summaryrefslogtreecommitdiffstats
path: root/drivers/gpu/drm/nouveau/nouveau_sgdma.c
blob: a33fe4019286e39294af123b7f8140973451c623 (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
#include "drmP.h"
#include "nouveau_drv.h"
#include <linux/pagemap.h>
#include <linux/slab.h>

#define NV_CTXDMA_PAGE_SHIFT 12
#define NV_CTXDMA_PAGE_SIZE  (1 << NV_CTXDMA_PAGE_SHIFT)
#define NV_CTXDMA_PAGE_MASK  (NV_CTXDMA_PAGE_SIZE - 1)

struct nouveau_sgdma_be {
	struct ttm_backend backend;
	struct drm_device *dev;

	dma_addr_t *pages;
	bool *ttm_alloced;
	unsigned nr_pages;

	u64 offset;
	bool bound;
};

static int
nouveau_sgdma_populate(struct ttm_backend *be, unsigned long num_pages,
		       struct page **pages, struct page *dummy_read_page,
		       dma_addr_t *dma_addrs)
{
	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
	struct drm_device *dev = nvbe->dev;

	NV_DEBUG(nvbe->dev, "num_pages = %ld\n", num_pages);

	if (nvbe->pages)
		return -EINVAL;

	nvbe->pages = kmalloc(sizeof(dma_addr_t) * num_pages, GFP_KERNEL);
	if (!nvbe->pages)
		return -ENOMEM;

	nvbe->ttm_alloced = kmalloc(sizeof(bool) * num_pages, GFP_KERNEL);
	if (!nvbe->ttm_alloced)
		return -ENOMEM;

	nvbe->nr_pages = 0;
	while (num_pages--) {
		if (dma_addrs[nvbe->nr_pages] != DMA_ERROR_CODE) {
			nvbe->pages[nvbe->nr_pages] =
					dma_addrs[nvbe->nr_pages];
		 	nvbe->ttm_alloced[nvbe->nr_pages] = true;
		} else {
			nvbe->pages[nvbe->nr_pages] =
				pci_map_page(dev->pdev, pages[nvbe->nr_pages], 0,
				     PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
			if (pci_dma_mapping_error(dev->pdev,
						  nvbe->pages[nvbe->nr_pages])) {
				be->func->clear(be);
				return -EFAULT;
			}
		}

		nvbe->nr_pages++;
	}

	return 0;
}

static void
nouveau_sgdma_clear(struct ttm_backend *be)
{
	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
	struct drm_device *dev;

	if (nvbe && nvbe->pages) {
		dev = nvbe->dev;
		NV_DEBUG(dev, "\n");

		if (nvbe->bound)
			be->func->unbind(be);

		while (nvbe->nr_pages--) {
			if (!nvbe->ttm_alloced[nvbe->nr_pages])
				pci_unmap_page(dev->pdev, nvbe->pages[nvbe->nr_pages],
				       PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
		}
		kfree(nvbe->pages);
		kfree(nvbe->ttm_alloced);
		nvbe->pages = NULL;
		nvbe->ttm_alloced = NULL;
		nvbe->nr_pages = 0;
	}
}

static void
nouveau_sgdma_destroy(struct ttm_backend *be)
{
	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;

	if (be) {
		NV_DEBUG(nvbe->dev, "\n");

		if (nvbe) {
			if (nvbe->pages)
				be->func->clear(be);
			kfree(nvbe);
		}
	}
}

static int
nv04_sgdma_bind(struct ttm_backend *be, struct ttm_mem_reg *mem)
{
	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
	struct drm_device *dev = nvbe->dev;
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	struct nouveau_gpuobj *gpuobj = dev_priv->gart_info.sg_ctxdma;
	unsigned i, j, pte;

	NV_DEBUG(dev, "pg=0x%lx\n", mem->start);

	nvbe->offset = mem->start << PAGE_SHIFT;
	pte = (nvbe->offset >> NV_CTXDMA_PAGE_SHIFT) + 2;
	for (i = 0; i < nvbe->nr_pages; i++) {
		dma_addr_t dma_offset = nvbe->pages[i];
		uint32_t offset_l = lower_32_bits(dma_offset);

		for (j = 0; j < PAGE_SIZE / NV_CTXDMA_PAGE_SIZE; j++, pte++) {
			nv_wo32(gpuobj, (pte * 4) + 0, offset_l | 3);
			dma_offset += NV_CTXDMA_PAGE_SIZE;
		}
	}

	nvbe->bound = true;
	return 0;
}

static int
nv04_sgdma_unbind(struct ttm_backend *be)
{
	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
	struct drm_device *dev = nvbe->dev;
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	struct nouveau_gpuobj *gpuobj = dev_priv->gart_info.sg_ctxdma;
	unsigned i, j, pte;

	NV_DEBUG(dev, "\n");

	if (!nvbe->bound)
		return 0;

	pte = (nvbe->offset >> NV_CTXDMA_PAGE_SHIFT) + 2;
	for (i = 0; i < nvbe->nr_pages; i++) {
		for (j = 0; j < PAGE_SIZE / NV_CTXDMA_PAGE_SIZE; j++, pte++)
			nv_wo32(gpuobj, (pte * 4) + 0, 0x00000000);
	}

	nvbe->bound = false;
	return 0;
}

static struct ttm_backend_func nv04_sgdma_backend = {
	.populate		= nouveau_sgdma_populate,
	.clear			= nouveau_sgdma_clear,
	.bind			= nv04_sgdma_bind,
	.unbind			= nv04_sgdma_unbind,
	.destroy		= nouveau_sgdma_destroy
};

static void
nv41_sgdma_flush(struct nouveau_sgdma_be *nvbe)
{
	struct drm_device *dev = nvbe->dev;

	nv_wr32(dev, 0x100810, 0x00000022);
	if (!nv_wait(dev, 0x100810, 0x00000100, 0x00000100))
		NV_ERROR(dev, "vm flush timeout: 0x%08x\n",
			 nv_rd32(dev, 0x100810));
	nv_wr32(dev, 0x100810, 0x00000000);
}

static int
nv41_sgdma_bind(struct ttm_backend *be, struct ttm_mem_reg *mem)
{
	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
	struct drm_nouveau_private *dev_priv = nvbe->dev->dev_private;
	struct nouveau_gpuobj *pgt = dev_priv->gart_info.sg_ctxdma;
	dma_addr_t *list = nvbe->pages;
	u32 pte = mem->start << 2;
	u32 cnt = nvbe->nr_pages;

	nvbe->offset = mem->start << PAGE_SHIFT;

	while (cnt--) {
		nv_wo32(pgt, pte, (*list++ >> 7) | 1);
		pte += 4;
	}

	nv41_sgdma_flush(nvbe);
	nvbe->bound = true;
	return 0;
}

static int
nv41_sgdma_unbind(struct ttm_backend *be)
{
	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
	struct drm_nouveau_private *dev_priv = nvbe->dev->dev_private;
	struct nouveau_gpuobj *pgt = dev_priv->gart_info.sg_ctxdma;
	u32 pte = (nvbe->offset >> 12) << 2;
	u32 cnt = nvbe->nr_pages;

	while (cnt--) {
		nv_wo32(pgt, pte, 0x00000000);
		pte += 4;
	}

	nv41_sgdma_flush(nvbe);
	nvbe->bound = false;
	return 0;
}

static struct ttm_backend_func nv41_sgdma_backend = {
	.populate		= nouveau_sgdma_populate,
	.clear			= nouveau_sgdma_clear,
	.bind			= nv41_sgdma_bind,
	.unbind			= nv41_sgdma_unbind,
	.destroy		= nouveau_sgdma_destroy
};

static void
nv44_sgdma_flush(struct nouveau_sgdma_be *nvbe)
{
	struct drm_device *dev = nvbe->dev;

	nv_wr32(dev, 0x100814, (nvbe->nr_pages - 1) << 12);
	nv_wr32(dev, 0x100808, nvbe->offset | 0x20);
	if (!nv_wait(dev, 0x100808, 0x00000001, 0x00000001))
		NV_ERROR(dev, "gart flush timeout: 0x%08x\n",
			 nv_rd32(dev, 0x100808));
	nv_wr32(dev, 0x100808, 0x00000000);
}

static void
nv44_sgdma_fill(struct nouveau_gpuobj *pgt, dma_addr_t *list, u32 base, u32 cnt)
{
	struct drm_nouveau_private *dev_priv = pgt->dev->dev_private;
	dma_addr_t dummy = dev_priv->gart_info.dummy.addr;
	u32 pte, tmp[4];

	pte   = base >> 2;
	base &= ~0x0000000f;

	tmp[0] = nv_ro32(pgt, base + 0x0);
	tmp[1] = nv_ro32(pgt, base + 0x4);
	tmp[2] = nv_ro32(pgt, base + 0x8);
	tmp[3] = nv_ro32(pgt, base + 0xc);
	while (cnt--) {
		u32 addr = list ? (*list++ >> 12) : (dummy >> 12);
		switch (pte++ & 0x3) {
		case 0:
			tmp[0] &= ~0x07ffffff;
			tmp[0] |= addr;
			break;
		case 1:
			tmp[0] &= ~0xf8000000;
			tmp[0] |= addr << 27;
			tmp[1] &= ~0x003fffff;
			tmp[1] |= addr >> 5;
			break;
		case 2:
			tmp[1] &= ~0xffc00000;
			tmp[1] |= addr << 22;
			tmp[2] &= ~0x0001ffff;
			tmp[2] |= addr >> 10;
			break;
		case 3:
			tmp[2] &= ~0xfffe0000;
			tmp[2] |= addr << 17;
			tmp[3] &= ~0x00000fff;
			tmp[3] |= addr >> 15;
			break;
		}
	}

	tmp[3] |= 0x40000000;

	nv_wo32(pgt, base + 0x0, tmp[0]);
	nv_wo32(pgt, base + 0x4, tmp[1]);
	nv_wo32(pgt, base + 0x8, tmp[2]);
	nv_wo32(pgt, base + 0xc, tmp[3]);
}

static int
nv44_sgdma_bind(struct ttm_backend *be, struct ttm_mem_reg *mem)
{
	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
	struct drm_nouveau_private *dev_priv = nvbe->dev->dev_private;
	struct nouveau_gpuobj *pgt = dev_priv->gart_info.sg_ctxdma;
	dma_addr_t *list = nvbe->pages;
	u32 pte = mem->start << 2, tmp[4];
	u32 cnt = nvbe->nr_pages;
	int i;

	nvbe->offset = mem->start << PAGE_SHIFT;

	if (pte & 0x0000000c) {
		u32  max = 4 - ((pte >> 2) & 0x3);
		u32 part = (cnt > max) ? max : cnt;
		nv44_sgdma_fill(pgt, list, pte, part);
		pte  += (part << 2);
		list += part;
		cnt  -= part;
	}

	while (cnt >= 4) {
		for (i = 0; i < 4; i++)
			tmp[i] = *list++ >> 12;
		nv_wo32(pgt, pte + 0x0, tmp[0] >>  0 | tmp[1] << 27);
		nv_wo32(pgt, pte + 0x4, tmp[1] >>  5 | tmp[2] << 22);
		nv_wo32(pgt, pte + 0x8, tmp[2] >> 10 | tmp[3] << 17);
		nv_wo32(pgt, pte + 0xc, tmp[3] >> 15 | 0x40000000);
		pte  += 0x10;
		cnt  -= 4;
	}

	if (cnt)
		nv44_sgdma_fill(pgt, list, pte, cnt);

	nv44_sgdma_flush(nvbe);
	nvbe->bound = true;
	return 0;
}

static int
nv44_sgdma_unbind(struct ttm_backend *be)
{
	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
	struct drm_nouveau_private *dev_priv = nvbe->dev->dev_private;
	struct nouveau_gpuobj *pgt = dev_priv->gart_info.sg_ctxdma;
	u32 pte = (nvbe->offset >> 12) << 2;
	u32 cnt = nvbe->nr_pages;

	if (pte & 0x0000000c) {
		u32  max = 4 - ((pte >> 2) & 0x3);
		u32 part = (cnt > max) ? max : cnt;
		nv44_sgdma_fill(pgt, NULL, pte, part);
		pte  += (part << 2);
		cnt  -= part;
	}

	while (cnt >= 4) {
		nv_wo32(pgt, pte + 0x0, 0x00000000);
		nv_wo32(pgt, pte + 0x4, 0x00000000);
		nv_wo32(pgt, pte + 0x8, 0x00000000);
		nv_wo32(pgt, pte + 0xc, 0x00000000);
		pte  += 0x10;
		cnt  -= 4;
	}

	if (cnt)
		nv44_sgdma_fill(pgt, NULL, pte, cnt);

	nv44_sgdma_flush(nvbe);
	nvbe->bound = false;
	return 0;
}

static struct ttm_backend_func nv44_sgdma_backend = {
	.populate		= nouveau_sgdma_populate,
	.clear			= nouveau_sgdma_clear,
	.bind			= nv44_sgdma_bind,
	.unbind			= nv44_sgdma_unbind,
	.destroy		= nouveau_sgdma_destroy
};

static int
nv50_sgdma_bind(struct ttm_backend *be, struct ttm_mem_reg *mem)
{
	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
	struct nouveau_mem *node = mem->mm_node;
	/* noop: bound in move_notify() */
	node->pages = nvbe->pages;
	nvbe->pages = (dma_addr_t *)node;
	nvbe->bound = true;
	return 0;
}

static int
nv50_sgdma_unbind(struct ttm_backend *be)
{
	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
	struct nouveau_mem *node = (struct nouveau_mem *)nvbe->pages;
	/* noop: unbound in move_notify() */
	nvbe->pages = node->pages;
	node->pages = NULL;
	nvbe->bound = false;
	return 0;
}

static struct ttm_backend_func nv50_sgdma_backend = {
	.populate		= nouveau_sgdma_populate,
	.clear			= nouveau_sgdma_clear,
	.bind			= nv50_sgdma_bind,
	.unbind			= nv50_sgdma_unbind,
	.destroy		= nouveau_sgdma_destroy
};

struct ttm_backend *
nouveau_sgdma_init_ttm(struct drm_device *dev)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	struct nouveau_sgdma_be *nvbe;

	nvbe = kzalloc(sizeof(*nvbe), GFP_KERNEL);
	if (!nvbe)
		return NULL;

	nvbe->dev = dev;

	nvbe->backend.func = dev_priv->gart_info.func;
	return &nvbe->backend;
}

int
nouveau_sgdma_init(struct drm_device *dev)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	struct nouveau_gpuobj *gpuobj = NULL;
	u32 aper_size, align;
	int ret;

	if (dev_priv->card_type >= NV_50 || drm_pci_device_is_pcie(dev))
		aper_size = 512 * 1024 * 1024;
	else
		aper_size = 64 * 1024 * 1024;

	/* Dear NVIDIA, NV44+ would like proper present bits in PTEs for
	 * christmas.  The cards before it have them, the cards after
	 * it have them, why is NV44 so unloved?
	 */
	dev_priv->gart_info.dummy.page = alloc_page(GFP_DMA32 | GFP_KERNEL);
	if (!dev_priv->gart_info.dummy.page)
		return -ENOMEM;

	dev_priv->gart_info.dummy.addr =
		pci_map_page(dev->pdev, dev_priv->gart_info.dummy.page,
			     0, PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
	if (pci_dma_mapping_error(dev->pdev, dev_priv->gart_info.dummy.addr)) {
		NV_ERROR(dev, "error mapping dummy page\n");
		__free_page(dev_priv->gart_info.dummy.page);
		dev_priv->gart_info.dummy.page = NULL;
		return -ENOMEM;
	}

	if (dev_priv->card_type >= NV_50) {
		dev_priv->gart_info.aper_base = 0;
		dev_priv->gart_info.aper_size = aper_size;
		dev_priv->gart_info.type = NOUVEAU_GART_HW;
		dev_priv->gart_info.func = &nv50_sgdma_backend;
	} else
	if (drm_pci_device_is_pcie(dev) &&
	    dev_priv->chipset != 0x40 && dev_priv->chipset != 0x45) {
		if (nv44_graph_class(dev)) {
			dev_priv->gart_info.func = &nv44_sgdma_backend;
			align = 512 * 1024;
		} else {
			dev_priv->gart_info.func = &nv41_sgdma_backend;
			align = 16;
		}

		ret = nouveau_gpuobj_new(dev, NULL, aper_size / 1024, align,
					 NVOBJ_FLAG_ZERO_ALLOC |
					 NVOBJ_FLAG_ZERO_FREE, &gpuobj);
		if (ret) {
			NV_ERROR(dev, "Error creating sgdma object: %d\n", ret);
			return ret;
		}

		dev_priv->gart_info.sg_ctxdma = gpuobj;
		dev_priv->gart_info.aper_base = 0;
		dev_priv->gart_info.aper_size = aper_size;
		dev_priv->gart_info.type = NOUVEAU_GART_HW;
	} else {
		ret = nouveau_gpuobj_new(dev, NULL, (aper_size / 1024) + 8, 16,
					 NVOBJ_FLAG_ZERO_ALLOC |
					 NVOBJ_FLAG_ZERO_FREE, &gpuobj);
		if (ret) {
			NV_ERROR(dev, "Error creating sgdma object: %d\n", ret);
			return ret;
		}

		nv_wo32(gpuobj, 0, NV_CLASS_DMA_IN_MEMORY |
				   (1 << 12) /* PT present */ |
				   (0 << 13) /* PT *not* linear */ |
				   (0 << 14) /* RW */ |
				   (2 << 16) /* PCI */);
		nv_wo32(gpuobj, 4, aper_size - 1);

		dev_priv->gart_info.sg_ctxdma = gpuobj;
		dev_priv->gart_info.aper_base = 0;
		dev_priv->gart_info.aper_size = aper_size;
		dev_priv->gart_info.type = NOUVEAU_GART_PDMA;
		dev_priv->gart_info.func = &nv04_sgdma_backend;
	}

	return 0;
}

void
nouveau_sgdma_takedown(struct drm_device *dev)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;

	nouveau_gpuobj_ref(NULL, &dev_priv->gart_info.sg_ctxdma);

	if (dev_priv->gart_info.dummy.page) {
		pci_unmap_page(dev->pdev, dev_priv->gart_info.dummy.addr,
			       PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
		__free_page(dev_priv->gart_info.dummy.page);
		dev_priv->gart_info.dummy.page = NULL;
	}
}

uint32_t
nouveau_sgdma_get_physical(struct drm_device *dev, uint32_t offset)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	struct nouveau_gpuobj *gpuobj = dev_priv->gart_info.sg_ctxdma;
	int pte = (offset >> NV_CTXDMA_PAGE_SHIFT) + 2;

	BUG_ON(dev_priv->card_type >= NV_50);

	return (nv_ro32(gpuobj, 4 * pte) & ~NV_CTXDMA_PAGE_MASK) |
		(offset & NV_CTXDMA_PAGE_MASK);
}
OpenPOWER on IntegriCloud