amdgpu_object.c 17 KB
Newer Older
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
/*
 * Copyright 2009 Jerome Glisse.
 * All Rights Reserved.
 *
 * 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, sub license, 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 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 NON-INFRINGEMENT. IN NO EVENT SHALL
 * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS 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.
 *
 * The above copyright notice and this permission notice (including the
 * next paragraph) shall be included in all copies or substantial portions
 * of the Software.
 *
 */
/*
 * Authors:
 *    Jerome Glisse <glisse@freedesktop.org>
 *    Thomas Hellstrom <thomas-at-tungstengraphics-dot-com>
 *    Dave Airlie
 */
#include <linux/list.h>
#include <linux/slab.h>
#include <drm/drmP.h>
#include <drm/amdgpu_drm.h>
#include "amdgpu.h"
#include "amdgpu_trace.h"


int amdgpu_ttm_init(struct amdgpu_device *adev);
void amdgpu_ttm_fini(struct amdgpu_device *adev);

static u64 amdgpu_get_vis_part_size(struct amdgpu_device *adev,
44
						struct ttm_mem_reg *mem)
45 46 47 48 49
{
	u64 ret = 0;
	if (mem->start << PAGE_SHIFT < adev->mc.visible_vram_size) {
		ret = (u64)((mem->start << PAGE_SHIFT) + mem->size) >
			   adev->mc.visible_vram_size ?
50
			   adev->mc.visible_vram_size - (mem->start << PAGE_SHIFT) :
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
			   mem->size;
	}
	return ret;
}

static void amdgpu_update_memory_usage(struct amdgpu_device *adev,
		       struct ttm_mem_reg *old_mem,
		       struct ttm_mem_reg *new_mem)
{
	u64 vis_size;
	if (!adev)
		return;

	if (new_mem) {
		switch (new_mem->mem_type) {
		case TTM_PL_TT:
			atomic64_add(new_mem->size, &adev->gtt_usage);
			break;
		case TTM_PL_VRAM:
			atomic64_add(new_mem->size, &adev->vram_usage);
			vis_size = amdgpu_get_vis_part_size(adev, new_mem);
			atomic64_add(vis_size, &adev->vram_vis_usage);
			break;
		}
	}

	if (old_mem) {
		switch (old_mem->mem_type) {
		case TTM_PL_TT:
			atomic64_sub(old_mem->size, &adev->gtt_usage);
			break;
		case TTM_PL_VRAM:
			atomic64_sub(old_mem->size, &adev->vram_usage);
			vis_size = amdgpu_get_vis_part_size(adev, old_mem);
			atomic64_sub(vis_size, &adev->vram_vis_usage);
			break;
		}
	}
}

static void amdgpu_ttm_bo_destroy(struct ttm_buffer_object *tbo)
{
	struct amdgpu_bo *bo;

	bo = container_of(tbo, struct amdgpu_bo, tbo);

	amdgpu_update_memory_usage(bo->adev, &bo->tbo.mem, NULL);

	mutex_lock(&bo->adev->gem.mutex);
	list_del_init(&bo->list);
	mutex_unlock(&bo->adev->gem.mutex);
	drm_gem_object_release(&bo->gem_base);
	kfree(bo->metadata);
	kfree(bo);
}

bool amdgpu_ttm_bo_is_amdgpu_bo(struct ttm_buffer_object *bo)
{
	if (bo->destroy == &amdgpu_ttm_bo_destroy)
		return true;
	return false;
}

114 115 116 117
static void amdgpu_ttm_placement_init(struct amdgpu_device *adev,
				      struct ttm_placement *placement,
				      struct ttm_place *placements,
				      u32 domain, u64 flags)
118 119
{
	u32 c = 0, i;
120 121 122

	placement->placement = placements;
	placement->busy_placement = placements;
123 124

	if (domain & AMDGPU_GEM_DOMAIN_VRAM) {
125 126 127 128 129 130
		if (flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS &&
			adev->mc.visible_vram_size < adev->mc.real_vram_size) {
			placements[c].fpfn =
				adev->mc.visible_vram_size >> PAGE_SHIFT;
			placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED |
				TTM_PL_FLAG_VRAM;
131
		}
132 133 134
		placements[c].fpfn = 0;
		placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED |
			TTM_PL_FLAG_VRAM;
135 136 137
	}

	if (domain & AMDGPU_GEM_DOMAIN_GTT) {
138 139 140 141
		if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC) {
			placements[c].fpfn = 0;
			placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_TT |
				TTM_PL_FLAG_UNCACHED;
142
		} else {
143 144
			placements[c].fpfn = 0;
			placements[c++].flags = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_TT;
145 146 147 148
		}
	}

	if (domain & AMDGPU_GEM_DOMAIN_CPU) {
149 150 151 152
		if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC) {
			placements[c].fpfn = 0;
			placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_SYSTEM |
				TTM_PL_FLAG_UNCACHED;
153
		} else {
154 155
			placements[c].fpfn = 0;
			placements[c++].flags = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_SYSTEM;
156 157 158 159
		}
	}

	if (domain & AMDGPU_GEM_DOMAIN_GDS) {
160 161 162
		placements[c].fpfn = 0;
		placements[c++].flags = TTM_PL_FLAG_UNCACHED |
			AMDGPU_PL_FLAG_GDS;
163 164
	}
	if (domain & AMDGPU_GEM_DOMAIN_GWS) {
165 166 167
		placements[c].fpfn = 0;
		placements[c++].flags = TTM_PL_FLAG_UNCACHED |
			AMDGPU_PL_FLAG_GWS;
168 169
	}
	if (domain & AMDGPU_GEM_DOMAIN_OA) {
170 171 172
		placements[c].fpfn = 0;
		placements[c++].flags = TTM_PL_FLAG_UNCACHED |
			AMDGPU_PL_FLAG_OA;
173 174 175
	}

	if (!c) {
176 177 178
		placements[c].fpfn = 0;
		placements[c++].flags = TTM_PL_MASK_CACHING |
			TTM_PL_FLAG_SYSTEM;
179
	}
180 181
	placement->num_placement = c;
	placement->num_busy_placement = c;
182 183

	for (i = 0; i < c; i++) {
184 185 186 187 188
		if ((flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) &&
			(placements[i].flags & TTM_PL_FLAG_VRAM) &&
			!placements[i].fpfn)
			placements[i].lpfn =
				adev->mc.visible_vram_size >> PAGE_SHIFT;
189
		else
190
			placements[i].lpfn = 0;
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
void amdgpu_ttm_placement_from_domain(struct amdgpu_bo *rbo, u32 domain)
{
	amdgpu_ttm_placement_init(rbo->adev, &rbo->placement,
				  rbo->placements, domain, rbo->flags);
}

static void amdgpu_fill_placement_to_bo(struct amdgpu_bo *bo,
					struct ttm_placement *placement)
{
	BUG_ON(placement->num_placement > (AMDGPU_GEM_DOMAIN_MAX + 1));

	memcpy(bo->placements, placement->placement,
	       placement->num_placement * sizeof(struct ttm_place));
	bo->placement.num_placement = placement->num_placement;
	bo->placement.num_busy_placement = placement->num_busy_placement;
	bo->placement.placement = bo->placements;
	bo->placement.busy_placement = bo->placements;
}

int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
				unsigned long size, int byte_align,
				bool kernel, u32 domain, u64 flags,
				struct sg_table *sg,
				struct ttm_placement *placement,
				struct amdgpu_bo **bo_ptr)
219 220 221 222 223 224 225 226 227 228 229
{
	struct amdgpu_bo *bo;
	enum ttm_bo_type type;
	unsigned long page_align;
	size_t acc_size;
	int r;

	/* VI has a hw bug where VM PTEs have to be allocated in groups of 8.
	 * do this as a temporary workaround
	 */
	if (!(domain & (AMDGPU_GEM_DOMAIN_GDS | AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA))) {
230
		if ((adev->asic_type >= CHIP_TOPAZ) && (adev->asic_type != CHIP_FIJI)) {
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
			if (byte_align & 0x7fff)
				byte_align = ALIGN(byte_align, 0x8000);
			if (size & 0x7fff)
				size = ALIGN(size, 0x8000);
		}
	}

	page_align = roundup(byte_align, PAGE_SIZE) >> PAGE_SHIFT;
	size = ALIGN(size, PAGE_SIZE);

	if (kernel) {
		type = ttm_bo_type_kernel;
	} else if (sg) {
		type = ttm_bo_type_sg;
	} else {
		type = ttm_bo_type_device;
	}
	*bo_ptr = NULL;

	acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size,
				       sizeof(struct amdgpu_bo));

	bo = kzalloc(sizeof(struct amdgpu_bo), GFP_KERNEL);
	if (bo == NULL)
		return -ENOMEM;
	r = drm_gem_object_init(adev->ddev, &bo->gem_base, size);
	if (unlikely(r)) {
		kfree(bo);
		return r;
	}
	bo->adev = adev;
	INIT_LIST_HEAD(&bo->list);
	INIT_LIST_HEAD(&bo->va);
	bo->initial_domain = domain & (AMDGPU_GEM_DOMAIN_VRAM |
				       AMDGPU_GEM_DOMAIN_GTT |
				       AMDGPU_GEM_DOMAIN_CPU |
				       AMDGPU_GEM_DOMAIN_GDS |
				       AMDGPU_GEM_DOMAIN_GWS |
				       AMDGPU_GEM_DOMAIN_OA);

	bo->flags = flags;
272
	amdgpu_fill_placement_to_bo(bo, placement);
273 274 275 276 277 278 279 280 281 282 283 284 285 286
	/* Kernel allocation are uninterruptible */
	r = ttm_bo_init(&adev->mman.bdev, &bo->tbo, size, type,
			&bo->placement, page_align, !kernel, NULL,
			acc_size, sg, NULL, &amdgpu_ttm_bo_destroy);
	if (unlikely(r != 0)) {
		return r;
	}
	*bo_ptr = bo;

	trace_amdgpu_bo_create(bo);

	return 0;
}

287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307
int amdgpu_bo_create(struct amdgpu_device *adev,
		     unsigned long size, int byte_align,
		     bool kernel, u32 domain, u64 flags,
		     struct sg_table *sg, struct amdgpu_bo **bo_ptr)
{
	struct ttm_placement placement = {0};
	struct ttm_place placements[AMDGPU_GEM_DOMAIN_MAX + 1];

	memset(&placements, 0,
	       (AMDGPU_GEM_DOMAIN_MAX + 1) * sizeof(struct ttm_place));

	amdgpu_ttm_placement_init(adev, &placement,
				  placements, domain, flags);

	return amdgpu_bo_create_restricted(adev, size, byte_align,
					   kernel, domain, flags,
					   sg,
					   &placement,
					   bo_ptr);
}

308 309 310 311 312
int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr)
{
	bool is_iomem;
	int r;

313 314 315
	if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)
		return -EPERM;

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
	if (bo->kptr) {
		if (ptr) {
			*ptr = bo->kptr;
		}
		return 0;
	}
	r = ttm_bo_kmap(&bo->tbo, 0, bo->tbo.num_pages, &bo->kmap);
	if (r) {
		return r;
	}
	bo->kptr = ttm_kmap_obj_virtual(&bo->kmap, &is_iomem);
	if (ptr) {
		*ptr = bo->kptr;
	}
	return 0;
}

void amdgpu_bo_kunmap(struct amdgpu_bo *bo)
{
	if (bo->kptr == NULL)
		return;
	bo->kptr = NULL;
	ttm_bo_kunmap(&bo->kmap);
}

struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo)
{
	if (bo == NULL)
		return NULL;

	ttm_bo_reference(&bo->tbo);
	return bo;
}

void amdgpu_bo_unref(struct amdgpu_bo **bo)
{
	struct ttm_buffer_object *tbo;

	if ((*bo) == NULL)
		return;

	tbo = &((*bo)->tbo);
	ttm_bo_unref(&tbo);
	if (tbo == NULL)
		*bo = NULL;
}

363 364
int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
			     u64 min_offset, u64 max_offset,
365 366 367
			     u64 *gpu_addr)
{
	int r, i;
368
	unsigned fpfn, lpfn;
369 370 371 372

	if (amdgpu_ttm_tt_has_userptr(bo->tbo.ttm))
		return -EPERM;

373 374 375
	if (WARN_ON_ONCE(min_offset > max_offset))
		return -EINVAL;

376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396
	if (bo->pin_count) {
		bo->pin_count++;
		if (gpu_addr)
			*gpu_addr = amdgpu_bo_gpu_offset(bo);

		if (max_offset != 0) {
			u64 domain_start;
			if (domain == AMDGPU_GEM_DOMAIN_VRAM)
				domain_start = bo->adev->mc.vram_start;
			else
				domain_start = bo->adev->mc.gtt_start;
			WARN_ON_ONCE(max_offset <
				     (amdgpu_bo_gpu_offset(bo) - domain_start));
		}

		return 0;
	}
	amdgpu_ttm_placement_from_domain(bo, domain);
	for (i = 0; i < bo->placement.num_placement; i++) {
		/* force to pin into visible video ram */
		if ((bo->placements[i].flags & TTM_PL_FLAG_VRAM) &&
397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
		    !(bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) &&
		    (!max_offset || max_offset > bo->adev->mc.visible_vram_size)) {
			if (WARN_ON_ONCE(min_offset >
					 bo->adev->mc.visible_vram_size))
				return -EINVAL;
			fpfn = min_offset >> PAGE_SHIFT;
			lpfn = bo->adev->mc.visible_vram_size >> PAGE_SHIFT;
		} else {
			fpfn = min_offset >> PAGE_SHIFT;
			lpfn = max_offset >> PAGE_SHIFT;
		}
		if (fpfn > bo->placements[i].fpfn)
			bo->placements[i].fpfn = fpfn;
		if (lpfn && lpfn < bo->placements[i].lpfn)
			bo->placements[i].lpfn = lpfn;
412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431
		bo->placements[i].flags |= TTM_PL_FLAG_NO_EVICT;
	}

	r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
	if (likely(r == 0)) {
		bo->pin_count = 1;
		if (gpu_addr != NULL)
			*gpu_addr = amdgpu_bo_gpu_offset(bo);
		if (domain == AMDGPU_GEM_DOMAIN_VRAM)
			bo->adev->vram_pin_size += amdgpu_bo_size(bo);
		else
			bo->adev->gart_pin_size += amdgpu_bo_size(bo);
	} else {
		dev_err(bo->adev->dev, "%p pin failed\n", bo);
	}
	return r;
}

int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain, u64 *gpu_addr)
{
432
	return amdgpu_bo_pin_restricted(bo, domain, 0, 0, gpu_addr);
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
}

int amdgpu_bo_unpin(struct amdgpu_bo *bo)
{
	int r, i;

	if (!bo->pin_count) {
		dev_warn(bo->adev->dev, "%p unpin not necessary\n", bo);
		return 0;
	}
	bo->pin_count--;
	if (bo->pin_count)
		return 0;
	for (i = 0; i < bo->placement.num_placement; i++) {
		bo->placements[i].lpfn = 0;
		bo->placements[i].flags &= ~TTM_PL_FLAG_NO_EVICT;
	}
	r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
	if (likely(r == 0)) {
		if (bo->tbo.mem.mem_type == TTM_PL_VRAM)
			bo->adev->vram_pin_size -= amdgpu_bo_size(bo);
		else
			bo->adev->gart_pin_size -= amdgpu_bo_size(bo);
	} else {
		dev_err(bo->adev->dev, "%p validate failed for unpin\n", bo);
	}
	return r;
}

int amdgpu_bo_evict_vram(struct amdgpu_device *adev)
{
	/* late 2.6.33 fix IGP hibernate - we need pm ops to do this correct */
465
	if (0 && (adev->flags & AMD_IS_APU)) {
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
		/* Useless to evict on IGP chips */
		return 0;
	}
	return ttm_bo_evict_mm(&adev->mman.bdev, TTM_PL_VRAM);
}

void amdgpu_bo_force_delete(struct amdgpu_device *adev)
{
	struct amdgpu_bo *bo, *n;

	if (list_empty(&adev->gem.objects)) {
		return;
	}
	dev_err(adev->dev, "Userspace still has active objects !\n");
	list_for_each_entry_safe(bo, n, &adev->gem.objects, list) {
		mutex_lock(&adev->ddev->struct_mutex);
		dev_err(adev->dev, "%p %p %lu %lu force free\n",
			&bo->gem_base, bo, (unsigned long)bo->gem_base.size,
			*((unsigned long *)&bo->gem_base.refcount));
		mutex_lock(&bo->adev->gem.mutex);
		list_del_init(&bo->list);
		mutex_unlock(&bo->adev->gem.mutex);
		/* this should unref the ttm bo */
		drm_gem_object_unreference(&bo->gem_base);
		mutex_unlock(&adev->ddev->struct_mutex);
	}
}

int amdgpu_bo_init(struct amdgpu_device *adev)
{
	/* Add an MTRR for the VRAM */
	adev->mc.vram_mtrr = arch_phys_wc_add(adev->mc.aper_base,
					      adev->mc.aper_size);
	DRM_INFO("Detected VRAM RAM=%lluM, BAR=%lluM\n",
		adev->mc.mc_vram_size >> 20,
		(unsigned long long)adev->mc.aper_size >> 20);
	DRM_INFO("RAM width %dbits DDR\n",
			adev->mc.vram_width);
	return amdgpu_ttm_init(adev);
}

void amdgpu_bo_fini(struct amdgpu_device *adev)
{
	amdgpu_ttm_fini(adev);
	arch_phys_wc_del(adev->mc.vram_mtrr);
}

int amdgpu_bo_fbdev_mmap(struct amdgpu_bo *bo,
			     struct vm_area_struct *vma)
{
	return ttm_fbdev_mmap(vma, &bo->tbo);
}

int amdgpu_bo_set_tiling_flags(struct amdgpu_bo *bo, u64 tiling_flags)
{
521
	if (AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT) > 6)
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
		return -EINVAL;

	bo->tiling_flags = tiling_flags;
	return 0;
}

void amdgpu_bo_get_tiling_flags(struct amdgpu_bo *bo, u64 *tiling_flags)
{
	lockdep_assert_held(&bo->tbo.resv->lock.base);

	if (tiling_flags)
		*tiling_flags = bo->tiling_flags;
}

int amdgpu_bo_set_metadata (struct amdgpu_bo *bo, void *metadata,
			    uint32_t metadata_size, uint64_t flags)
{
	void *buffer;

	if (!metadata_size) {
		if (bo->metadata_size) {
			kfree(bo->metadata);
			bo->metadata_size = 0;
		}
		return 0;
	}

	if (metadata == NULL)
		return -EINVAL;

	buffer = kzalloc(metadata_size, GFP_KERNEL);
	if (buffer == NULL)
		return -ENOMEM;

	memcpy(buffer, metadata, metadata_size);

	kfree(bo->metadata);
	bo->metadata_flags = flags;
	bo->metadata = buffer;
	bo->metadata_size = metadata_size;

	return 0;
}

int amdgpu_bo_get_metadata(struct amdgpu_bo *bo, void *buffer,
			   size_t buffer_size, uint32_t *metadata_size,
			   uint64_t *flags)
{
	if (!buffer && !metadata_size)
		return -EINVAL;

	if (buffer) {
		if (buffer_size < bo->metadata_size)
			return -EINVAL;

		if (bo->metadata_size)
			memcpy(buffer, bo->metadata, bo->metadata_size);
	}

	if (metadata_size)
		*metadata_size = bo->metadata_size;
	if (flags)
		*flags = bo->metadata_flags;

	return 0;
}

void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
			   struct ttm_mem_reg *new_mem)
{
	struct amdgpu_bo *rbo;

	if (!amdgpu_ttm_bo_is_amdgpu_bo(bo))
		return;

	rbo = container_of(bo, struct amdgpu_bo, tbo);
	amdgpu_vm_bo_invalidate(rbo->adev, rbo);

	/* update statistics */
	if (!new_mem)
		return;

	/* move_notify is called before move happens */
	amdgpu_update_memory_usage(rbo->adev, &bo->mem, new_mem);
}

int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
{
	struct amdgpu_device *adev;
611 612 613
	struct amdgpu_bo *abo;
	unsigned long offset, size, lpfn;
	int i, r;
614 615 616

	if (!amdgpu_ttm_bo_is_amdgpu_bo(bo))
		return 0;
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

	abo = container_of(bo, struct amdgpu_bo, tbo);
	adev = abo->adev;
	if (bo->mem.mem_type != TTM_PL_VRAM)
		return 0;

	size = bo->mem.num_pages << PAGE_SHIFT;
	offset = bo->mem.start << PAGE_SHIFT;
	if ((offset + size) <= adev->mc.visible_vram_size)
		return 0;

	/* hurrah the memory is not visible ! */
	amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM);
	lpfn =	adev->mc.visible_vram_size >> PAGE_SHIFT;
	for (i = 0; i < abo->placement.num_placement; i++) {
		/* Force into visible VRAM */
		if ((abo->placements[i].flags & TTM_PL_FLAG_VRAM) &&
		    (!abo->placements[i].lpfn || abo->placements[i].lpfn > lpfn))
			abo->placements[i].lpfn = lpfn;
	}
	r = ttm_bo_validate(bo, &abo->placement, false, false);
	if (unlikely(r == -ENOMEM)) {
		amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT);
		return ttm_bo_validate(bo, &abo->placement, false, false);
	} else if (unlikely(r != 0)) {
		return r;
643
	}
644 645 646 647 648 649

	offset = bo->mem.start << PAGE_SHIFT;
	/* this should never happen */
	if ((offset + size) > adev->mc.visible_vram_size)
		return -EINVAL;

650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670
	return 0;
}

/**
 * amdgpu_bo_fence - add fence to buffer object
 *
 * @bo: buffer object in question
 * @fence: fence to add
 * @shared: true if fence should be added shared
 *
 */
void amdgpu_bo_fence(struct amdgpu_bo *bo, struct amdgpu_fence *fence,
		     bool shared)
{
	struct reservation_object *resv = bo->tbo.resv;

	if (shared)
		reservation_object_add_shared_fence(resv, &fence->base);
	else
		reservation_object_add_excl_fence(resv, &fence->base);
}