summaryrefslogtreecommitdiffstats
path: root/drivers/gpu/drm/nouveau/nouveau_mem.c
blob: 02755712ed3d3aa03641478f52c0ef65ab7b9a29 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
/*
 * Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
 * Copyright 2005 Stephane Marchesin
 *
 * The Weather Channel (TM) funded Tungsten Graphics to develop the
 * initial release of the Radeon 8500 driver under the XFree86 license.
 * This notice must be preserved.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the next
 * paragraph) shall be included in all copies or substantial portions of the
 * Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
 * DEALINGS IN THE SOFTWARE.
 *
 * Authors:
 *    Keith Whitwell <keith@tungstengraphics.com>
 */


#include "drmP.h"
#include "drm.h"
#include "drm_sarea.h"
#include "nouveau_drv.h"

static struct mem_block *
split_block(struct mem_block *p, uint64_t start, uint64_t size,
	    struct drm_file *file_priv)
{
	/* Maybe cut off the start of an existing block */
	if (start > p->start) {
		struct mem_block *newblock =
			kmalloc(sizeof(*newblock), GFP_KERNEL);
		if (!newblock)
			goto out;
		newblock->start = start;
		newblock->size = p->size - (start - p->start);
		newblock->file_priv = NULL;
		newblock->next = p->next;
		newblock->prev = p;
		p->next->prev = newblock;
		p->next = newblock;
		p->size -= newblock->size;
		p = newblock;
	}

	/* Maybe cut off the end of an existing block */
	if (size < p->size) {
		struct mem_block *newblock =
			kmalloc(sizeof(*newblock), GFP_KERNEL);
		if (!newblock)
			goto out;
		newblock->start = start + size;
		newblock->size = p->size - size;
		newblock->file_priv = NULL;
		newblock->next = p->next;
		newblock->prev = p;
		p->next->prev = newblock;
		p->next = newblock;
		p->size = size;
	}

out:
	/* Our block is in the middle */
	p->file_priv = file_priv;
	return p;
}

struct mem_block *
nouveau_mem_alloc_block(struct mem_block *heap, uint64_t size,
			int align2, struct drm_file *file_priv, int tail)
{
	struct mem_block *p;
	uint64_t mask = (1 << align2) - 1;

	if (!heap)
		return NULL;

	if (tail) {
		list_for_each_prev(p, heap) {
			uint64_t start = ((p->start + p->size) - size) & ~mask;

			if (p->file_priv == NULL && start >= p->start &&
			    start + size <= p->start + p->size)
				return split_block(p, start, size, file_priv);
		}
	} else {
		list_for_each(p, heap) {
			uint64_t start = (p->start + mask) & ~mask;

			if (p->file_priv == NULL &&
			    start + size <= p->start + p->size)
				return split_block(p, start, size, file_priv);
		}
	}

	return NULL;
}

void nouveau_mem_free_block(struct mem_block *p)
{
	p->file_priv = NULL;

	/* Assumes a single contiguous range.  Needs a special file_priv in
	 * 'heap' to stop it being subsumed.
	 */
	if (p->next->file_priv == NULL) {
		struct mem_block *q = p->next;
		p->size += q->size;
		p->next = q->next;
		p->next->prev = p;
		kfree(q);
	}

	if (p->prev->file_priv == NULL) {
		struct mem_block *q = p->prev;
		q->size += p->size;
		q->next = p->next;
		q->next->prev = q;
		kfree(p);
	}
}

/* Initialize.  How to check for an uninitialized heap?
 */
int nouveau_mem_init_heap(struct mem_block **heap, uint64_t start,
			  uint64_t size)
{
	struct mem_block *blocks = kmalloc(sizeof(*blocks), GFP_KERNEL);

	if (!blocks)
		return -ENOMEM;

	*heap = kmalloc(sizeof(**heap), GFP_KERNEL);
	if (!*heap) {
		kfree(blocks);
		return -ENOMEM;
	}

	blocks->start = start;
	blocks->size = size;
	blocks->file_priv = NULL;
	blocks->next = blocks->prev = *heap;

	memset(*heap, 0, sizeof(**heap));
	(*heap)->file_priv = (struct drm_file *) -1;
	(*heap)->next = (*heap)->prev = blocks;
	return 0;
}

/*
 * Free all blocks associated with the releasing file_priv
 */
void nouveau_mem_release(struct drm_file *file_priv, struct mem_block *heap)
{
	struct mem_block *p;

	if (!heap || !heap->next)
		return;

	list_for_each(p, heap) {
		if (p->file_priv == file_priv)
			p->file_priv = NULL;
	}

	/* Assumes a single contiguous range.  Needs a special file_priv in
	 * 'heap' to stop it being subsumed.
	 */
	list_for_each(p, heap) {
		while ((p->file_priv == NULL) &&
					(p->next->file_priv == NULL) &&
					(p->next != heap)) {
			struct mem_block *q = p->next;
			p->size += q->size;
			p->next = q->next;
			p->next->prev = p;
			kfree(q);
		}
	}
}

/*
 * NV50 VM helpers
 */
int
nv50_mem_vm_bind_linear(struct drm_device *dev, uint64_t virt, uint32_t size,
			uint32_t flags, uint64_t phys)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	struct nouveau_gpuobj **pgt;
	unsigned psz, pfl, pages;

	if (virt >= dev_priv->vm_gart_base &&
	    (virt + size) < (dev_priv->vm_gart_base + dev_priv->vm_gart_size)) {
		psz = 12;
		pgt = &dev_priv->gart_info.sg_ctxdma;
		pfl = 0x21;
		virt -= dev_priv->vm_gart_base;
	} else
	if (virt >= dev_priv->vm_vram_base &&
	    (virt + size) < (dev_priv->vm_vram_base + dev_priv->vm_vram_size)) {
		psz = 16;
		pgt = dev_priv->vm_vram_pt;
		pfl = 0x01;
		virt -= dev_priv->vm_vram_base;
	} else {
		NV_ERROR(dev, "Invalid address: 0x%16llx-0x%16llx\n",
			 virt, virt + size - 1);
		return -EINVAL;
	}

	pages = size >> psz;

	dev_priv->engine.instmem.prepare_access(dev, true);
	if (flags & 0x80000000) {
		while (pages--) {
			struct nouveau_gpuobj *pt = pgt[virt >> 29];
			unsigned pte = ((virt & 0x1fffffffULL) >> psz) << 1;

			nv_wo32(dev, pt, pte++, 0x00000000);
			nv_wo32(dev, pt, pte++, 0x00000000);

			virt += (1 << psz);
		}
	} else {
		while (pages--) {
			struct nouveau_gpuobj *pt = pgt[virt >> 29];
			unsigned pte = ((virt & 0x1fffffffULL) >> psz) << 1;
			unsigned offset_h = upper_32_bits(phys) & 0xff;
			unsigned offset_l = lower_32_bits(phys);

			nv_wo32(dev, pt, pte++, offset_l | pfl);
			nv_wo32(dev, pt, pte++, offset_h | flags);

			phys += (1 << psz);
			virt += (1 << psz);
		}
	}
	dev_priv->engine.instmem.finish_access(dev);

	nv_wr32(dev, 0x100c80, 0x00050001);
	if (!nv_wait(0x100c80, 0x00000001, 0x00000000)) {
		NV_ERROR(dev, "timeout: (0x100c80 & 1) == 0 (2)\n");
		NV_ERROR(dev, "0x100c80 = 0x%08x\n", nv_rd32(dev, 0x100c80));
		return -EBUSY;
	}

	nv_wr32(dev, 0x100c80, 0x00000001);
	if (!nv_wait(0x100c80, 0x00000001, 0x00000000)) {
		NV_ERROR(dev, "timeout: (0x100c80 & 1) == 0 (2)\n");
		NV_ERROR(dev, "0x100c80 = 0x%08x\n", nv_rd32(dev, 0x100c80));
		return -EBUSY;
	}

	return 0;
}

void
nv50_mem_vm_unbind(struct drm_device *dev, uint64_t virt, uint32_t size)
{
	nv50_mem_vm_bind_linear(dev, virt, size, 0x80000000, 0);
}

/*
 * Cleanup everything
 */
void nouveau_mem_takedown(struct mem_block **heap)
{
	struct mem_block *p;

	if (!*heap)
		return;

	for (p = (*heap)->next; p != *heap;) {
		struct mem_block *q = p;
		p = p->next;
		kfree(q);
	}

	kfree(*heap);
	*heap = NULL;
}

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

	if (dev_priv->ttm.bdev.man[TTM_PL_PRIV0].has_type)
		ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_PRIV0);
	ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_VRAM);

	ttm_bo_device_release(&dev_priv->ttm.bdev);

	nouveau_ttm_global_release(dev_priv);

	if (drm_core_has_AGP(dev) && dev->agp &&
	    drm_core_check_feature(dev, DRIVER_MODESET)) {
		struct drm_agp_mem *entry, *tempe;

		/* Remove AGP resources, but leave dev->agp
		   intact until drv_cleanup is called. */
		list_for_each_entry_safe(entry, tempe, &dev->agp->memory, head) {
			if (entry->bound)
				drm_unbind_agp(entry->memory);
			drm_free_agp(entry->memory, entry->pages);
			kfree(entry);
		}
		INIT_LIST_HEAD(&dev->agp->memory);

		if (dev->agp->acquired)
			drm_agp_release(dev);

		dev->agp->acquired = 0;
		dev->agp->enabled = 0;
	}

	if (dev_priv->fb_mtrr) {
		drm_mtrr_del(dev_priv->fb_mtrr, drm_get_resource_start(dev, 1),
			     drm_get_resource_len(dev, 1), DRM_MTRR_WC);
		dev_priv->fb_mtrr = 0;
	}
}

/*XXX won't work on BSD because of pci_read_config_dword */
static uint32_t
nouveau_mem_fb_amount_igp(struct drm_device *dev)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	struct pci_dev *bridge;
	uint32_t mem;

	bridge = pci_get_bus_and_slot(0, PCI_DEVFN(0, 1));
	if (!bridge) {
		NV_ERROR(dev, "no bridge device\n");
		return 0;
	}

	if (dev_priv->flags&NV_NFORCE) {
		pci_read_config_dword(bridge, 0x7C, &mem);
		return (uint64_t)(((mem >> 6) & 31) + 1)*1024*1024;
	} else
	if (dev_priv->flags&NV_NFORCE2) {
		pci_read_config_dword(bridge, 0x84, &mem);
		return (uint64_t)(((mem >> 4) & 127) + 1)*1024*1024;
	}

	NV_ERROR(dev, "impossible!\n");
	return 0;
}

/* returns the amount of FB ram in bytes */
uint64_t nouveau_mem_fb_amount(struct drm_device *dev)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	uint32_t boot0;

	switch (dev_priv->card_type) {
	case NV_04:
		boot0 = nv_rd32(dev, NV03_BOOT_0);
		if (boot0 & 0x00000100)
			return (((boot0 >> 12) & 0xf) * 2 + 2) * 1024 * 1024;

		switch (boot0 & NV03_BOOT_0_RAM_AMOUNT) {
		case NV04_BOOT_0_RAM_AMOUNT_32MB:
			return 32 * 1024 * 1024;
		case NV04_BOOT_0_RAM_AMOUNT_16MB:
			return 16 * 1024 * 1024;
		case NV04_BOOT_0_RAM_AMOUNT_8MB:
			return 8 * 1024 * 1024;
		case NV04_BOOT_0_RAM_AMOUNT_4MB:
			return 4 * 1024 * 1024;
		}
		break;
	case NV_10:
	case NV_20:
	case NV_30:
	case NV_40:
	case NV_50:
	default:
		if (dev_priv->flags & (NV_NFORCE | NV_NFORCE2)) {
			return nouveau_mem_fb_amount_igp(dev);
		} else {
			uint64_t mem;
			mem = (nv_rd32(dev, NV04_FIFO_DATA) &
					NV10_FIFO_DATA_RAM_AMOUNT_MB_MASK) >>
					NV10_FIFO_DATA_RAM_AMOUNT_MB_SHIFT;
			return mem * 1024 * 1024;
		}
		break;
	}

	NV_ERROR(dev,
		"Unable to detect video ram size. Please report your setup to "
							DRIVER_EMAIL "\n");
	return 0;
}

static void nouveau_mem_reset_agp(struct drm_device *dev)
{
	uint32_t saved_pci_nv_1, saved_pci_nv_19, pmc_enable;

	saved_pci_nv_1 = nv_rd32(dev, NV04_PBUS_PCI_NV_1);
	saved_pci_nv_19 = nv_rd32(dev, NV04_PBUS_PCI_NV_19);

	/* clear busmaster bit */
	nv_wr32(dev, NV04_PBUS_PCI_NV_1, saved_pci_nv_1 & ~0x4);
	/* clear SBA and AGP bits */
	nv_wr32(dev, NV04_PBUS_PCI_NV_19, saved_pci_nv_19 & 0xfffff0ff);

	/* power cycle pgraph, if enabled */
	pmc_enable = nv_rd32(dev, NV03_PMC_ENABLE);
	if (pmc_enable & NV_PMC_ENABLE_PGRAPH) {
		nv_wr32(dev, NV03_PMC_ENABLE,
				pmc_enable & ~NV_PMC_ENABLE_PGRAPH);
		nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) |
				NV_PMC_ENABLE_PGRAPH);
	}

	/* and restore (gives effect of resetting AGP) */
	nv_wr32(dev, NV04_PBUS_PCI_NV_19, saved_pci_nv_19);
	nv_wr32(dev, NV04_PBUS_PCI_NV_1, saved_pci_nv_1);
}

int
nouveau_mem_init_agp(struct drm_device *dev)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	struct drm_agp_info info;
	struct drm_agp_mode mode;
	int ret;

	if (nouveau_noagp)
		return 0;

	nouveau_mem_reset_agp(dev);

	if (!dev->agp->acquired) {
		ret = drm_agp_acquire(dev);
		if (ret) {
			NV_ERROR(dev, "Unable to acquire AGP: %d\n", ret);
			return ret;
		}
	}

	ret = drm_agp_info(dev, &info);
	if (ret) {
		NV_ERROR(dev, "Unable to get AGP info: %d\n", ret);
		return ret;
	}

	/* see agp.h for the AGPSTAT_* modes available */
	mode.mode = info.mode;
	ret = drm_agp_enable(dev, mode);
	if (ret) {
		NV_ERROR(dev, "Unable to enable AGP: %d\n", ret);
		return ret;
	}

	dev_priv->gart_info.type	= NOUVEAU_GART_AGP;
	dev_priv->gart_info.aper_base	= info.aperture_base;
	dev_priv->gart_info.aper_size	= info.aperture_size;
	return 0;
}

int
nouveau_mem_init(struct drm_device *dev)
{
	struct drm_nouveau_private *dev_priv = dev->dev_private;
	struct ttm_bo_device *bdev = &dev_priv->ttm.bdev;
	int ret, dma_bits = 32;

	dev_priv->fb_phys = drm_get_resource_start(dev, 1);
	dev_priv->gart_info.type = NOUVEAU_GART_NONE;

	if (dev_priv->card_type >= NV_50 &&
	    pci_dma_supported(dev->pdev, DMA_BIT_MASK(40)))
		dma_bits = 40;

	ret = pci_set_dma_mask(dev->pdev, DMA_BIT_MASK(dma_bits));
	if (ret) {
		NV_ERROR(dev, "Error setting DMA mask: %d\n", ret);
		return ret;
	}

	ret = nouveau_ttm_global_init(dev_priv);
	if (ret)
		return ret;

	ret = ttm_bo_device_init(&dev_priv->ttm.bdev,
				 dev_priv->ttm.bo_global_ref.ref.object,
				 &nouveau_bo_driver, DRM_FILE_PAGE_OFFSET,
				 dma_bits <= 32 ? true : false);
	if (ret) {
		NV_ERROR(dev, "Error initialising bo driver: %d\n", ret);
		return ret;
	}

	INIT_LIST_HEAD(&dev_priv->ttm.bo_list);
	spin_lock_init(&dev_priv->ttm.bo_list_lock);

	dev_priv->fb_available_size = nouveau_mem_fb_amount(dev);

	dev_priv->fb_mappable_pages = dev_priv->fb_available_size;
	if (dev_priv->fb_mappable_pages > drm_get_resource_len(dev, 1))
		dev_priv->fb_mappable_pages = drm_get_resource_len(dev, 1);
	dev_priv->fb_mappable_pages >>= PAGE_SHIFT;

	NV_INFO(dev, "%d MiB VRAM\n", (int)(dev_priv->fb_available_size >> 20));

	/* remove reserved space at end of vram from available amount */
	dev_priv->fb_available_size -= dev_priv->ramin_rsvd_vram;
	dev_priv->fb_aper_free = dev_priv->fb_available_size;

	/* mappable vram */
	ret = ttm_bo_init_mm(bdev, TTM_PL_VRAM,
			     dev_priv->fb_available_size >> PAGE_SHIFT);
	if (ret) {
		NV_ERROR(dev, "Failed VRAM mm init: %d\n", ret);
		return ret;
	}

	/* GART */
#if !defined(__powerpc__) && !defined(__ia64__)
	if (drm_device_is_agp(dev) && dev->agp) {
		ret = nouveau_mem_init_agp(dev);
		if (ret)
			NV_ERROR(dev, "Error initialising AGP: %d\n", ret);
	}
#endif

	if (dev_priv->gart_info.type == NOUVEAU_GART_NONE) {
		ret = nouveau_sgdma_init(dev);
		if (ret) {
			NV_ERROR(dev, "Error initialising PCI(E): %d\n", ret);
			return ret;
		}
	}

	NV_INFO(dev, "%d MiB GART (aperture)\n",
		(int)(dev_priv->gart_info.aper_size >> 20));
	dev_priv->gart_info.aper_free = dev_priv->gart_info.aper_size;

	ret = ttm_bo_init_mm(bdev, TTM_PL_TT,
			     dev_priv->gart_info.aper_size >> PAGE_SHIFT);
	if (ret) {
		NV_ERROR(dev, "Failed TT mm init: %d\n", ret);
		return ret;
	}

	dev_priv->fb_mtrr = drm_mtrr_add(drm_get_resource_start(dev, 1),
					 drm_get_resource_len(dev, 1),
					 DRM_MTRR_WC);
	return 0;
}


OpenPOWER on IntegriCloud