sure no one forces an issue_pending
* while we're waiting.
*/
spin_lock_bh(&ioat_chan->desc_lock);
ioat_chan->pending = INT_MIN;
writeb(IOAT_CHANCMD_RESET,
ioat_chan->reg_base
+ IOAT_CHANCMD_OFFSET(ioat_chan->device->version));
spin_unlock_bh(&ioat_chan->desc_lock);
/* schedule the 2nd half instead of sleeping a long time */
schedule_delayed_work(&ioat_chan->work, RESET_DELAY);
}
/**
* ioat_dma_chan_watchdog - watch for stuck channels
*/
static void ioat_dma_chan_watchdog(struct work_struct *work)
{
struct ioatdma_device *device =
container_of(work, struct ioatdma_device, work.work);
struct ioat_dma_chan *ioat_chan;
int i;
union {
u64 full;
struct {
u32 low;
u32 high;
};
} completion_hw;
unsigned long compl_desc_addr_hw;
for (i = 0; i < device->common.chancnt; i++) {
ioat_chan = ioat_lookup_chan_by_index(device, i);
if (ioat_chan->device->version == IOAT_VER_1_2
/* have we started processing anything yet */
&& ioat_chan->last_completion
/* have we completed any since last watchdog cycle? */
&& (ioat_chan->last_completion ==
ioat_chan->watchdog_completion)
/* has TCP stuck on one cookie since last watchdog? */
&& (ioat_chan->watchdog_tcp_cookie ==
ioat_chan->watchdog_last_tcp_cookie)
&& (ioat_chan->watchdog_tcp_cookie !=
ioat_chan->completed_cookie)
/* is there something in the chain to be processed? */
/* CB1 chain always has at least the last one processed */
&& (ioat_chan->used_desc.prev != ioat_chan->used_desc.next)
&& ioat_chan->pending == 0) {
/*
* check CHANSTS register for completed
* descriptor address.
* if it is different than completion writeback,
* it is not zero
* and it has changed since the last watchdog
* we can assume that channel
* is still working correctly
* and the problem is in completion writeback.
* update completion writeback
* with actual CHANSTS value
* else
* try resetting the channel
*/
completion_hw.low = readl(ioat_chan->reg_base +
IOAT_CHANSTS_OFFSET_LOW(ioat_chan->device->version));
completion_hw.high = readl(ioat_chan->reg_base +
IOAT_CHANSTS_OFFSET_HIGH(ioat_chan->device->version));
#if (BITS_PER_LONG == 64)
compl_desc_addr_hw =
completion_hw.full
& IOAT_CHANSTS_COMPLETED_DESCRIPTOR_ADDR;
#else
compl_desc_addr_hw =
completion_hw.low & IOAT_LOW_COMPLETION_MASK;
#endif
if ((compl_desc_addr_hw != 0)
&& (compl_desc_addr_hw != ioat_chan->watchdog_completion)
&& (compl_desc_addr_hw != ioat_chan->last_compl_desc_addr_hw)) {
ioat_chan->last_compl_desc_addr_hw = compl_desc_addr_hw;
ioat_chan->completion_virt->low = completion_hw.low;
ioat_chan->completion_virt->high = completion_hw.high;
} else {
ioat_dma_reset_channel(ioat_chan);
ioat_chan->watchdog_completion = 0;
ioat_chan->last_compl_desc_addr_hw = 0;
}
/*
* for version 2.0 if there are descriptors yet to be processed
* and the last completed hasn't changed since the last watchdog
* if they haven't hit the pending level
* issue the pending to push them through
* else
* try resetting the channel
*/
} else if (ioat_chan->device->version == IOAT_VER_2_0
&& ioat_chan->used_desc.prev
&& ioat_chan->last_completion
&& ioat_chan->last_completion == ioat_chan->watchdog_completion) {
if (ioat_chan->pending < ioat_pending_level)
ioat2_dma_memcpy_issue_pending(&ioat_chan->common);
else {
ioat_dma_reset_channel(ioat_chan);
ioat_chan->watchdog_completion = 0;
}
} else {
ioat_chan->last_compl_desc_addr_hw = 0;
ioat_chan->watchdog_completion
= ioat_chan->last_completion;
}
ioat_chan->watchdog_last_tcp_cookie =
ioat_chan->watchdog_tcp_cookie;
}
schedule_delayed_work(&device->work, WATCHDOG_DELAY);
}
static dma_cookie_t ioat1_tx_submit(struct dma_async_tx_descriptor *tx)
{
struct ioat_dma_chan *ioat_chan = to_ioat_chan(tx->chan);
struct ioat_desc_sw *first = tx_to_ioat_desc(tx);
struct ioat_desc_sw *prev, *new;
struct ioat_dma_descriptor *hw;
dma_cookie_t cookie;
LIST_HEAD(new_chain);
u32 copy;
size_t len;
dma_addr_t src, dst;
unsigned long orig_flags;
unsigned int desc_count = 0;
/* src and dest and len are stored in the initial descriptor */
len = first->len;
src = first->src;
dst = first->dst;
orig_flags = first->async_tx.flags;
new = first;
spin_lock_bh(&ioat_chan->desc_lock);
prev = to_ioat_desc(ioat_chan->used_desc.prev);
prefetch(prev->hw);
do {
copy = min_t(size_t, len, ioat_chan->xfercap);
async_tx_ack(&new->async_tx);
hw = new->hw;
hw->size = copy;
hw->ctl = 0;
hw->src_addr = src;
hw->dst_addr = dst;
hw->next = 0;
/* chain together the physical address list for the HW */
wmb();
prev->hw->next = (u64) new->async_tx.phys;
len -= copy;
dst += copy;
src += copy;
list_add_tail(&new->node, &new_chain);
desc_count++;
prev = new;
} while (len && (new = ioat1_dma_get_next_descriptor(ioat_chan)));
if (!new) {
dev_err(&ioat_chan->device->pdev->dev,
"tx submit failed\n");
spin_unlock_bh(&ioat_chan->desc_lock);
return -ENOMEM;
}
hw->ctl = IOAT_DMA_DESCRIPTOR_CTL_CP_STS;
if (first->async_tx.callback) {
hw->ctl |= IOAT_DMA_DESCRIPTOR_CTL_INT_GN;
if (first != new) {
/* move callback into to last desc */
new->async_tx.callback = first->async_tx.callback;
new->async_tx.callback_param
= first->async_tx.callback_param;
first->async_tx.callback = NULL;
first->async_tx.callback_param = NULL;
}
}
new->tx_cnt = desc_count;
new->async_tx.flags = orig_flags; /* client is in control of this ack */
/* store the original values for use in later cleanup */
if (new != first) {
new->src = first->src;
new->dst = first->dst;
new->len = first->len;
}
/* cookie incr and addition to used_list must be atomic */
cookie = ioat_chan->common.cookie;
cookie++;
if (cookie < 0)
cookie = 1;
ioat_chan->common.cookie = new->async_tx.cookie = cookie;
/* write address into NextDescriptor field of last desc in chain */
to_ioat_desc(ioat_chan->used_desc.prev)->hw->next =
first->async_tx.phys;
list_splice_tail(&new_chain, &ioat_chan->used_desc);
ioat_chan->dmacount += desc_count;
ioat_chan->pending += desc_count;
if (ioat_chan->pending >= ioat_pending_level)
__ioat1_dma_memcpy_issue_pending(ioat_chan);
spin_unlock_bh(&ioat_chan->desc_lock);
return cookie;
}
static dma_cookie_t ioat2_tx_submit(struct dma_async_tx_descriptor *tx)
{
struct ioat_dma_chan *ioat_chan = to_ioat_chan(tx->chan);
struct ioat_desc_sw *first = tx_to_ioat_desc(tx);
struct ioat_desc_sw *new;
struct ioat_dma_descriptor *hw;
dma_cookie_t cookie;
u32 copy;
size_t len;
dma_addr_t src, dst;
unsigned long orig_flags;
unsigned int desc_count = 0;
/* src and dest and len are stored in the initial descriptor */
len = first->len;
src = first->src;
dst = first->dst;
orig_flags = first->async_tx.flags;
new = first;
/*
* ioat_chan->desc_lock is still in force in version 2 path
* it gets unlocked at end of this function
*/
do {
copy = min_t(size_t, len, ioat_chan->xfercap);
async_tx_ack(&new->async_tx);
hw = new->hw;
hw->size = copy;
hw->ctl = 0;
hw->src_addr = src;
hw->dst_addr = dst;
len -= copy;
dst += copy;
src += copy;
desc_count++;
} while (len && (new = ioat2_dma_get_next_descriptor(ioat_chan)));
if (!new) {
dev_err(&ioat_chan->device->pdev->dev,
"tx submit failed\n");
spin_unlock_bh(&ioat_chan->desc_lock);
return -ENOMEM;
}
hw->ctl |= IOAT_DMA_DESCRIPTOR_CTL_CP_STS;
if (first->async_tx.callback) {
hw->ctl |= IOAT_DMA_DESCRIPTOR_CTL_INT_GN;
if (first != new) {
/* move callback into to last desc */
new->async_tx.callback = first->async_tx.callback;
new->async_tx.callback_param
= first->async_tx.callback_param;
first->async_tx.callback = NULL;
first->async_tx.callback_param = NULL;
}
}
new->tx_cnt = desc_count;
new->async_tx.flags = orig_flags; /* client is in control of this ack */
/* store the original values for use in later cleanup */
if (new != first) {
new->src = first->src;
new->dst = first->dst;
new->len = first->len;
}
/* cookie incr and addition to used_list must be atomic */
cookie = ioat_chan->common.cookie;
cookie++;
if (cookie < 0)
cookie = 1;
ioat_chan->common.cookie = new->async_tx.cookie = cookie;
ioat_chan->dmacount += desc_count;
ioat_chan->pending += desc_count;
if (ioat_chan->pending >= ioat_pending_level)
__ioat2_dma_memcpy_issue_pending(ioat_chan);
spin_unlock_bh(&ioat_chan->desc_lock);
return cookie;
}
/**
* ioat_dma_alloc_descriptor - allocate and return a sw and hw descriptor pair
* @ioat_chan: the channel supplying the memory pool for the descriptors
* @flags: allocation flags
*/
static struct ioat_desc_sw *ioat_dma_alloc_descriptor(
struct ioat_dma_chan *ioat_chan,
gfp_t flags)
{
struct ioat_dma_descriptor *desc;
struct ioat_desc_sw *desc_sw;
struct ioatdma_device *ioatdma_device;
dma_addr_t phys;
ioatdma_device = to_ioatdma_device(ioat_chan->common.device);
desc = pci_pool_alloc(ioatdma_device->dma_pool, flags, &phys);
if (unlikely(!desc))
return NULL;
desc_sw = kzalloc(sizeof(*desc_sw), flags);
if (unlikely(!desc_sw)) {
pci_pool_free(ioatdma_device->dma_pool, desc, phys);
return NULL;
}
memset(desc, 0, sizeof(*desc));
dma_async_tx_descriptor_init(&desc_sw->async_tx, &ioat_chan->common);
switch (ioat_chan->device->version) {
case IOAT_VER_1_2:
desc_sw->async_tx.tx_submit = ioat1_tx_submit;
break;
case IOAT_VER_2_0:
case IOAT_VER_3_0:
desc_sw->async_tx.tx_submit = ioat2_tx_submit;
break;
}
desc_sw->hw = desc;
desc_sw->async_tx.phys = phys;
return desc_sw;
}
static int ioat_initial_desc_count = 256;
module_param(ioat_initial_desc_count, int, 0644);
MODULE_PARM_DESC(ioat_initial_desc_count,
"initial descriptors per channel (default: 256)");
/**
* ioat2_dma_massage_chan_desc - link the descriptors into a circle
* @ioat_chan: the channel to be massaged
*/
static void ioat2_dma_massage_chan_desc(struct ioat_dma_chan *ioat_chan)
{
struct ioat_desc_sw *desc, *_desc;
/* setup used_desc */
ioat_chan->used_desc.next = ioat_chan->free_desc.next;
ioat_chan->used_desc.prev = NULL;
/* pull free_desc out of the circle so that every node is a hw
* descriptor, but leave it pointing to the list
*/
ioat_chan->free_desc.prev->next = ioat_chan->free_desc.next;
ioat_chan->free_desc.next->prev = ioat_chan->free_desc.prev;
/* circle link the hw descriptors */
desc = to_ioat_desc(ioat_chan->free_desc.next);
desc->hw->next = to_ioat_desc(desc->node.next)->async_tx.phys;
list_for_each_entry_safe(desc, _desc, ioat_chan->free_desc.next, node) {
desc->hw->next = to_ioat_desc(desc->node.next)->async_tx.phys;
}
}
/**
* ioat_dma_alloc_chan_resources - returns the number of allocated descriptors
* @chan: the channel to be filled out
*/
static int ioat_dma_alloc_chan_resources(struct dma_chan *chan)
{
struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);
struct ioat_desc_sw *desc;
u16 chanctrl;
u32 chanerr;
int i;
LIST_HEAD(tmp_list);
/* have we already been set up? */
if (!list_empty(&ioat_chan->free_desc))
return ioat_chan->desccount;
/* Setup register to interrupt and write completion status on error */
chanctrl = IOAT_CHANCTRL_ERR_INT_EN |
IOAT_CHANCTRL_ANY_ERR_ABORT_EN |
IOAT_CHANCTRL_ERR_COMPLETION_EN;
writew(chanctrl, ioat_chan->reg_base + IOAT_CHANCTRL_OFFSET);
chanerr = readl(ioat_chan->reg_base + IOAT_CHANERR_OFFSET);
if (chanerr) {
dev_err(&ioat_chan->device->pdev->dev,
"CHANERR = %x, clearing\n", chanerr);
writel(chanerr, ioat_chan->reg_base + IOAT_CHANERR_OFFSET);
}
/* Allocate descriptors */
for (i = 0; i < ioat_initial_desc_count; i++) {
desc = ioat_dma_alloc_descriptor(ioat_chan, GFP_KERNEL);
if (!desc) {
dev_err(&ioat_chan->device->pdev->dev,
"Only %d initial descriptors\n", i);
break;
}
list_add_tail(&desc->node, &tmp_list);
}
spin_lock_bh(&ioat_chan->desc_lock);
ioat_chan->desccount = i;
list_splice(&tmp_list, &ioat_chan->free_desc);
if (ioat_chan->device->version != IOAT_VER_1_2)
ioat2_dma_massage_chan_desc(ioat_chan);
spin_unlock_bh(&ioat_chan->desc_lock);
/* allocate a completion writeback area */
/* doing 2 32bit writes to mmio since 1 64b write doesn't work */
ioat_chan->completion_virt =
pci_pool_alloc(ioat_chan->device->completion_pool,
GFP_KERNEL,
&ioat_chan->completion_addr);
memset(ioat_chan->completion_virt, 0,
sizeof(*ioat_chan->completion_virt));
writel(((u64) ioat_chan->completion_addr) & 0x00000000FFFFFFFF,
ioat_chan->reg_base + IOAT_CHANCMP_OFFSET_LOW);
writel(((u64) ioat_chan->completion_addr) >> 32,
ioat_chan->reg_base + IOAT_CHANCMP_OFFSET_HIGH);
tasklet_enable(&ioat_chan->cleanup_task);
ioat_dma_start_null_desc(ioat_chan); /* give chain to dma device */
return ioat_chan->desccount;
}
/**
* ioat_dma_free_chan_resources - release all the descriptors
* @chan: the channel to be cleaned
*/
static void ioat_dma_free_chan_resources(struct dma_chan *chan)
{
struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);
struct ioatdma_device *ioatdma_device = to_ioatdma_device(chan->device);
struct ioat_desc_sw *desc, *_desc;
int in_use_descs = 0;
/* Before freeing channel resources first check
* if they have been previously allocated for this channel.
*/
if (ioat_chan->desccount == 0)
return;
tasklet_disable(&ioat_chan->cleanup_task);
ioat_dma_memcpy_cleanup(ioat_chan);
/* Delay 100ms after reset to allow internal DMA logic to quiesce
* before removing DMA descriptor resources.
*/
writeb(IOAT_CHANCMD_RESET,
ioat_chan->reg_base
+ IOAT_CHANCMD_OFFSET(ioat_chan->device->version));
mdelay(100);
spin_lock_bh(&ioat_chan->desc_lock);
switch (ioat_chan->device->version) {
case IOAT_VER_1_2:
list_for_each_entry_safe(desc, _desc,
&ioat_chan->used_desc, node) {
in_use_descs++;
list_del(&desc->node);
pci_pool_free(ioatdma_device->dma_pool, desc->hw,
desc->async_tx.phys);
kfree(desc);
}
list_for_each_entry_safe(desc, _desc,
&ioat_chan->free_desc, node) {
list_del(&desc->node);
pci_pool_free(ioatdma_device->dma_pool, desc->hw,
desc->async_tx.phys);
kfree(desc);
}
break;
case IOAT_VER_2_0:
case IOAT_VER_3_0:
list_for_each_entry_safe(desc, _desc,
ioat_chan->free_desc.next, node) {
list_del(&desc->node);
pci_pool_free(ioatdma_device->dma_pool, desc->hw,
desc->async_tx.phys);
kfree(desc);
}
desc = to_ioat_desc(ioat_chan->free_desc.next);
pci_pool_free(ioatdma_device->dma_pool, desc->hw,
desc->async_tx.phys);
kfree(desc);
INIT_LIST_HEAD(&ioat_chan->free_desc);
INIT_LIST_HEAD(&ioat_chan->used_desc);
break;
}
spin_unlock_bh(&ioat_chan->desc_lock);
pci_pool_free(ioatdma_device->completion_pool,
ioat_chan->completion_virt,
ioat_chan->completion_addr);
/* one is ok since we left it on there on purpose */
if (in_use_descs > 1)
dev_err(&ioat_chan->device->pdev->dev,
"Freeing %d in use descriptors!\n",
in_use_descs - 1);
ioat_chan->last_completion = ioat_chan->completion_addr = 0;
ioat_chan->pending = 0;
ioat_chan->dmacount = 0;
ioat_chan->desccount = 0;
ioat_chan->watchdog_completion = 0;
ioat_chan->last_compl_desc_addr_hw = 0;
ioat_chan->watchdog_tcp_cookie =
ioat_chan->watchdog_last_tcp_cookie = 0;
}
/**
* ioat_dma_get_next_descriptor - return the next available descriptor
* @ioat_chan: IOAT DMA channel handle
*
* Gets the next descriptor from the chain, and must be called with the
* channel's desc_lock held. Allocates more descriptors if the channel
* has run out.
*/
static struct ioat_desc_sw *
ioat1_dma_get_next_descriptor(struct ioat_dma_chan *ioat_chan)
{
struct ioat_desc_sw *new;
if (!list_empty(&ioat_chan->free_desc)) {
new = to_ioat_desc(ioat_chan->free_desc.next);
list_del(&new->node);
} else {
/* try to get another desc */
new = ioat_dma_alloc_descriptor(ioat_chan, GFP_ATOMIC);
if (!new) {
dev_err(&ioat_chan->device->pdev->dev,
"alloc failed\n");
return NULL;
}
}
prefetch(new->hw);
return new;
}
static struct ioat_desc_sw *
ioat2_dma_get_next_descriptor(struct ioat_dma_chan *ioat_chan)
{
struct ioat_desc_sw *new;
/*
* used.prev points to where to start processing
* used.next points to next free descriptor
* if used.prev == NULL, there are none waiting to be processed
* if used.next == used.prev.prev, there is only one free descriptor,
* and we need to use it to as a noop descriptor before
* linking in a new set of descriptors, since the device
* has probably already read the pointer to it
*/
if (ioat_chan->used_desc.prev &&
ioat_chan->used_desc.next == ioat_chan->used_desc.prev->prev) {
struct ioat_desc_sw *desc;
struct ioat_desc_sw *noop_desc;
int i;
/* set up the noop descriptor */
noop_desc = to_ioat_desc(ioat_chan->used_desc.next);
/* set size to non-zero value (channel returns error when size is 0) */
noop_desc->hw->size = NULL_DESC_BUFFER_SIZE;
noop_desc->hw->ctl = IOAT_DMA_DESCRIPTOR_NUL;
noop_desc->hw->src_addr = 0;
noop_desc->hw->dst_addr = 0;
ioat_chan->used_desc.next = ioat_chan->used_desc.next->next;
ioat_chan->pending++;
ioat_chan->dmacount++;
/* try to get a few more descriptors */
for (i = 16; i; i--) {
desc = ioat_dma_alloc_descriptor(ioat_chan, GFP_ATOMIC);
if (!desc) {
dev_err(&ioat_chan->device->pdev->dev,
"alloc failed\n");
break;
}
list_add_tail(&desc->node, ioat_chan->used_desc.next);
desc->hw->next
= to_ioat_desc(desc->node.next)->async_tx.phys;
to_ioat_desc(desc->node.prev)->hw->next
= desc->async_tx.phys;
ioat_chan->desccount++;
}
ioat_chan->used_desc.next = noop_desc->node.next;
}
new = to_ioat_desc(ioat_chan->used_desc.next);
prefetch(new);
ioat_chan->used_desc.next = new->node.next;
if (ioat_chan->used_desc.prev == NULL)
ioat_chan->used_desc.prev = &new->node;
prefetch(new->hw);
return new;
}
static struct ioat_desc_sw *ioat_dma_get_next_descriptor(
struct ioat_dma_chan *ioat_chan)
{
if (!ioat_chan)
return NULL;
switch (ioat_chan->device->version) {
case IOAT_VER_1_2:
return ioat1_dma_get_next_descriptor(ioat_chan);
case IOAT_VER_2_0:
case IOAT_VER_3_0:
return ioat2_dma_get_next_descriptor(ioat_chan);
}
return NULL;
}
static struct dma_async_tx_descriptor *ioat1_dma_prep_memcpy(
struct dma_chan *chan,
dma_addr_t dma_dest,
dma_addr_t dma_src,
size_t len,
unsigned long flags)
{