radeon_ttm.c 22.9 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
/*
 * 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 <ttm/ttm_bo_api.h>
#include <ttm/ttm_bo_driver.h>
#include <ttm/ttm_placement.h>
#include <ttm/ttm_module.h>
36
#include <ttm/ttm_page_alloc.h>
37 38
#include <drm/drmP.h>
#include <drm/radeon_drm.h>
39
#include <linux/seq_file.h>
40
#include <linux/slab.h>
41 42 43 44 45
#include "radeon_reg.h"
#include "radeon.h"

#define DRM_FILE_PAGE_OFFSET (0x100000000ULL >> PAGE_SHIFT)

46 47
static int radeon_ttm_debugfs_init(struct radeon_device *rdev);

48 49 50 51 52 53 54 55 56 57 58 59 60 61
static struct radeon_device *radeon_get_rdev(struct ttm_bo_device *bdev)
{
	struct radeon_mman *mman;
	struct radeon_device *rdev;

	mman = container_of(bdev, struct radeon_mman, bdev);
	rdev = container_of(mman, struct radeon_device, mman);
	return rdev;
}


/*
 * Global memory.
 */
62
static int radeon_ttm_mem_global_init(struct drm_global_reference *ref)
63 64 65 66
{
	return ttm_mem_global_init(ref->object);
}

67
static void radeon_ttm_mem_global_release(struct drm_global_reference *ref)
68 69 70 71 72 73
{
	ttm_mem_global_release(ref->object);
}

static int radeon_ttm_global_init(struct radeon_device *rdev)
{
74
	struct drm_global_reference *global_ref;
75 76 77 78
	int r;

	rdev->mman.mem_global_referenced = false;
	global_ref = &rdev->mman.mem_global_ref;
79
	global_ref->global_type = DRM_GLOBAL_TTM_MEM;
80 81 82
	global_ref->size = sizeof(struct ttm_mem_global);
	global_ref->init = &radeon_ttm_mem_global_init;
	global_ref->release = &radeon_ttm_mem_global_release;
83
	r = drm_global_item_ref(global_ref);
84
	if (r != 0) {
85 86
		DRM_ERROR("Failed setting up TTM memory accounting "
			  "subsystem.\n");
87 88
		return r;
	}
89 90 91 92

	rdev->mman.bo_global_ref.mem_glob =
		rdev->mman.mem_global_ref.object;
	global_ref = &rdev->mman.bo_global_ref.ref;
93
	global_ref->global_type = DRM_GLOBAL_TTM_BO;
94
	global_ref->size = sizeof(struct ttm_bo_global);
95 96
	global_ref->init = &ttm_bo_global_init;
	global_ref->release = &ttm_bo_global_release;
97
	r = drm_global_item_ref(global_ref);
98 99
	if (r != 0) {
		DRM_ERROR("Failed setting up TTM BO subsystem.\n");
100
		drm_global_item_unref(&rdev->mman.mem_global_ref);
101 102 103
		return r;
	}

104 105 106 107 108 109 110
	rdev->mman.mem_global_referenced = true;
	return 0;
}

static void radeon_ttm_global_fini(struct radeon_device *rdev)
{
	if (rdev->mman.mem_global_referenced) {
111 112
		drm_global_item_unref(&rdev->mman.bo_global_ref.ref);
		drm_global_item_unref(&rdev->mman.mem_global_ref);
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
		rdev->mman.mem_global_referenced = false;
	}
}

struct ttm_backend *radeon_ttm_backend_create(struct radeon_device *rdev);

static struct ttm_backend*
radeon_create_ttm_backend_entry(struct ttm_bo_device *bdev)
{
	struct radeon_device *rdev;

	rdev = radeon_get_rdev(bdev);
#if __OS_HAS_AGP
	if (rdev->flags & RADEON_IS_AGP) {
		return ttm_agp_backend_init(bdev, rdev->ddev->agp->bridge);
	} else
#endif
	{
		return radeon_ttm_backend_create(rdev);
	}
}

static int radeon_invalidate_caches(struct ttm_bo_device *bdev, uint32_t flags)
{
	return 0;
}

static int radeon_init_mem_type(struct ttm_bo_device *bdev, uint32_t type,
				struct ttm_mem_type_manager *man)
{
	struct radeon_device *rdev;

	rdev = radeon_get_rdev(bdev);

	switch (type) {
	case TTM_PL_SYSTEM:
		/* System memory */
		man->flags = TTM_MEMTYPE_FLAG_MAPPABLE;
		man->available_caching = TTM_PL_MASK_CACHING;
		man->default_caching = TTM_PL_FLAG_CACHED;
		break;
	case TTM_PL_TT:
155
		man->func = &ttm_bo_manager_func;
156
		man->gpu_offset = rdev->mc.gtt_start;
157 158
		man->available_caching = TTM_PL_MASK_CACHING;
		man->default_caching = TTM_PL_FLAG_CACHED;
159
		man->flags = TTM_MEMTYPE_FLAG_MAPPABLE | TTM_MEMTYPE_FLAG_CMA;
160 161 162 163 164 165 166
#if __OS_HAS_AGP
		if (rdev->flags & RADEON_IS_AGP) {
			if (!(drm_core_has_AGP(rdev->ddev) && rdev->ddev->agp)) {
				DRM_ERROR("AGP is not enabled for memory type %u\n",
					  (unsigned)type);
				return -EINVAL;
			}
167
			if (!rdev->ddev->agp->cant_use_aperture)
168
				man->flags = TTM_MEMTYPE_FLAG_MAPPABLE;
169 170 171 172
			man->available_caching = TTM_PL_FLAG_UNCACHED |
						 TTM_PL_FLAG_WC;
			man->default_caching = TTM_PL_FLAG_WC;
		}
173
#endif
174 175 176
		break;
	case TTM_PL_VRAM:
		/* "On-card" video ram */
177
		man->func = &ttm_bo_manager_func;
178
		man->gpu_offset = rdev->mc.vram_start;
179 180 181 182 183 184 185 186 187 188 189 190
		man->flags = TTM_MEMTYPE_FLAG_FIXED |
			     TTM_MEMTYPE_FLAG_MAPPABLE;
		man->available_caching = TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_WC;
		man->default_caching = TTM_PL_FLAG_WC;
		break;
	default:
		DRM_ERROR("Unsupported memory type %u\n", (unsigned)type);
		return -EINVAL;
	}
	return 0;
}

191 192
static void radeon_evict_flags(struct ttm_buffer_object *bo,
				struct ttm_placement *placement)
193
{
194 195 196 197 198 199 200 201 202 203 204 205 206
	struct radeon_bo *rbo;
	static u32 placements = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM;

	if (!radeon_ttm_bo_is_radeon_bo(bo)) {
		placement->fpfn = 0;
		placement->lpfn = 0;
		placement->placement = &placements;
		placement->busy_placement = &placements;
		placement->num_placement = 1;
		placement->num_busy_placement = 1;
		return;
	}
	rbo = container_of(bo, struct radeon_bo, tbo);
207
	switch (bo->mem.mem_type) {
208
	case TTM_PL_VRAM:
209 210 211 212
		if (rbo->rdev->cp.ready == false)
			radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_CPU);
		else
			radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_GTT);
213 214
		break;
	case TTM_PL_TT:
215
	default:
216
		radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_CPU);
217
	}
218
	*placement = rbo->placement;
219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236
}

static int radeon_verify_access(struct ttm_buffer_object *bo, struct file *filp)
{
	return 0;
}

static void radeon_move_null(struct ttm_buffer_object *bo,
			     struct ttm_mem_reg *new_mem)
{
	struct ttm_mem_reg *old_mem = &bo->mem;

	BUG_ON(old_mem->mm_node != NULL);
	*old_mem = *new_mem;
	new_mem->mm_node = NULL;
}

static int radeon_move_blit(struct ttm_buffer_object *bo,
237 238 239
			bool evict, int no_wait_reserve, bool no_wait_gpu,
			struct ttm_mem_reg *new_mem,
			struct ttm_mem_reg *old_mem)
240 241 242 243 244 245 246 247 248 249 250
{
	struct radeon_device *rdev;
	uint64_t old_start, new_start;
	struct radeon_fence *fence;
	int r;

	rdev = radeon_get_rdev(bo->bdev);
	r = radeon_fence_create(rdev, &fence);
	if (unlikely(r)) {
		return r;
	}
251 252
	old_start = old_mem->start << PAGE_SHIFT;
	new_start = new_mem->start << PAGE_SHIFT;
253 254 255

	switch (old_mem->mem_type) {
	case TTM_PL_VRAM:
256
		old_start += rdev->mc.vram_start;
257 258
		break;
	case TTM_PL_TT:
259
		old_start += rdev->mc.gtt_start;
260 261 262 263 264 265 266
		break;
	default:
		DRM_ERROR("Unknown placement %d\n", old_mem->mem_type);
		return -EINVAL;
	}
	switch (new_mem->mem_type) {
	case TTM_PL_VRAM:
267
		new_start += rdev->mc.vram_start;
268 269
		break;
	case TTM_PL_TT:
270
		new_start += rdev->mc.gtt_start;
271 272 273 274 275 276 277 278 279
		break;
	default:
		DRM_ERROR("Unknown placement %d\n", old_mem->mem_type);
		return -EINVAL;
	}
	if (!rdev->cp.ready) {
		DRM_ERROR("Trying to move memory with CP turned off.\n");
		return -EINVAL;
	}
280 281 282 283 284 285

	BUILD_BUG_ON((PAGE_SIZE % RADEON_GPU_PAGE_SIZE) != 0);

	r = radeon_copy(rdev, old_start, new_start,
			new_mem->num_pages * (PAGE_SIZE / RADEON_GPU_PAGE_SIZE), /* GPU pages */
			fence);
286 287
	/* FIXME: handle copy error */
	r = ttm_bo_move_accel_cleanup(bo, (void *)fence, NULL,
288
				      evict, no_wait_reserve, no_wait_gpu, new_mem);
289 290 291 292 293
	radeon_fence_unref(&fence);
	return r;
}

static int radeon_move_vram_ram(struct ttm_buffer_object *bo,
294 295
				bool evict, bool interruptible,
				bool no_wait_reserve, bool no_wait_gpu,
296 297 298 299 300
				struct ttm_mem_reg *new_mem)
{
	struct radeon_device *rdev;
	struct ttm_mem_reg *old_mem = &bo->mem;
	struct ttm_mem_reg tmp_mem;
301 302
	u32 placements;
	struct ttm_placement placement;
303 304 305 306 307
	int r;

	rdev = radeon_get_rdev(bo->bdev);
	tmp_mem = *new_mem;
	tmp_mem.mm_node = NULL;
308 309 310 311 312 313 314 315
	placement.fpfn = 0;
	placement.lpfn = 0;
	placement.num_placement = 1;
	placement.placement = &placements;
	placement.num_busy_placement = 1;
	placement.busy_placement = &placements;
	placements = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
	r = ttm_bo_mem_space(bo, &placement, &tmp_mem,
316
			     interruptible, no_wait_reserve, no_wait_gpu);
317 318 319
	if (unlikely(r)) {
		return r;
	}
320 321 322 323 324 325

	r = ttm_tt_set_placement_caching(bo->ttm, tmp_mem.placement);
	if (unlikely(r)) {
		goto out_cleanup;
	}

326 327 328 329
	r = ttm_tt_bind(bo->ttm, &tmp_mem);
	if (unlikely(r)) {
		goto out_cleanup;
	}
330
	r = radeon_move_blit(bo, true, no_wait_reserve, no_wait_gpu, &tmp_mem, old_mem);
331 332 333
	if (unlikely(r)) {
		goto out_cleanup;
	}
334
	r = ttm_bo_move_ttm(bo, true, no_wait_reserve, no_wait_gpu, new_mem);
335
out_cleanup:
336
	ttm_bo_mem_put(bo, &tmp_mem);
337 338 339 340
	return r;
}

static int radeon_move_ram_vram(struct ttm_buffer_object *bo,
341 342
				bool evict, bool interruptible,
				bool no_wait_reserve, bool no_wait_gpu,
343 344 345 346 347
				struct ttm_mem_reg *new_mem)
{
	struct radeon_device *rdev;
	struct ttm_mem_reg *old_mem = &bo->mem;
	struct ttm_mem_reg tmp_mem;
348 349
	struct ttm_placement placement;
	u32 placements;
350 351 352 353 354
	int r;

	rdev = radeon_get_rdev(bo->bdev);
	tmp_mem = *new_mem;
	tmp_mem.mm_node = NULL;
355 356 357 358 359 360 361
	placement.fpfn = 0;
	placement.lpfn = 0;
	placement.num_placement = 1;
	placement.placement = &placements;
	placement.num_busy_placement = 1;
	placement.busy_placement = &placements;
	placements = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
362
	r = ttm_bo_mem_space(bo, &placement, &tmp_mem, interruptible, no_wait_reserve, no_wait_gpu);
363 364 365
	if (unlikely(r)) {
		return r;
	}
366
	r = ttm_bo_move_ttm(bo, true, no_wait_reserve, no_wait_gpu, &tmp_mem);
367 368 369
	if (unlikely(r)) {
		goto out_cleanup;
	}
370
	r = radeon_move_blit(bo, true, no_wait_reserve, no_wait_gpu, new_mem, old_mem);
371 372 373 374
	if (unlikely(r)) {
		goto out_cleanup;
	}
out_cleanup:
375
	ttm_bo_mem_put(bo, &tmp_mem);
376 377 378 379
	return r;
}

static int radeon_bo_move(struct ttm_buffer_object *bo,
380 381 382
			bool evict, bool interruptible,
			bool no_wait_reserve, bool no_wait_gpu,
			struct ttm_mem_reg *new_mem)
383 384 385 386 387 388 389 390 391 392 393 394 395 396
{
	struct radeon_device *rdev;
	struct ttm_mem_reg *old_mem = &bo->mem;
	int r;

	rdev = radeon_get_rdev(bo->bdev);
	if (old_mem->mem_type == TTM_PL_SYSTEM && bo->ttm == NULL) {
		radeon_move_null(bo, new_mem);
		return 0;
	}
	if ((old_mem->mem_type == TTM_PL_TT &&
	     new_mem->mem_type == TTM_PL_SYSTEM) ||
	    (old_mem->mem_type == TTM_PL_SYSTEM &&
	     new_mem->mem_type == TTM_PL_TT)) {
397
		/* bind is enough */
398 399 400
		radeon_move_null(bo, new_mem);
		return 0;
	}
401
	if (!rdev->cp.ready || rdev->asic->copy == NULL) {
402
		/* use memcpy */
403
		goto memcpy;
404 405 406 407
	}

	if (old_mem->mem_type == TTM_PL_VRAM &&
	    new_mem->mem_type == TTM_PL_SYSTEM) {
408
		r = radeon_move_vram_ram(bo, evict, interruptible,
409
					no_wait_reserve, no_wait_gpu, new_mem);
410 411
	} else if (old_mem->mem_type == TTM_PL_SYSTEM &&
		   new_mem->mem_type == TTM_PL_VRAM) {
412
		r = radeon_move_ram_vram(bo, evict, interruptible,
413
					    no_wait_reserve, no_wait_gpu, new_mem);
414
	} else {
415
		r = radeon_move_blit(bo, evict, no_wait_reserve, no_wait_gpu, new_mem, old_mem);
416
	}
417 418 419

	if (r) {
memcpy:
420
		r = ttm_bo_move_memcpy(bo, evict, no_wait_reserve, no_wait_gpu, new_mem);
421
	}
422 423 424
	return r;
}

425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
static int radeon_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem)
{
	struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type];
	struct radeon_device *rdev = radeon_get_rdev(bdev);

	mem->bus.addr = NULL;
	mem->bus.offset = 0;
	mem->bus.size = mem->num_pages << PAGE_SHIFT;
	mem->bus.base = 0;
	mem->bus.is_iomem = false;
	if (!(man->flags & TTM_MEMTYPE_FLAG_MAPPABLE))
		return -EINVAL;
	switch (mem->mem_type) {
	case TTM_PL_SYSTEM:
		/* system memory */
		return 0;
	case TTM_PL_TT:
#if __OS_HAS_AGP
		if (rdev->flags & RADEON_IS_AGP) {
			/* RADEON_IS_AGP is set only if AGP is active */
445
			mem->bus.offset = mem->start << PAGE_SHIFT;
446
			mem->bus.base = rdev->mc.agp_base;
447
			mem->bus.is_iomem = !rdev->ddev->agp->cant_use_aperture;
448 449 450 451
		}
#endif
		break;
	case TTM_PL_VRAM:
452
		mem->bus.offset = mem->start << PAGE_SHIFT;
453 454 455 456 457
		/* check if it's visible */
		if ((mem->bus.offset + mem->bus.size) > rdev->mc.visible_vram_size)
			return -EINVAL;
		mem->bus.base = rdev->mc.aper_base;
		mem->bus.is_iomem = true;
458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480
#ifdef __alpha__
		/*
		 * Alpha: use bus.addr to hold the ioremap() return,
		 * so we can modify bus.base below.
		 */
		if (mem->placement & TTM_PL_FLAG_WC)
			mem->bus.addr =
				ioremap_wc(mem->bus.base + mem->bus.offset,
					   mem->bus.size);
		else
			mem->bus.addr =
				ioremap_nocache(mem->bus.base + mem->bus.offset,
						mem->bus.size);

		/*
		 * Alpha: Use just the bus offset plus
		 * the hose/domain memory base for bus.base.
		 * It then can be used to build PTEs for VRAM
		 * access, as done in ttm_bo_vm_fault().
		 */
		mem->bus.base = (mem->bus.base & 0x0ffffffffUL) +
			rdev->ddev->hose->dense_mem_base;
#endif
481 482 483 484 485 486 487 488 489 490 491
		break;
	default:
		return -EINVAL;
	}
	return 0;
}

static void radeon_ttm_io_mem_free(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem)
{
}

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
static int radeon_sync_obj_wait(void *sync_obj, void *sync_arg,
				bool lazy, bool interruptible)
{
	return radeon_fence_wait((struct radeon_fence *)sync_obj, interruptible);
}

static int radeon_sync_obj_flush(void *sync_obj, void *sync_arg)
{
	return 0;
}

static void radeon_sync_obj_unref(void **sync_obj)
{
	radeon_fence_unref((struct radeon_fence **)sync_obj);
}

static void *radeon_sync_obj_ref(void *sync_obj)
{
	return radeon_fence_ref((struct radeon_fence *)sync_obj);
}

static bool radeon_sync_obj_signaled(void *sync_obj, void *sync_arg)
{
	return radeon_fence_signaled((struct radeon_fence *)sync_obj);
}

static struct ttm_bo_driver radeon_bo_driver = {
	.create_ttm_backend_entry = &radeon_create_ttm_backend_entry,
	.invalidate_caches = &radeon_invalidate_caches,
	.init_mem_type = &radeon_init_mem_type,
	.evict_flags = &radeon_evict_flags,
	.move = &radeon_bo_move,
	.verify_access = &radeon_verify_access,
	.sync_obj_signaled = &radeon_sync_obj_signaled,
	.sync_obj_wait = &radeon_sync_obj_wait,
	.sync_obj_flush = &radeon_sync_obj_flush,
	.sync_obj_unref = &radeon_sync_obj_unref,
	.sync_obj_ref = &radeon_sync_obj_ref,
530 531
	.move_notify = &radeon_bo_move_notify,
	.fault_reserve_notify = &radeon_bo_fault_reserve_notify,
532 533
	.io_mem_reserve = &radeon_ttm_io_mem_reserve,
	.io_mem_free = &radeon_ttm_io_mem_free,
534 535 536 537 538 539 540 541 542 543 544 545
};

int radeon_ttm_init(struct radeon_device *rdev)
{
	int r;

	r = radeon_ttm_global_init(rdev);
	if (r) {
		return r;
	}
	/* No others user of address space so set it to 0 */
	r = ttm_bo_device_init(&rdev->mman.bdev,
546
			       rdev->mman.bo_global_ref.ref.object,
D
Dave Airlie 已提交
547 548
			       &radeon_bo_driver, DRM_FILE_PAGE_OFFSET,
			       rdev->need_dma32);
549 550 551 552
	if (r) {
		DRM_ERROR("failed initializing buffer object driver(%d).\n", r);
		return r;
	}
553
	rdev->mman.initialized = true;
554
	r = ttm_bo_init_mm(&rdev->mman.bdev, TTM_PL_VRAM,
555
				rdev->mc.real_vram_size >> PAGE_SHIFT);
556 557 558 559
	if (r) {
		DRM_ERROR("Failed initializing VRAM heap.\n");
		return r;
	}
560
	r = radeon_bo_create(rdev, 256 * 1024, PAGE_SIZE, true,
561 562
				RADEON_GEM_DOMAIN_VRAM,
				&rdev->stollen_vga_memory);
563 564 565
	if (r) {
		return r;
	}
566 567 568 569 570
	r = radeon_bo_reserve(rdev->stollen_vga_memory, false);
	if (r)
		return r;
	r = radeon_bo_pin(rdev->stollen_vga_memory, RADEON_GEM_DOMAIN_VRAM, NULL);
	radeon_bo_unreserve(rdev->stollen_vga_memory);
571
	if (r) {
572
		radeon_bo_unref(&rdev->stollen_vga_memory);
573 574 575
		return r;
	}
	DRM_INFO("radeon: %uM of VRAM memory ready\n",
576
		 (unsigned)rdev->mc.real_vram_size / (1024 * 1024));
577
	r = ttm_bo_init_mm(&rdev->mman.bdev, TTM_PL_TT,
578
				rdev->mc.gtt_size >> PAGE_SHIFT);
579 580 581 582 583
	if (r) {
		DRM_ERROR("Failed initializing GTT heap.\n");
		return r;
	}
	DRM_INFO("radeon: %uM of GTT memory ready.\n",
584
		 (unsigned)(rdev->mc.gtt_size / (1024 * 1024)));
585 586 587
	if (unlikely(rdev->mman.bdev.dev_mapping == NULL)) {
		rdev->mman.bdev.dev_mapping = rdev->ddev->dev_mapping;
	}
588 589 590 591 592 593

	r = radeon_ttm_debugfs_init(rdev);
	if (r) {
		DRM_ERROR("Failed to init debugfs\n");
		return r;
	}
594 595 596 597 598
	return 0;
}

void radeon_ttm_fini(struct radeon_device *rdev)
{
599 600
	int r;

601 602
	if (!rdev->mman.initialized)
		return;
603
	if (rdev->stollen_vga_memory) {
604 605 606 607 608 609
		r = radeon_bo_reserve(rdev->stollen_vga_memory, false);
		if (r == 0) {
			radeon_bo_unpin(rdev->stollen_vga_memory);
			radeon_bo_unreserve(rdev->stollen_vga_memory);
		}
		radeon_bo_unref(&rdev->stollen_vga_memory);
610 611 612 613 614 615
	}
	ttm_bo_clean_mm(&rdev->mman.bdev, TTM_PL_VRAM);
	ttm_bo_clean_mm(&rdev->mman.bdev, TTM_PL_TT);
	ttm_bo_device_release(&rdev->mman.bdev);
	radeon_gart_fini(rdev);
	radeon_ttm_global_fini(rdev);
616
	rdev->mman.initialized = false;
617 618 619
	DRM_INFO("radeon: ttm finalized\n");
}

620 621 622 623 624 625 626 627 628 629 630 631 632 633
/* this should only be called at bootup or when userspace
 * isn't running */
void radeon_ttm_set_active_vram_size(struct radeon_device *rdev, u64 size)
{
	struct ttm_mem_type_manager *man;

	if (!rdev->mman.initialized)
		return;

	man = &rdev->mman.bdev.man[TTM_PL_VRAM];
	/* this just adjusts TTM size idea, which sets lpfn to the correct value */
	man->size = size >> PAGE_SHIFT;
}

634
static struct vm_operations_struct radeon_ttm_vm_ops;
635
static const struct vm_operations_struct *ttm_vm_ops = NULL;
636 637 638 639

static int radeon_ttm_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
{
	struct ttm_buffer_object *bo;
640
	struct radeon_device *rdev;
641 642
	int r;

643
	bo = (struct ttm_buffer_object *)vma->vm_private_data;	
644 645 646
	if (bo == NULL) {
		return VM_FAULT_NOPAGE;
	}
647 648
	rdev = radeon_get_rdev(bo->bdev);
	mutex_lock(&rdev->vram_mutex);
649
	r = ttm_vm_ops->fault(vma, vmf);
650
	mutex_unlock(&rdev->vram_mutex);
651 652 653 654 655 656 657 658 659 660 661 662 663
	return r;
}

int radeon_mmap(struct file *filp, struct vm_area_struct *vma)
{
	struct drm_file *file_priv;
	struct radeon_device *rdev;
	int r;

	if (unlikely(vma->vm_pgoff < DRM_FILE_PAGE_OFFSET)) {
		return drm_mmap(filp, vma);
	}

664
	file_priv = filp->private_data;
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
	rdev = file_priv->minor->dev->dev_private;
	if (rdev == NULL) {
		return -EINVAL;
	}
	r = ttm_bo_mmap(filp, vma, &rdev->mman.bdev);
	if (unlikely(r != 0)) {
		return r;
	}
	if (unlikely(ttm_vm_ops == NULL)) {
		ttm_vm_ops = vma->vm_ops;
		radeon_ttm_vm_ops = *ttm_vm_ops;
		radeon_ttm_vm_ops.fault = &radeon_ttm_fault;
	}
	vma->vm_ops = &radeon_ttm_vm_ops;
	return 0;
}


/*
 * TTM backend functions.
 */
struct radeon_ttm_backend {
	struct ttm_backend		backend;
	struct radeon_device		*rdev;
	unsigned long			num_pages;
	struct page			**pages;
	struct page			*dummy_read_page;
692
	dma_addr_t			*dma_addrs;
693 694 695 696 697 698 699 700
	bool				populated;
	bool				bound;
	unsigned			offset;
};

static int radeon_ttm_backend_populate(struct ttm_backend *backend,
				       unsigned long num_pages,
				       struct page **pages,
701 702
				       struct page *dummy_read_page,
				       dma_addr_t *dma_addrs)
703 704 705 706 707
{
	struct radeon_ttm_backend *gtt;

	gtt = container_of(backend, struct radeon_ttm_backend, backend);
	gtt->pages = pages;
708
	gtt->dma_addrs = dma_addrs;
709 710 711 712 713 714 715 716 717 718 719 720
	gtt->num_pages = num_pages;
	gtt->dummy_read_page = dummy_read_page;
	gtt->populated = true;
	return 0;
}

static void radeon_ttm_backend_clear(struct ttm_backend *backend)
{
	struct radeon_ttm_backend *gtt;

	gtt = container_of(backend, struct radeon_ttm_backend, backend);
	gtt->pages = NULL;
721
	gtt->dma_addrs = NULL;
722 723 724 725 726 727 728 729 730 731 732 733 734 735
	gtt->num_pages = 0;
	gtt->dummy_read_page = NULL;
	gtt->populated = false;
	gtt->bound = false;
}


static int radeon_ttm_backend_bind(struct ttm_backend *backend,
				   struct ttm_mem_reg *bo_mem)
{
	struct radeon_ttm_backend *gtt;
	int r;

	gtt = container_of(backend, struct radeon_ttm_backend, backend);
736
	gtt->offset = bo_mem->start << PAGE_SHIFT;
737
	if (!gtt->num_pages) {
J
Joe Perches 已提交
738 739
		WARN(1, "nothing to bind %lu pages for mreg %p back %p!\n",
		     gtt->num_pages, bo_mem, backend);
740 741
	}
	r = radeon_gart_bind(gtt->rdev, gtt->offset,
742
			     gtt->num_pages, gtt->pages, gtt->dma_addrs);
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
	if (r) {
		DRM_ERROR("failed to bind %lu pages at 0x%08X\n",
			  gtt->num_pages, gtt->offset);
		return r;
	}
	gtt->bound = true;
	return 0;
}

static int radeon_ttm_backend_unbind(struct ttm_backend *backend)
{
	struct radeon_ttm_backend *gtt;

	gtt = container_of(backend, struct radeon_ttm_backend, backend);
	radeon_gart_unbind(gtt->rdev, gtt->offset, gtt->num_pages);
	gtt->bound = false;
	return 0;
}

static void radeon_ttm_backend_destroy(struct ttm_backend *backend)
{
	struct radeon_ttm_backend *gtt;

	gtt = container_of(backend, struct radeon_ttm_backend, backend);
	if (gtt->bound) {
		radeon_ttm_backend_unbind(backend);
	}
	kfree(gtt);
}

static struct ttm_backend_func radeon_backend_func = {
	.populate = &radeon_ttm_backend_populate,
	.clear = &radeon_ttm_backend_clear,
	.bind = &radeon_ttm_backend_bind,
	.unbind = &radeon_ttm_backend_unbind,
	.destroy = &radeon_ttm_backend_destroy,
};

struct ttm_backend *radeon_ttm_backend_create(struct radeon_device *rdev)
{
	struct radeon_ttm_backend *gtt;

	gtt = kzalloc(sizeof(struct radeon_ttm_backend), GFP_KERNEL);
	if (gtt == NULL) {
		return NULL;
	}
	gtt->backend.bdev = &rdev->mman.bdev;
	gtt->backend.func = &radeon_backend_func;
	gtt->rdev = rdev;
	gtt->pages = NULL;
	gtt->num_pages = 0;
	gtt->dummy_read_page = NULL;
	gtt->populated = false;
	gtt->bound = false;
	return &gtt->backend;
}
799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820

#define RADEON_DEBUGFS_MEM_TYPES 2

#if defined(CONFIG_DEBUG_FS)
static int radeon_mm_dump_table(struct seq_file *m, void *data)
{
	struct drm_info_node *node = (struct drm_info_node *)m->private;
	struct drm_mm *mm = (struct drm_mm *)node->info_ent->data;
	struct drm_device *dev = node->minor->dev;
	struct radeon_device *rdev = dev->dev_private;
	int ret;
	struct ttm_bo_global *glob = rdev->mman.bdev.glob;

	spin_lock(&glob->lru_lock);
	ret = drm_mm_dump_table(m, mm);
	spin_unlock(&glob->lru_lock);
	return ret;
}
#endif

static int radeon_ttm_debugfs_init(struct radeon_device *rdev)
{
821
#if defined(CONFIG_DEBUG_FS)
822 823
	static struct drm_info_list radeon_mem_types_list[RADEON_DEBUGFS_MEM_TYPES+1];
	static char radeon_mem_types_names[RADEON_DEBUGFS_MEM_TYPES+1][32];
824 825 826 827 828 829 830 831 832 833 834
	unsigned i;

	for (i = 0; i < RADEON_DEBUGFS_MEM_TYPES; i++) {
		if (i == 0)
			sprintf(radeon_mem_types_names[i], "radeon_vram_mm");
		else
			sprintf(radeon_mem_types_names[i], "radeon_gtt_mm");
		radeon_mem_types_list[i].name = radeon_mem_types_names[i];
		radeon_mem_types_list[i].show = &radeon_mm_dump_table;
		radeon_mem_types_list[i].driver_features = 0;
		if (i == 0)
835
			radeon_mem_types_list[i].data = rdev->mman.bdev.man[TTM_PL_VRAM].priv;
836
		else
837
			radeon_mem_types_list[i].data = rdev->mman.bdev.man[TTM_PL_TT].priv;
838 839

	}
840 841 842 843 844 845 846
	/* Add ttm page pool to debugfs */
	sprintf(radeon_mem_types_names[i], "ttm_page_pool");
	radeon_mem_types_list[i].name = radeon_mem_types_names[i];
	radeon_mem_types_list[i].show = &ttm_page_alloc_debugfs;
	radeon_mem_types_list[i].driver_features = 0;
	radeon_mem_types_list[i].data = NULL;
	return radeon_debugfs_add_files(rdev, radeon_mem_types_list, RADEON_DEBUGFS_MEM_TYPES+1);
847 848 849 850

#endif
	return 0;
}