aboutsummaryrefslogtreecommitdiffstats
path: root/arch/tile/kernel/pci-dma.c
blob: b9fe80ec108989fa341060542816d9760fd62d63 (plain) (blame)
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
/*
 * Copyright 2010 Tilera Corporation. All Rights Reserved.
 *
 *   This program is free software; you can redistribute it and/or
 *   modify it under the terms of the GNU General Public License
 *   as published by the Free Software Foundation, version 2.
 *
 *   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, GOOD TITLE or
 *   NON INFRINGEMENT.  See the GNU General Public License for
 *   more details.
 */

#include <linux/mm.h>
#include <linux/dma-mapping.h>
#include <linux/swiotlb.h>
#include <linux/vmalloc.h>
#include <linux/export.h>
#include <asm/tlbflush.h>
#include <asm/homecache.h>

/* Generic DMA mapping functions: */

/*
 * Allocate what Linux calls "coherent" memory.  On TILEPro this is
 * uncached memory; on TILE-Gx it is hash-for-home memory.
 */
#ifdef __tilepro__
#define PAGE_HOME_DMA PAGE_HOME_UNCACHED
#else
#define PAGE_HOME_DMA PAGE_HOME_HASH
#endif

static void *tile_dma_alloc_coherent(struct device *dev, size_t size,
				     dma_addr_t *dma_handle, gfp_t gfp,
				     struct dma_attrs *attrs)
{
	u64 dma_mask = dev->coherent_dma_mask ?: DMA_BIT_MASK(32);
	int node = dev_to_node(dev);
	int order = get_order(size);
	struct page *pg;
	dma_addr_t addr;

	gfp |= __GFP_ZERO;

	/*
	 * If the mask specifies that the memory be in the first 4 GB, then
	 * we force the allocation to come from the DMA zone.  We also
	 * force the node to 0 since that's the only node where the DMA
	 * zone isn't empty.  If the mask size is smaller than 32 bits, we
	 * may still not be able to guarantee a suitable memory address, in
	 * which case we will return NULL.  But such devices are uncommon.
	 */
	if (dma_mask <= DMA_BIT_MASK(32)) {
		gfp |= GFP_DMA;
		node = 0;
	}

	pg = homecache_alloc_pages_node(node, gfp, order, PAGE_HOME_DMA);
	if (pg == NULL)
		return NULL;

	addr = page_to_phys(pg);
	if (addr + size > dma_mask) {
		__homecache_free_pages(pg, order);
		return NULL;
	}

	*dma_handle = addr;

	return page_address(pg);
}

/*
 * Free memory that was allocated with tile_dma_alloc_coherent.
 */
static void tile_dma_free_coherent(struct device *dev, size_t size,
				   void *vaddr, dma_addr_t dma_handle,
				   struct dma_attrs *attrs)
{
	homecache_free_pages((unsigned long)vaddr, get_order(size));
}

/*
 * The map routines "map" the specified address range for DMA
 * accesses.  The memory belongs to the device after this call is
 * issued, until it is unmapped with dma_unmap_single.
 *
 * We don't need to do any mapping, we just flush the address range
 * out of the cache and return a DMA address.
 *
 * The unmap routines do whatever is necessary before the processor
 * accesses the memory again, and must be called before the driver
 * touches the memory.  We can get away with a cache invalidate if we
 * can count on nothing having been touched.
 */

/* Set up a single page for DMA access. */
static void __dma_prep_page(struct page *page, unsigned long offset,
			    size_t size, enum dma_data_direction direction)
{
	/*
	 * Flush the page from cache if necessary.
	 * On tilegx, data is delivered to hash-for-home L3; on tilepro,
	 * data is delivered direct to memory.
	 *
	 * NOTE: If we were just doing DMA_TO_DEVICE we could optimize
	 * this to be a "flush" not a "finv" and keep some of the
	 * state in cache across the DMA operation, but it doesn't seem
	 * worth creating the necessary flush_buffer_xxx() infrastructure.
	 */
	int home = page_home(page);
	switch (home) {
	case PAGE_HOME_HASH:
#ifdef __tilegx__
		return;
#endif
		break;
	case PAGE_HOME_UNCACHED:
#ifdef __tilepro__
		return;
#endif
		break;
	case PAGE_HOME_IMMUTABLE:
		/* Should be going to the device only. */
		BUG_ON(direction == DMA_FROM_DEVICE ||
		       direction == DMA_BIDIRECTIONAL);
		return;
	case PAGE_HOME_INCOHERENT:
		/* Incoherent anyway, so no need to work hard here. */
		return;
	default:
		BUG_ON(home < 0 || home >= NR_CPUS);
		break;
	}
	homecache_finv_page(page);

#ifdef DEBUG_ALIGNMENT
	/* Warn if the region isn't cacheline aligned. */
	if (offset & (L2_CACHE_BYTES - 1) || (size & (L2_CACHE_BYTES - 1)))
		pr_warn("Unaligned DMA to non-hfh memory: PA %#llx/%#lx\n",
			PFN_PHYS(page_to_pfn(page)) + offset, size);
#endif
}

/* Make the page ready to be read by the core. */
static void __dma_complete_page(struct page *page, unsigned long offset,
				size_t size, enum dma_data_direction direction)
{
#ifdef __tilegx__
	switch (page_home(page)) {
	case PAGE_HOME_HASH:
		/* I/O device delivered data the way the cpu wanted it. */
		break;
	case PAGE_HOME_INCOHERENT:
		/* Incoherent anyway, so no need to work hard here. */
		break;
	case PAGE_HOME_IMMUTABLE:
		/* Extra read-only copies are not a problem. */
		break;
	default:
		/* Flush the bogus hash-for-home I/O entries to memory. */
		homecache_finv_map_page(page, PAGE_HOME_HASH);
		break;
	}
#endif
}

static void __dma_prep_pa_range(dma_addr_t dma_addr, size_t size,
				enum dma_data_direction direction)
{
	struct page *page = pfn_to_page(PFN_DOWN(dma_addr));
	unsigned long offset = dma_addr & (PAGE_SIZE - 1);
	size_t bytes = min(size, (size_t)(PAGE_SIZE - offset));

	while (size != 0) {
		__dma_prep_page(page, offset, bytes, direction);
		size -= bytes;
		++page;
		offset = 0;
		bytes = min((size_t)PAGE_SIZE, size);
	}
}

static void __dma_complete_pa_range(dma_addr_t dma_addr, size_t size,
				    enum dma_data_direction direction)
{
	struct page *page = pfn_to_page(PFN_DOWN(dma_addr));
	unsigned long offset = dma_addr & (PAGE_SIZE - 1);
	size_t bytes = min(size, (size_t)(PAGE_SIZE - offset));

	while (size != 0) {
		__dma_complete_page(page, offset, bytes, direction);
		size -= bytes;
		++page;
		offset = 0;
		bytes = min((size_t)PAGE_SIZE, size);
	}
}

static int tile_dma_map_sg(struct device *dev, struct scatterlist *sglist,
			   int nents, enum dma_data_direction direction,
			   struct dma_attrs *attrs)
{
	struct scatterlist *sg;
	int i;

	BUG_ON(!valid_dma_direction(direction));

	WARN_ON(nents == 0 || sglist->length == 0);

	for_each_sg(sglist, sg, nents, i) {
		sg->dma_address = sg_phys(sg);
		__dma_prep_pa_range(sg->dma_address, sg->length, direction);
#ifdef CONFIG_NEED_SG_DMA_LENGTH
		sg->dma_length = sg->length;
#endif
	}

	return nents;
}

static void tile_dma_unmap_sg(struct device *dev, struct scatterlist *sglist,
			      int nents, enum dma_data_direction direction,
			      struct dma_attrs *attrs)
{
	struct scatterlist *sg;
	int i;

	BUG_ON(!valid_dma_direction(direction));
	for_each_sg(sglist, sg, nents, i) {
		sg->dma_address = sg_phys(sg);
		__dma_complete_pa_range(sg->dma_address, sg->length,
					direction);
	}
}

static dma_addr_t tile_dma_map_page(struct device *dev, struct page *page,
				    unsigned long offset, size_t size,
				    enum dma_data_direction direction,
				    struct dma_attrs *attrs)
{
	BUG_ON(!valid_dma_direction(direction));

	BUG_ON(offset + size > PAGE_SIZE);
	__dma_prep_page(page, offset, size, direction);

	return page_to_pa(page) + offset;
}

static void tile_dma_unmap_page(struct device *dev, dma_addr_t dma_address,
				size_t size, enum dma_data_direction direction,
				struct dma_attrs *attrs)
{
	BUG_ON(!valid_dma_direction(direction));

	__dma_complete_page(pfn_to_page(PFN_DOWN(dma_address)),
			    dma_address & PAGE_OFFSET, size, direction);
}

static void tile_dma_sync_single_for_cpu(struct device *dev,
					 dma_addr_t dma_handle,
					 size_t size,
					 enum dma_data_direction direction)
{
	BUG_ON(!valid_dma_direction(direction));

	__dma_complete_pa_range(dma_handle, size, direction);
}

static void tile_dma_sync_single_for_device(struct device *dev,
					    dma_addr_t dma_handle, size_t size,
					    enum dma_data_direction direction)
{
	__dma_prep_pa_range(dma_handle, size, direction);
}

static void tile_dma_sync_sg_for_cpu(struct device *dev,
				     struct scatterlist *sglist, int nelems,
				     enum dma_data_direction direction)
{
	struct scatterlist *sg;
	int i;

	BUG_ON(!valid_dma_direction(direction));
	WARN_ON(nelems == 0 || sglist->length == 0);

	for_each_sg(sglist, sg, nelems, i) {
		dma_sync_single_for_cpu(dev, sg->dma_address,
					sg_dma_len(sg), direction);
	}
}

static void tile_dma_sync_sg_for_device(struct device *dev,
					struct scatterlist *sglist, int nelems,
					enum dma_data_direction direction)
{
	struct scatterlist *sg;
	int i;

	BUG_ON(!valid_dma_direction(direction));
	WARN_ON(nelems == 0 || sglist->length == 0);

	for_each_sg(sglist, sg, nelems, i) {
		dma_sync_single_for_device(dev, sg->dma_address,
					   sg_dma_len(sg), direction);
	}
}

static inline int
tile_dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
{
	return 0;
}

static inline int
tile_dma_supported(struct device *dev, u64 mask)
{
	return 1;
}

static struct dma_map_ops tile_default_dma_map_ops = {
	.alloc = tile_dma_alloc_coherent,
	.free = tile_dma_free_coherent,
	.map_page = tile_dma_map_page,
	.unmap_page = tile_dma_unmap_page,
	.map_sg = tile_dma_map_sg,
	.unmap_sg = tile_dma_unmap_sg,
	.sync_single_for_cpu = tile_dma_sync_single_for_cpu,
	.sync_single_for_device = tile_dma_sync_single_for_device,
	.sync_sg_for_cpu = tile_dma_sync_sg_for_cpu,
	.sync_sg_for_device = tile_dma_sync_sg_for_device,
	.mapping_error = tile_dma_mapping_error,
	.dma_supported = tile_dma_supported
};

struct dma_map_ops *tile_dma_map_ops = &tile_default_dma_map_ops;
EXPORT_SYMBOL(tile_dma_map_ops);

/* Generic PCI DMA mapping functions */

static void *tile_pci_dma_alloc_coherent(struct device *dev, size_t size,
					 dma_addr_t *dma_handle, gfp_t gfp,
					 struct dma_attrs *attrs)
{
	int node = dev_to_node(dev);
	int order = get_order(size);
	struct page *pg;
	dma_addr_t addr;

	gfp |= __GFP_ZERO;

	pg = homecache_alloc_pages_node(node, gfp, order, PAGE_HOME_DMA);
	if (pg == NULL)
		return NULL;

	addr = page_to_phys(pg);

	*dma_handle = phys_to_dma(dev, addr);

	return page_address(pg);
}

/*
 * Free memory that was allocated with tile_pci_dma_alloc_coherent.
 */
static void tile_pci_dma_free_coherent(struct device *dev, size_t size,
				       void *vaddr, dma_addr_t dma_handle,
				       struct dma_attrs *attrs)
{
	homecache_free_pages((unsigned long)vaddr, get_order(size));
}

static int tile_pci_dma_map_sg(struct device *dev, struct scatterlist *sglist,
			       int nents, enum dma_data_direction direction,
			       struct dma_attrs *attrs)
{
	struct scatterlist *sg;
	int i;

	BUG_ON(!valid_dma_direction(direction));

	WARN_ON(nents == 0 || sglist->length == 0);

	for_each_sg(sglist, sg, nents, i) {
		sg->dma_address = sg_phys(sg);
		__dma_prep_pa_range(sg->dma_address, sg->length, direction);

		sg->dma_address = phys_to_dma(dev, sg->dma_address);
#ifdef CONFIG_NEED_SG_DMA_LENGTH
		sg->dma_length = sg->length;
#endif
	}

	return nents;
}

static void tile_pci_dma_unmap_sg(struct device *dev,
				  struct scatterlist *sglist, int nents,
				  enum dma_data_direction direction,
				  struct dma_attrs *attrs)
{
	struct scatterlist *sg;
	int i;

	BUG_ON(!valid_dma_direction(direction));
	for_each_sg(sglist, sg, nents, i) {
		sg->dma_address = sg_phys(sg);
		__dma_complete_pa_range(sg->dma_address, sg->length,
					direction);
	}
}

static dma_addr_t tile_pci_dma_map_page(struct device *dev, struct page *page,
					unsigned long offset, size_t size,
					enum dma_data_direction direction,
					struct dma_attrs *attrs)
{
	BUG_ON(!valid_dma_direction(direction));

	BUG_ON(offset + size > PAGE_SIZE);
	__dma_prep_page(page, offset, size, direction);

	return phys_to_dma(dev, page_to_pa(page) + offset);
}

static void tile_pci_dma_unmap_page(struct device *dev, dma_addr_t dma_address,
				    size_t size,
				    enum dma_data_direction direction,
				    struct dma_attrs *attrs)
{
	BUG_ON(!valid_dma_direction(direction));

	dma_address = dma_to_phys(dev, dma_address);

	__dma_complete_page(pfn_to_page(PFN_DOWN(dma_address)),
			    dma_address & PAGE_OFFSET, size, direction);
}

static void tile_pci_dma_sync_single_for_cpu(struct device *dev,
					     dma_addr_t dma_handle,
					     size_t size,
					     enum dma_data_direction direction)
{
	BUG_ON(!valid_dma_direction(direction));

	dma_handle = dma_to_phys(dev, dma_handle);

	__dma_complete_pa_range(dma_handle, size, direction);
}

static void tile_pci_dma_sync_single_for_device(struct device *dev,
						dma_addr_t dma_handle,
						size_t size,
						enum dma_data_direction
						direction)
{
	dma_handle = dma_to_phys(dev, dma_handle);

	__dma_prep_pa_range(dma_handle, size, direction);
}

static void tile_pci_dma_sync_sg_for_cpu(struct device *dev,
					 struct scatterlist *sglist,
					 int nelems,
					 enum dma_data_direction direction)
{
	struct scatterlist *sg;
	int i;

	BUG_ON(!valid_dma_direction(direction));
	WARN_ON(nelems == 0 || sglist->length == 0);

	for_each_sg(sglist, sg, nelems, i) {
		dma_sync_single_for_cpu(dev, sg->dma_address,
					sg_dma_len(sg), direction);
	}
}

static void tile_pci_dma_sync_sg_for_device(struct device *dev,
					    struct scatterlist *sglist,
					    int nelems,
					    enum dma_data_direction direction)
{
	struct scatterlist *sg;
	int i;

	BUG_ON(!valid_dma_direction(direction));
	WARN_ON(nelems == 0 || sglist->length == 0);

	for_each_sg(sglist, sg, nelems, i) {
		dma_sync_single_for_device(dev, sg->dma_address,
					   sg_dma_len(sg), direction);
	}
}

static inline int
tile_pci_dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
{
	return 0;
}

static inline int
tile_pci_dma_supported(struct device *dev, u64 mask)
{
	return 1;
}

static struct dma_map_ops tile_pci_default_dma_map_ops = {
	.alloc = tile_pci_dma_alloc_coherent,
	.free = tile_pci_dma_free_coherent,
	.map_page = tile_pci_dma_map_page,
	.unmap_page = tile_pci_dma_unmap_page,
	.map_sg = tile_pci_dma_map_sg,
	.unmap_sg = tile_pci_dma_unmap_sg,
	.sync_single_for_cpu = tile_pci_dma_sync_single_for_cpu,
	.sync_single_for_device = tile_pci_dma_sync_single_for_device,
	.sync_sg_for_cpu = tile_pci_dma_sync_sg_for_cpu,
	.sync_sg_for_device = tile_pci_dma_sync_sg_for_device,
	.mapping_error = tile_pci_dma_mapping_error,
	.dma_supported = tile_pci_dma_supported
};

struct dma_map_ops *gx_pci_dma_map_ops = &tile_pci_default_dma_map_ops;
EXPORT_SYMBOL(gx_pci_dma_map_ops);

/* PCI DMA mapping functions for legacy PCI devices */

#ifdef CONFIG_SWIOTLB
static void *tile_swiotlb_alloc_coherent(struct device *dev, size_t size,
					 dma_addr_t *dma_handle, gfp_t gfp,
					 struct dma_attrs *attrs)
{
	gfp |= GFP_DMA;
	return swiotlb_alloc_coherent(dev, size, dma_handle, gfp);
}

static void tile_swiotlb_free_coherent(struct device *dev, size_t size,
				       void *vaddr, dma_addr_t dma_addr,
				       struct dma_attrs *attrs)
{
	swiotlb_free_coherent(dev, size, vaddr, dma_addr);
}

static struct dma_map_ops pci_swiotlb_dma_ops = {
	.alloc = tile_swiotlb_alloc_coherent,
	.free = tile_swiotlb_free_coherent,
	.map_page = swiotlb_map_page,
	.unmap_page = swiotlb_unmap_page,
	.map_sg = swiotlb_map_sg_attrs,
	.unmap_sg = swiotlb_unmap_sg_attrs,
	.sync_single_for_cpu = swiotlb_sync_single_for_cpu,
	.sync_single_for_device = swiotlb_sync_single_for_device,
	.sync_sg_for_cpu = swiotlb_sync_sg_for_cpu,
	.sync_sg_for_device = swiotlb_sync_sg_for_device,
	.dma_supported = swiotlb_dma_supported,
	.mapping_error = swiotlb_dma_mapping_error,
};

struct dma_map_ops *gx_legacy_pci_dma_map_ops = &pci_swiotlb_dma_ops;
#else
struct dma_map_ops *gx_legacy_pci_dma_map_ops;
#endif
EXPORT_SYMBOL(gx_legacy_pci_dma_map_ops);

#ifdef CONFIG_ARCH_HAS_DMA_SET_COHERENT_MASK
int dma_set_coherent_mask(struct device *dev, u64 mask)
{
	struct dma_map_ops *dma_ops = get_dma_ops(dev);

	/* Handle legacy PCI devices with limited memory addressability. */
	if (((dma_ops == gx_pci_dma_map_ops) ||
	    (dma_ops == gx_legacy_pci_dma_map_ops)) &&
	    (mask <= DMA_BIT_MASK(32))) {
		if (mask > dev->archdata.max_direct_dma_addr)
			mask = dev->archdata.max_direct_dma_addr;
	}

	if (!dma_supported(dev, mask))
		return -EIO;
	dev->coherent_dma_mask = mask;
	return 0;
}
EXPORT_SYMBOL(dma_set_coherent_mask);
#endif