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
|
/*-
* Copyright (c) 2013 The FreeBSD Foundation
* All rights reserved.
*
* This software was developed by Konstantin Belousov <kib@FreeBSD.org>
* under sponsorship from the FreeBSD Foundation.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/bus.h>
#include <sys/conf.h>
#include <sys/interrupt.h>
#include <sys/kernel.h>
#include <sys/ktr.h>
#include <sys/lock.h>
#include <sys/proc.h>
#include <sys/memdesc.h>
#include <sys/mutex.h>
#include <sys/sysctl.h>
#include <sys/rman.h>
#include <sys/taskqueue.h>
#include <sys/tree.h>
#include <sys/uio.h>
#include <dev/pci/pcireg.h>
#include <dev/pci/pcivar.h>
#include <vm/vm.h>
#include <vm/vm_extern.h>
#include <vm/vm_kern.h>
#include <vm/vm_object.h>
#include <vm/vm_page.h>
#include <vm/vm_map.h>
#include <machine/atomic.h>
#include <machine/bus.h>
#include <machine/md_var.h>
#include <machine/specialreg.h>
#include <x86/include/busdma_impl.h>
#include <x86/iommu/intel_reg.h>
#include <x86/iommu/busdma_dmar.h>
#include <x86/iommu/intel_dmar.h>
/*
* busdma_dmar.c, the implementation of the busdma(9) interface using
* DMAR units from Intel VT-d.
*/
static bool
dmar_bus_dma_is_dev_disabled(int domain, int bus, int slot, int func)
{
char str[128], *env;
snprintf(str, sizeof(str), "hw.busdma.pci%d.%d.%d.%d.bounce",
domain, bus, slot, func);
env = getenv(str);
if (env == NULL)
return (false);
freeenv(env);
return (true);
}
/*
* Given original device, find the requester ID that will be seen by
* the DMAR unit and used for page table lookup. PCI bridges may take
* ownership of transactions from downstream devices, so it may not be
* the same as the BSF of the target device. In those cases, all
* devices downstream of the bridge must share a single mapping
* domain, and must collectively be assigned to use either DMAR or
* bounce mapping.
*/
static device_t
dmar_get_requester(device_t dev, int *bus, int *slot, int *func)
{
devclass_t pci_class;
device_t pci, pcib, requester;
int cap_offset;
pci_class = devclass_find("pci");
requester = dev;
*bus = pci_get_bus(dev);
*slot = pci_get_slot(dev);
*func = pci_get_function(dev);
/*
* Walk the bridge hierarchy from the target device to the
* host port to find the translating bridge nearest the DMAR
* unit.
*/
for (;;) {
pci = device_get_parent(dev);
KASSERT(pci != NULL, ("NULL parent for pci%d:%d:%d:%d",
pci_get_domain(dev), pci_get_bus(dev), pci_get_slot(dev),
pci_get_function(dev)));
KASSERT(device_get_devclass(pci) == pci_class,
("Non-pci parent for pci%d:%d:%d:%d",
pci_get_domain(dev), pci_get_bus(dev), pci_get_slot(dev),
pci_get_function(dev)));
pcib = device_get_parent(pci);
KASSERT(pcib != NULL, ("NULL bridge for pci%d:%d:%d:%d",
pci_get_domain(dev), pci_get_bus(dev), pci_get_slot(dev),
pci_get_function(dev)));
/*
* The parent of our "bridge" isn't another PCI bus,
* so pcib isn't a PCI->PCI bridge but rather a host
* port, and the requester ID won't be translated
* further.
*/
if (device_get_devclass(device_get_parent(pcib)) != pci_class)
break;
if (pci_find_cap(dev, PCIY_EXPRESS, &cap_offset) != 0) {
/*
* Device is not PCIe, it cannot be seen as a
* requester by DMAR unit.
*/
requester = pcib;
/* Check whether the bus above is PCIe. */
if (pci_find_cap(pcib, PCIY_EXPRESS,
&cap_offset) == 0) {
/*
* The current device is not PCIe, but
* the bridge above it is. This is a
* PCIe->PCI bridge. Assume that the
* requester ID will be the secondary
* bus number with slot and function
* set to zero.
*
* XXX: Doesn't handle the case where
* the bridge is PCIe->PCI-X, and the
* bridge will only take ownership of
* requests in some cases. We should
* provide context entries with the
* same page tables for taken and
* non-taken transactions.
*/
*bus = pci_get_bus(dev);
*slot = *func = 0;
} else {
/*
* Neither the device nor the bridge
* above it are PCIe. This is a
* conventional PCI->PCI bridge, which
* will use the bridge's BSF as the
* requester ID.
*/
*bus = pci_get_bus(pcib);
*slot = pci_get_slot(pcib);
*func = pci_get_function(pcib);
}
}
/*
* Do not stop the loop even if the target device is
* PCIe, because it is possible (but unlikely) to have
* a PCI->PCIe bridge somewhere in the hierarchy.
*/
dev = pcib;
}
return (requester);
}
struct dmar_ctx *
dmar_instantiate_ctx(struct dmar_unit *dmar, device_t dev, bool rmrr)
{
device_t requester;
struct dmar_ctx *ctx;
bool disabled;
int bus, slot, func;
requester = dmar_get_requester(dev, &bus, &slot, &func);
/*
* If the user requested the IOMMU disabled for the device, we
* cannot disable the DMAR, due to possibility of other
* devices on the same DMAR still requiring translation.
* Instead provide the identity mapping for the device
* context.
*/
disabled = dmar_bus_dma_is_dev_disabled(pci_get_domain(dev), bus,
slot, func);
ctx = dmar_get_ctx(dmar, requester, bus, slot, func, disabled, rmrr);
if (ctx == NULL)
return (NULL);
if (disabled) {
/*
* Keep the first reference on context, release the
* later refs.
*/
DMAR_LOCK(dmar);
if ((ctx->flags & DMAR_CTX_DISABLED) == 0) {
ctx->flags |= DMAR_CTX_DISABLED;
DMAR_UNLOCK(dmar);
} else {
dmar_free_ctx_locked(dmar, ctx);
}
ctx = NULL;
}
return (ctx);
}
bus_dma_tag_t
dmar_get_dma_tag(device_t dev, device_t child)
{
struct dmar_unit *dmar;
struct dmar_ctx *ctx;
bus_dma_tag_t res;
dmar = dmar_find(child);
/* Not in scope of any DMAR ? */
if (dmar == NULL)
return (NULL);
dmar_quirks_pre_use(dmar);
dmar_instantiate_rmrr_ctxs(dmar);
ctx = dmar_instantiate_ctx(dmar, child, false);
res = ctx == NULL ? NULL : (bus_dma_tag_t)&ctx->ctx_tag;
return (res);
}
static MALLOC_DEFINE(M_DMAR_DMAMAP, "dmar_dmamap", "Intel DMAR DMA Map");
static void dmar_bus_schedule_dmamap(struct dmar_unit *unit,
struct bus_dmamap_dmar *map);
static int
dmar_bus_dma_tag_create(bus_dma_tag_t parent, bus_size_t alignment,
bus_addr_t boundary, bus_addr_t lowaddr, bus_addr_t highaddr,
bus_dma_filter_t *filter, void *filterarg, bus_size_t maxsize,
int nsegments, bus_size_t maxsegsz, int flags, bus_dma_lock_t *lockfunc,
void *lockfuncarg, bus_dma_tag_t *dmat)
{
struct bus_dma_tag_dmar *newtag, *oldtag;
int error;
*dmat = NULL;
error = common_bus_dma_tag_create(parent != NULL ?
&((struct bus_dma_tag_dmar *)parent)->common : NULL, alignment,
boundary, lowaddr, highaddr, filter, filterarg, maxsize,
nsegments, maxsegsz, flags, lockfunc, lockfuncarg,
sizeof(struct bus_dma_tag_dmar), (void **)&newtag);
if (error != 0)
goto out;
oldtag = (struct bus_dma_tag_dmar *)parent;
newtag->common.impl = &bus_dma_dmar_impl;
newtag->ctx = oldtag->ctx;
newtag->owner = oldtag->owner;
*dmat = (bus_dma_tag_t)newtag;
out:
CTR4(KTR_BUSDMA, "%s returned tag %p tag flags 0x%x error %d",
__func__, newtag, (newtag != NULL ? newtag->common.flags : 0),
error);
return (error);
}
static int
dmar_bus_dma_tag_destroy(bus_dma_tag_t dmat1)
{
struct bus_dma_tag_dmar *dmat, *dmat_copy, *parent;
int error;
error = 0;
dmat_copy = dmat = (struct bus_dma_tag_dmar *)dmat1;
if (dmat != NULL) {
if (dmat->map_count != 0) {
error = EBUSY;
goto out;
}
while (dmat != NULL) {
parent = (struct bus_dma_tag_dmar *)dmat->common.parent;
if (atomic_fetchadd_int(&dmat->common.ref_count, -1) ==
1) {
if (dmat == &dmat->ctx->ctx_tag)
dmar_free_ctx(dmat->ctx);
free(dmat->segments, M_DMAR_DMAMAP);
free(dmat, M_DEVBUF);
dmat = parent;
} else
dmat = NULL;
}
}
out:
CTR3(KTR_BUSDMA, "%s tag %p error %d", __func__, dmat_copy, error);
return (error);
}
static int
dmar_bus_dmamap_create(bus_dma_tag_t dmat, int flags, bus_dmamap_t *mapp)
{
struct bus_dma_tag_dmar *tag;
struct bus_dmamap_dmar *map;
tag = (struct bus_dma_tag_dmar *)dmat;
map = malloc(sizeof(*map), M_DMAR_DMAMAP, M_NOWAIT | M_ZERO);
if (map == NULL) {
*mapp = NULL;
return (ENOMEM);
}
if (tag->segments == NULL) {
tag->segments = malloc(sizeof(bus_dma_segment_t) *
tag->common.nsegments, M_DMAR_DMAMAP, M_NOWAIT);
if (tag->segments == NULL) {
free(map, M_DMAR_DMAMAP);
*mapp = NULL;
return (ENOMEM);
}
}
TAILQ_INIT(&map->map_entries);
map->tag = tag;
map->locked = true;
map->cansleep = false;
tag->map_count++;
*mapp = (bus_dmamap_t)map;
return (0);
}
static int
dmar_bus_dmamap_destroy(bus_dma_tag_t dmat, bus_dmamap_t map1)
{
struct bus_dma_tag_dmar *tag;
struct bus_dmamap_dmar *map;
tag = (struct bus_dma_tag_dmar *)dmat;
map = (struct bus_dmamap_dmar *)map1;
if (map != NULL) {
DMAR_CTX_LOCK(tag->ctx);
if (!TAILQ_EMPTY(&map->map_entries)) {
DMAR_CTX_UNLOCK(tag->ctx);
return (EBUSY);
}
DMAR_CTX_UNLOCK(tag->ctx);
free(map, M_DMAR_DMAMAP);
}
tag->map_count--;
return (0);
}
static int
dmar_bus_dmamem_alloc(bus_dma_tag_t dmat, void** vaddr, int flags,
bus_dmamap_t *mapp)
{
struct bus_dma_tag_dmar *tag;
struct bus_dmamap_dmar *map;
int error, mflags;
vm_memattr_t attr;
error = dmar_bus_dmamap_create(dmat, flags, mapp);
if (error != 0)
return (error);
mflags = (flags & BUS_DMA_NOWAIT) != 0 ? M_NOWAIT : M_WAITOK;
mflags |= (flags & BUS_DMA_ZERO) != 0 ? M_ZERO : 0;
attr = (flags & BUS_DMA_NOCACHE) != 0 ? VM_MEMATTR_UNCACHEABLE :
VM_MEMATTR_DEFAULT;
tag = (struct bus_dma_tag_dmar *)dmat;
map = (struct bus_dmamap_dmar *)*mapp;
if (tag->common.maxsize < PAGE_SIZE &&
tag->common.alignment <= tag->common.maxsize &&
attr == VM_MEMATTR_DEFAULT) {
*vaddr = malloc(tag->common.maxsize, M_DEVBUF, mflags);
map->flags |= BUS_DMAMAP_DMAR_MALLOC;
} else {
*vaddr = (void *)kmem_alloc_attr(kernel_arena,
tag->common.maxsize, mflags, 0ul, BUS_SPACE_MAXADDR,
attr);
map->flags |= BUS_DMAMAP_DMAR_KMEM_ALLOC;
}
if (*vaddr == NULL) {
dmar_bus_dmamap_destroy(dmat, *mapp);
*mapp = NULL;
return (ENOMEM);
}
return (0);
}
static void
dmar_bus_dmamem_free(bus_dma_tag_t dmat, void *vaddr, bus_dmamap_t map1)
{
struct bus_dma_tag_dmar *tag;
struct bus_dmamap_dmar *map;
tag = (struct bus_dma_tag_dmar *)dmat;
map = (struct bus_dmamap_dmar *)map1;
if ((map->flags & BUS_DMAMAP_DMAR_MALLOC) != 0) {
free(vaddr, M_DEVBUF);
map->flags &= ~BUS_DMAMAP_DMAR_MALLOC;
} else {
KASSERT((map->flags & BUS_DMAMAP_DMAR_KMEM_ALLOC) != 0,
("dmar_bus_dmamem_free for non alloced map %p", map));
kmem_free(kernel_arena, (vm_offset_t)vaddr, tag->common.maxsize);
map->flags &= ~BUS_DMAMAP_DMAR_KMEM_ALLOC;
}
dmar_bus_dmamap_destroy(dmat, map1);
}
static int
dmar_bus_dmamap_load_something1(struct bus_dma_tag_dmar *tag,
struct bus_dmamap_dmar *map, vm_page_t *ma, int offset, bus_size_t buflen,
int flags, bus_dma_segment_t *segs, int *segp,
struct dmar_map_entries_tailq *unroll_list)
{
struct dmar_ctx *ctx;
struct dmar_map_entry *entry;
dmar_gaddr_t size;
bus_size_t buflen1;
int error, idx, gas_flags, seg;
if (segs == NULL)
segs = tag->segments;
ctx = tag->ctx;
seg = *segp;
error = 0;
idx = 0;
while (buflen > 0) {
seg++;
if (seg >= tag->common.nsegments) {
error = EFBIG;
break;
}
buflen1 = buflen > tag->common.maxsegsz ?
tag->common.maxsegsz : buflen;
buflen -= buflen1;
size = round_page(offset + buflen1);
/*
* (Too) optimistically allow split if there are more
* then one segments left.
*/
gas_flags = map->cansleep ? DMAR_GM_CANWAIT : 0;
if (seg + 1 < tag->common.nsegments)
gas_flags |= DMAR_GM_CANSPLIT;
error = dmar_gas_map(ctx, &tag->common, size,
DMAR_MAP_ENTRY_READ | DMAR_MAP_ENTRY_WRITE,
gas_flags, ma + idx, &entry);
if (error != 0)
break;
if ((gas_flags & DMAR_GM_CANSPLIT) != 0) {
KASSERT(size >= entry->end - entry->start,
("split increased entry size %jx %jx %jx",
(uintmax_t)size, (uintmax_t)entry->start,
(uintmax_t)entry->end));
size = entry->end - entry->start;
if (buflen1 > size)
buflen1 = size;
} else {
KASSERT(entry->end - entry->start == size,
("no split allowed %jx %jx %jx",
(uintmax_t)size, (uintmax_t)entry->start,
(uintmax_t)entry->end));
}
KASSERT(((entry->start + offset) & (tag->common.alignment - 1))
== 0,
("alignment failed: ctx %p start 0x%jx offset %x "
"align 0x%jx", ctx, (uintmax_t)entry->start, offset,
(uintmax_t)tag->common.alignment));
KASSERT(entry->end <= tag->common.lowaddr ||
entry->start >= tag->common.highaddr,
("entry placement failed: ctx %p start 0x%jx end 0x%jx "
"lowaddr 0x%jx highaddr 0x%jx", ctx,
(uintmax_t)entry->start, (uintmax_t)entry->end,
(uintmax_t)tag->common.lowaddr,
(uintmax_t)tag->common.highaddr));
KASSERT(dmar_test_boundary(entry->start, entry->end -
entry->start, tag->common.boundary),
("boundary failed: ctx %p start 0x%jx end 0x%jx "
"boundary 0x%jx", ctx, (uintmax_t)entry->start,
(uintmax_t)entry->end, (uintmax_t)tag->common.boundary));
KASSERT(buflen1 <= tag->common.maxsegsz,
("segment too large: ctx %p start 0x%jx end 0x%jx "
"maxsegsz 0x%jx", ctx, (uintmax_t)entry->start,
(uintmax_t)entry->end, (uintmax_t)tag->common.maxsegsz));
DMAR_CTX_LOCK(ctx);
TAILQ_INSERT_TAIL(&map->map_entries, entry, dmamap_link);
entry->flags |= DMAR_MAP_ENTRY_MAP;
DMAR_CTX_UNLOCK(ctx);
TAILQ_INSERT_TAIL(unroll_list, entry, unroll_link);
segs[seg].ds_addr = entry->start + offset;
segs[seg].ds_len = buflen1;
idx += OFF_TO_IDX(trunc_page(offset + buflen1));
offset += buflen1;
offset &= DMAR_PAGE_MASK;
}
if (error == 0)
*segp = seg;
return (error);
}
static int
dmar_bus_dmamap_load_something(struct bus_dma_tag_dmar *tag,
struct bus_dmamap_dmar *map, vm_page_t *ma, int offset, bus_size_t buflen,
int flags, bus_dma_segment_t *segs, int *segp)
{
struct dmar_ctx *ctx;
struct dmar_map_entry *entry, *entry1;
struct dmar_map_entries_tailq unroll_list;
int error;
ctx = tag->ctx;
atomic_add_long(&ctx->loads, 1);
TAILQ_INIT(&unroll_list);
error = dmar_bus_dmamap_load_something1(tag, map, ma, offset,
buflen, flags, segs, segp, &unroll_list);
if (error != 0) {
/*
* The busdma interface does not allow us to report
* partial buffer load, so unfortunately we have to
* revert all work done.
*/
DMAR_CTX_LOCK(ctx);
TAILQ_FOREACH_SAFE(entry, &unroll_list, unroll_link,
entry1) {
/*
* No entries other than what we have created
* during the failed run might have been
* inserted there in between, since we own ctx
* pglock.
*/
TAILQ_REMOVE(&map->map_entries, entry, dmamap_link);
TAILQ_REMOVE(&unroll_list, entry, unroll_link);
TAILQ_INSERT_TAIL(&ctx->unload_entries, entry,
dmamap_link);
}
DMAR_CTX_UNLOCK(ctx);
taskqueue_enqueue(ctx->dmar->delayed_taskqueue,
&ctx->unload_task);
}
if (error == ENOMEM && (flags & BUS_DMA_NOWAIT) == 0 &&
!map->cansleep)
error = EINPROGRESS;
if (error == EINPROGRESS)
dmar_bus_schedule_dmamap(ctx->dmar, map);
return (error);
}
static int
dmar_bus_dmamap_load_ma(bus_dma_tag_t dmat, bus_dmamap_t map1,
struct vm_page **ma, bus_size_t tlen, int ma_offs, int flags,
bus_dma_segment_t *segs, int *segp)
{
struct bus_dma_tag_dmar *tag;
struct bus_dmamap_dmar *map;
tag = (struct bus_dma_tag_dmar *)dmat;
map = (struct bus_dmamap_dmar *)map1;
return (dmar_bus_dmamap_load_something(tag, map, ma, ma_offs, tlen,
flags, segs, segp));
}
static int
dmar_bus_dmamap_load_phys(bus_dma_tag_t dmat, bus_dmamap_t map1,
vm_paddr_t buf, bus_size_t buflen, int flags, bus_dma_segment_t *segs,
int *segp)
{
struct bus_dma_tag_dmar *tag;
struct bus_dmamap_dmar *map;
vm_page_t *ma;
vm_paddr_t pstart, pend;
int error, i, ma_cnt, offset;
tag = (struct bus_dma_tag_dmar *)dmat;
map = (struct bus_dmamap_dmar *)map1;
pstart = trunc_page(buf);
pend = round_page(buf + buflen);
offset = buf & PAGE_MASK;
ma_cnt = OFF_TO_IDX(pend - pstart);
ma = malloc(sizeof(vm_page_t) * ma_cnt, M_DEVBUF, map->cansleep ?
M_WAITOK : M_NOWAIT);
if (ma == NULL)
return (ENOMEM);
for (i = 0; i < ma_cnt; i++)
ma[i] = PHYS_TO_VM_PAGE(pstart + i * PAGE_SIZE);
error = dmar_bus_dmamap_load_something(tag, map, ma, offset, buflen,
flags, segs, segp);
free(ma, M_DEVBUF);
return (error);
}
static int
dmar_bus_dmamap_load_buffer(bus_dma_tag_t dmat, bus_dmamap_t map1, void *buf,
bus_size_t buflen, pmap_t pmap, int flags, bus_dma_segment_t *segs,
int *segp)
{
struct bus_dma_tag_dmar *tag;
struct bus_dmamap_dmar *map;
vm_page_t *ma, fma;
vm_paddr_t pstart, pend, paddr;
int error, i, ma_cnt, offset;
tag = (struct bus_dma_tag_dmar *)dmat;
map = (struct bus_dmamap_dmar *)map1;
pstart = trunc_page((vm_offset_t)buf);
pend = round_page((vm_offset_t)buf + buflen);
offset = (vm_offset_t)buf & PAGE_MASK;
ma_cnt = OFF_TO_IDX(pend - pstart);
ma = malloc(sizeof(vm_page_t) * ma_cnt, M_DEVBUF, map->cansleep ?
M_WAITOK : M_NOWAIT);
if (ma == NULL)
return (ENOMEM);
if (dumping) {
/*
* If dumping, do not attempt to call
* PHYS_TO_VM_PAGE() at all. It may return non-NULL
* but the vm_page returned might be not initialized,
* e.g. for the kernel itself.
*/
KASSERT(pmap == kernel_pmap, ("non-kernel address write"));
fma = malloc(sizeof(struct vm_page) * ma_cnt, M_DEVBUF,
M_ZERO | (map->cansleep ? M_WAITOK : M_NOWAIT));
if (fma == NULL) {
free(ma, M_DEVBUF);
return (ENOMEM);
}
for (i = 0; i < ma_cnt; i++, pstart += PAGE_SIZE) {
paddr = pmap_kextract(pstart);
vm_page_initfake(&fma[i], paddr, VM_MEMATTR_DEFAULT);
ma[i] = &fma[i];
}
} else {
fma = NULL;
for (i = 0; i < ma_cnt; i++, pstart += PAGE_SIZE) {
if (pmap == kernel_pmap)
paddr = pmap_kextract(pstart);
else
paddr = pmap_extract(pmap, pstart);
ma[i] = PHYS_TO_VM_PAGE(paddr);
KASSERT(VM_PAGE_TO_PHYS(ma[i]) == paddr,
("PHYS_TO_VM_PAGE failed %jx %jx m %p",
(uintmax_t)paddr, (uintmax_t)VM_PAGE_TO_PHYS(ma[i]),
ma[i]));
}
}
error = dmar_bus_dmamap_load_something(tag, map, ma, offset, buflen,
flags, segs, segp);
free(ma, M_DEVBUF);
free(fma, M_DEVBUF);
return (error);
}
static void
dmar_bus_dmamap_waitok(bus_dma_tag_t dmat, bus_dmamap_t map1,
struct memdesc *mem, bus_dmamap_callback_t *callback, void *callback_arg)
{
struct bus_dmamap_dmar *map;
if (map1 == NULL)
return;
map = (struct bus_dmamap_dmar *)map1;
map->mem = *mem;
map->tag = (struct bus_dma_tag_dmar *)dmat;
map->callback = callback;
map->callback_arg = callback_arg;
}
static bus_dma_segment_t *
dmar_bus_dmamap_complete(bus_dma_tag_t dmat, bus_dmamap_t map1,
bus_dma_segment_t *segs, int nsegs, int error)
{
struct bus_dma_tag_dmar *tag;
struct bus_dmamap_dmar *map;
tag = (struct bus_dma_tag_dmar *)dmat;
map = (struct bus_dmamap_dmar *)map1;
if (!map->locked) {
KASSERT(map->cansleep,
("map not locked and not sleepable context %p", map));
/*
* We are called from the delayed context. Relock the
* driver.
*/
(tag->common.lockfunc)(tag->common.lockfuncarg, BUS_DMA_LOCK);
map->locked = true;
}
if (segs == NULL)
segs = tag->segments;
return (segs);
}
/*
* The limitations of busdma KPI forces the dmar to perform the actual
* unload, consisting of the unmapping of the map entries page tables,
* from the delayed context on i386, since page table page mapping
* might require a sleep to be successfull. The unfortunate
* consequence is that the DMA requests can be served some time after
* the bus_dmamap_unload() call returned.
*
* On amd64, we assume that sf allocation cannot fail.
*/
static void
dmar_bus_dmamap_unload(bus_dma_tag_t dmat, bus_dmamap_t map1)
{
struct bus_dma_tag_dmar *tag;
struct bus_dmamap_dmar *map;
struct dmar_ctx *ctx;
#if defined(__amd64__)
struct dmar_map_entries_tailq entries;
#endif
tag = (struct bus_dma_tag_dmar *)dmat;
map = (struct bus_dmamap_dmar *)map1;
ctx = tag->ctx;
atomic_add_long(&ctx->unloads, 1);
#if defined(__i386__)
DMAR_CTX_LOCK(ctx);
TAILQ_CONCAT(&ctx->unload_entries, &map->map_entries, dmamap_link);
DMAR_CTX_UNLOCK(ctx);
taskqueue_enqueue(ctx->dmar->delayed_taskqueue, &ctx->unload_task);
#else /* defined(__amd64__) */
TAILQ_INIT(&entries);
DMAR_CTX_LOCK(ctx);
TAILQ_CONCAT(&entries, &map->map_entries, dmamap_link);
DMAR_CTX_UNLOCK(ctx);
THREAD_NO_SLEEPING();
dmar_ctx_unload(ctx, &entries, false);
THREAD_SLEEPING_OK();
KASSERT(TAILQ_EMPTY(&entries), ("lazy dmar_ctx_unload %p", ctx));
#endif
}
static void
dmar_bus_dmamap_sync(bus_dma_tag_t dmat, bus_dmamap_t map,
bus_dmasync_op_t op)
{
}
struct bus_dma_impl bus_dma_dmar_impl = {
.tag_create = dmar_bus_dma_tag_create,
.tag_destroy = dmar_bus_dma_tag_destroy,
.map_create = dmar_bus_dmamap_create,
.map_destroy = dmar_bus_dmamap_destroy,
.mem_alloc = dmar_bus_dmamem_alloc,
.mem_free = dmar_bus_dmamem_free,
.load_phys = dmar_bus_dmamap_load_phys,
.load_buffer = dmar_bus_dmamap_load_buffer,
.load_ma = dmar_bus_dmamap_load_ma,
.map_waitok = dmar_bus_dmamap_waitok,
.map_complete = dmar_bus_dmamap_complete,
.map_unload = dmar_bus_dmamap_unload,
.map_sync = dmar_bus_dmamap_sync
};
static void
dmar_bus_task_dmamap(void *arg, int pending)
{
struct bus_dma_tag_dmar *tag;
struct bus_dmamap_dmar *map;
struct dmar_unit *unit;
struct dmar_ctx *ctx;
unit = arg;
DMAR_LOCK(unit);
while ((map = TAILQ_FIRST(&unit->delayed_maps)) != NULL) {
TAILQ_REMOVE(&unit->delayed_maps, map, delay_link);
DMAR_UNLOCK(unit);
tag = map->tag;
ctx = map->tag->ctx;
map->cansleep = true;
map->locked = false;
bus_dmamap_load_mem((bus_dma_tag_t)tag, (bus_dmamap_t)map,
&map->mem, map->callback, map->callback_arg,
BUS_DMA_WAITOK);
map->cansleep = false;
if (map->locked) {
(tag->common.lockfunc)(tag->common.lockfuncarg,
BUS_DMA_UNLOCK);
} else
map->locked = true;
map->cansleep = false;
DMAR_LOCK(unit);
}
DMAR_UNLOCK(unit);
}
static void
dmar_bus_schedule_dmamap(struct dmar_unit *unit, struct bus_dmamap_dmar *map)
{
struct dmar_ctx *ctx;
ctx = map->tag->ctx;
map->locked = false;
DMAR_LOCK(unit);
TAILQ_INSERT_TAIL(&unit->delayed_maps, map, delay_link);
DMAR_UNLOCK(unit);
taskqueue_enqueue(unit->delayed_taskqueue, &unit->dmamap_load_task);
}
int
dmar_init_busdma(struct dmar_unit *unit)
{
TAILQ_INIT(&unit->delayed_maps);
TASK_INIT(&unit->dmamap_load_task, 0, dmar_bus_task_dmamap, unit);
unit->delayed_taskqueue = taskqueue_create("dmar", M_WAITOK,
taskqueue_thread_enqueue, &unit->delayed_taskqueue);
taskqueue_start_threads(&unit->delayed_taskqueue, 1, PI_DISK,
"dmar%d busdma taskq", unit->unit);
return (0);
}
void
dmar_fini_busdma(struct dmar_unit *unit)
{
if (unit->delayed_taskqueue == NULL)
return;
taskqueue_drain(unit->delayed_taskqueue, &unit->dmamap_load_task);
taskqueue_free(unit->delayed_taskqueue);
unit->delayed_taskqueue = NULL;
}
|