i915_gem_gtt.h 20.1 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
/*
 * Copyright © 2014 Intel Corporation
 *
 * 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 OR COPYRIGHT HOLDERS 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.
 *
 * Please try to maintain the following order within this file unless it makes
 * sense to do otherwise. From top to bottom:
 * 1. typedefs
 * 2. #defines, and macros
 * 3. structure definitions
 * 4. function prototypes
 *
 * Within each section, please try to order by generation in ascending order,
 * from top to bottom (ie. gen6 on the top, gen8 on the bottom).
 */

#ifndef __I915_GEM_GTT_H__
#define __I915_GEM_GTT_H__

37
#include <linux/io-mapping.h>
J
Joonas Lahtinen 已提交
38
#include <linux/mm.h>
39
#include <linux/pagevec.h>
40

J
Joonas Lahtinen 已提交
41
#include "i915_gem_timeline.h"
42
#include "i915_gem_request.h"
43
#include "i915_selftest.h"
44

45 46 47 48 49 50 51
#define I915_GTT_PAGE_SIZE_4K BIT(12)
#define I915_GTT_PAGE_SIZE_64K BIT(16)
#define I915_GTT_PAGE_SIZE_2M BIT(21)

#define I915_GTT_PAGE_SIZE I915_GTT_PAGE_SIZE_4K
#define I915_GTT_MAX_PAGE_SIZE I915_GTT_PAGE_SIZE_2M

52 53
#define I915_GTT_MIN_ALIGNMENT I915_GTT_PAGE_SIZE

54 55 56 57 58
#define I915_FENCE_REG_NONE -1
#define I915_MAX_NUM_FENCES 32
/* 32 fences + sign bit for FENCE_REG_NONE */
#define I915_MAX_NUM_FENCE_BITS 6

59
struct drm_i915_file_private;
60
struct drm_i915_fence_reg;
61

62 63 64 65 66
typedef u32 gen6_pte_t;
typedef u64 gen8_pte_t;
typedef u64 gen8_pde_t;
typedef u64 gen8_ppgtt_pdpe_t;
typedef u64 gen8_ppgtt_pml4e_t;
67

68
#define ggtt_total_entries(ggtt) ((ggtt)->base.total >> PAGE_SHIFT)
69 70 71 72 73 74 75 76 77

/* gen6-hsw has bit 11-4 for physical addr bit 39-32 */
#define GEN6_GTT_ADDR_ENCODE(addr)	((addr) | (((addr) >> 28) & 0xff0))
#define GEN6_PTE_ADDR_ENCODE(addr)	GEN6_GTT_ADDR_ENCODE(addr)
#define GEN6_PDE_ADDR_ENCODE(addr)	GEN6_GTT_ADDR_ENCODE(addr)
#define GEN6_PTE_CACHE_LLC		(2 << 1)
#define GEN6_PTE_UNCACHED		(1 << 1)
#define GEN6_PTE_VALID			(1 << 0)

78
#define I915_PTES(pte_len)		((unsigned int)(PAGE_SIZE / (pte_len)))
79 80 81
#define I915_PTE_MASK(pte_len)		(I915_PTES(pte_len) - 1)
#define I915_PDES			512
#define I915_PDE_MASK			(I915_PDES - 1)
82
#define NUM_PTE(pde_shift)     (1 << (pde_shift - PAGE_SHIFT))
83 84 85

#define GEN6_PTES			I915_PTES(sizeof(gen6_pte_t))
#define GEN6_PD_SIZE		        (I915_PDES * PAGE_SIZE)
86
#define GEN6_PD_ALIGN			(PAGE_SIZE * 16)
87
#define GEN6_PDE_SHIFT			22
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109
#define GEN6_PDE_VALID			(1 << 0)

#define GEN7_PTE_CACHE_L3_LLC		(3 << 1)

#define BYT_PTE_SNOOPED_BY_CPU_CACHES	(1 << 2)
#define BYT_PTE_WRITEABLE		(1 << 1)

/* Cacheability Control is a 4-bit value. The low three bits are stored in bits
 * 3:1 of the PTE, while the fourth bit is stored in bit 11 of the PTE.
 */
#define HSW_CACHEABILITY_CONTROL(bits)	((((bits) & 0x7) << 1) | \
					 (((bits) & 0x8) << (11 - 3)))
#define HSW_WB_LLC_AGE3			HSW_CACHEABILITY_CONTROL(0x2)
#define HSW_WB_LLC_AGE0			HSW_CACHEABILITY_CONTROL(0x3)
#define HSW_WB_ELLC_LLC_AGE3		HSW_CACHEABILITY_CONTROL(0x8)
#define HSW_WB_ELLC_LLC_AGE0		HSW_CACHEABILITY_CONTROL(0xb)
#define HSW_WT_ELLC_LLC_AGE3		HSW_CACHEABILITY_CONTROL(0x7)
#define HSW_WT_ELLC_LLC_AGE0		HSW_CACHEABILITY_CONTROL(0x6)
#define HSW_PTE_UNCACHED		(0)
#define HSW_GTT_ADDR_ENCODE(addr)	((addr) | (((addr) >> 28) & 0x7f0))
#define HSW_PTE_ADDR_ENCODE(addr)	HSW_GTT_ADDR_ENCODE(addr)

110
/* GEN8 32b style address is defined as a 3 level page table:
111 112 113 114
 * 31:30 | 29:21 | 20:12 |  11:0
 * PDPE  |  PDE  |  PTE  | offset
 * The difference as compared to normal x86 3 level page table is the PDPEs are
 * programmed via register.
115 116 117 118 119 120 121 122 123
 */
#define GEN8_3LVL_PDPES			4
#define GEN8_PDE_SHIFT			21
#define GEN8_PDE_MASK			0x1ff
#define GEN8_PTE_SHIFT			12
#define GEN8_PTE_MASK			0x1ff
#define GEN8_PTES			I915_PTES(sizeof(gen8_pte_t))

/* GEN8 48b style address is defined as a 4 level page table:
124 125
 * 47:39 | 38:30 | 29:21 | 20:12 |  11:0
 * PML4E | PDPE  |  PDE  |  PTE  | offset
126
 */
127 128
#define GEN8_PML4ES_PER_PML4		512
#define GEN8_PML4E_SHIFT		39
129
#define GEN8_PML4E_MASK			(GEN8_PML4ES_PER_PML4 - 1)
130
#define GEN8_PDPE_SHIFT			30
131 132 133
/* NB: GEN8_PDPE_MASK is untrue for 32b platforms, but it has no impact on 32b page
 * tables */
#define GEN8_PDPE_MASK			0x1ff
134

135 136 137 138
#define PPAT_UNCACHED			(_PAGE_PWT | _PAGE_PCD)
#define PPAT_CACHED_PDE			0 /* WB LLC */
#define PPAT_CACHED			_PAGE_PAT /* WB LLCeLLC */
#define PPAT_DISPLAY_ELLC		_PAGE_PCD /* WT eLLC */
139

140
#define CHV_PPAT_SNOOP			(1<<6)
141
#define GEN8_PPAT_AGE(x)		((x)<<4)
142 143 144 145 146 147 148 149
#define GEN8_PPAT_LLCeLLC		(3<<2)
#define GEN8_PPAT_LLCELLC		(2<<2)
#define GEN8_PPAT_LLC			(1<<2)
#define GEN8_PPAT_WB			(3<<0)
#define GEN8_PPAT_WT			(2<<0)
#define GEN8_PPAT_WC			(1<<0)
#define GEN8_PPAT_UC			(0<<0)
#define GEN8_PPAT_ELLC_OVERRIDE		(0<<2)
150
#define GEN8_PPAT(i, x)			((u64)(x) << ((i) * 8))
151

152 153 154 155 156
#define GEN8_PPAT_GET_CA(x) ((x) & 3)
#define GEN8_PPAT_GET_TC(x) ((x) & (3 << 2))
#define GEN8_PPAT_GET_AGE(x) ((x) & (3 << 4))
#define CHV_PPAT_GET_SNOOP(x) ((x) & (1 << 6))

J
Joonas Lahtinen 已提交
157 158
struct sg_table;

159
struct intel_rotation_info {
160
	struct intel_rotation_plane_info {
161
		/* tiles */
162
		unsigned int width, height, stride, offset;
163
	} plane[2];
164 165 166 167 168 169
} __packed;

static inline void assert_intel_rotation_info_is_packed(void)
{
	BUILD_BUG_ON(sizeof(struct intel_rotation_info) != 8*sizeof(unsigned int));
}
170

171 172 173
struct intel_partial_info {
	u64 offset;
	unsigned int size;
174 175 176 177 178 179
} __packed;

static inline void assert_intel_partial_info_is_packed(void)
{
	BUILD_BUG_ON(sizeof(struct intel_partial_info) != sizeof(u64) + sizeof(unsigned int));
}
180

181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200
enum i915_ggtt_view_type {
	I915_GGTT_VIEW_NORMAL = 0,
	I915_GGTT_VIEW_ROTATED = sizeof(struct intel_rotation_info),
	I915_GGTT_VIEW_PARTIAL = sizeof(struct intel_partial_info),
};

static inline void assert_i915_ggtt_view_type_is_unique(void)
{
	/* As we encode the size of each branch inside the union into its type,
	 * we have to be careful that each branch has a unique size.
	 */
	switch ((enum i915_ggtt_view_type)0) {
	case I915_GGTT_VIEW_NORMAL:
	case I915_GGTT_VIEW_PARTIAL:
	case I915_GGTT_VIEW_ROTATED:
		/* gcc complains if these are identical cases */
		break;
	}
}

201 202
struct i915_ggtt_view {
	enum i915_ggtt_view_type type;
203
	union {
204
		/* Members need to contain no holes/padding */
205
		struct intel_partial_info partial;
206
		struct intel_rotation_info rotated;
207
	};
208 209
};

210
enum i915_cache_level;
211

J
Joonas Lahtinen 已提交
212
struct i915_vma;
213

214
struct i915_page_dma {
B
Ben Widawsky 已提交
215
	struct page *page;
216 217 218 219 220 221
	union {
		dma_addr_t daddr;

		/* For gen6/gen7 only. This is the offset in the GGTT
		 * where the page directory entries for PPGTT begin
		 */
222
		u32 ggtt_offset;
223 224 225
	};
};

226 227 228 229
#define px_base(px) (&(px)->base)
#define px_page(px) (px_base(px)->page)
#define px_dma(px) (px_base(px)->daddr)

230 231
struct i915_page_table {
	struct i915_page_dma base;
232
	unsigned int used_ptes;
B
Ben Widawsky 已提交
233 234
};

235
struct i915_page_directory {
236
	struct i915_page_dma base;
237

238
	struct i915_page_table *page_table[I915_PDES]; /* PDEs */
239
	unsigned int used_pdes;
B
Ben Widawsky 已提交
240 241
};

242
struct i915_page_directory_pointer {
243 244
	struct i915_page_dma base;
	struct i915_page_directory **page_directory;
245
	unsigned int used_pdpes;
B
Ben Widawsky 已提交
246 247
};

248 249 250 251 252
struct i915_pml4 {
	struct i915_page_dma base;
	struct i915_page_directory_pointer *pdps[GEN8_PML4ES_PER_PML4];
};

253 254
struct i915_address_space {
	struct drm_mm mm;
C
Chris Wilson 已提交
255
	struct i915_gem_timeline timeline;
256
	struct drm_i915_private *i915;
257
	struct device *dma;
258 259 260 261 262 263 264 265 266
	/* Every address space belongs to a struct file - except for the global
	 * GTT that is owned by the driver (and so @file is set to NULL). In
	 * principle, no information should leak from one context to another
	 * (or between files/processes etc) unless explicitly shared by the
	 * owner. Tracking the owner is important in order to free up per-file
	 * objects along with the file, to aide resource tracking, and to
	 * assign blame.
	 */
	struct drm_i915_file_private *file;
267
	struct list_head global_link;
268
	u64 total;		/* size addr space maps (ex. 2GB for ggtt) */
269
	u64 reserved;		/* size addr space reserved */
270

271 272
	bool closed;

273
	struct i915_page_dma scratch_page;
274 275
	struct i915_page_table *scratch_pt;
	struct i915_page_directory *scratch_pd;
276
	struct i915_page_directory_pointer *scratch_pdp; /* GEN8+ & 48b PPGTT */
277 278 279 280 281

	/**
	 * List of objects currently involved in rendering.
	 *
	 * Includes buffers having the contents of their GPU caches
282
	 * flushed, not necessarily primitives. last_read_req
283 284 285 286 287 288 289 290 291 292
	 * represents when the rendering involved will be completed.
	 *
	 * A reference is held on the buffer while on this list.
	 */
	struct list_head active_list;

	/**
	 * LRU list of objects which are not in the ringbuffer and
	 * are ready to unbind, but are still in the GTT.
	 *
293
	 * last_read_req is NULL while an object is in this list.
294 295 296 297 298 299 300
	 *
	 * A reference is not held on the buffer while on this list,
	 * as merely being GTT-bound shouldn't prevent its being
	 * freed, and we'll pull it off the list in the free path.
	 */
	struct list_head inactive_list;

301 302 303 304 305 306 307
	/**
	 * List of vma that have been unbound.
	 *
	 * A reference is not held on the buffer while on this list.
	 */
	struct list_head unbound_list;

308 309 310
	struct pagevec free_pages;
	bool pt_kmap_wc;

311
	/* FIXME: Need a more generic return type */
312 313
	gen6_pte_t (*pte_encode)(dma_addr_t addr,
				 enum i915_cache_level level,
314
				 u32 flags); /* Create a valid PTE */
315 316
	/* flags for pte_encode */
#define PTE_READ_ONLY	(1<<0)
317
	int (*allocate_va_range)(struct i915_address_space *vm,
318
				 u64 start, u64 length);
319
	void (*clear_range)(struct i915_address_space *vm,
320
			    u64 start, u64 length);
321 322
	void (*insert_page)(struct i915_address_space *vm,
			    dma_addr_t addr,
323
			    u64 offset,
324 325
			    enum i915_cache_level cache_level,
			    u32 flags);
326
	void (*insert_entries)(struct i915_address_space *vm,
327
			       struct i915_vma *vma,
328 329
			       enum i915_cache_level cache_level,
			       u32 flags);
330
	void (*cleanup)(struct i915_address_space *vm);
331 332 333 334
	/** Unmap an object from an address space. This usually consists of
	 * setting the valid PTE entries to a reserved scratch page. */
	void (*unbind_vma)(struct i915_vma *vma);
	/* Map an object into an address space with the given cache flags. */
335 336 337
	int (*bind_vma)(struct i915_vma *vma,
			enum i915_cache_level cache_level,
			u32 flags);
338 339
	int (*set_pages)(struct i915_vma *vma);
	void (*clear_pages)(struct i915_vma *vma);
340 341

	I915_SELFTEST_DECLARE(struct fault_attr fault_attr);
342 343
};

344
#define i915_is_ggtt(V) (!(V)->file)
345

346 347 348 349 350 351
static inline bool
i915_vm_is_48bit(const struct i915_address_space *vm)
{
	return (vm->total - 1) >> 32;
}

352 353 354 355 356 357 358
/* The Graphics Translation Table is the way in which GEN hardware translates a
 * Graphics Virtual Address into a Physical Address. In addition to the normal
 * collateral associated with any va->pa translations GEN hardware also has a
 * portion of the GTT which can be mapped by the CPU and remain both coherent
 * and correct (in cases like swizzling). That region is referred to as GMADR in
 * the spec.
 */
359
struct i915_ggtt {
360
	struct i915_address_space base;
361
	struct io_mapping mappable;	/* Mapping to our CPU mappable region */
362

363 364 365
	phys_addr_t mappable_base;	/* PA of our GMADR */
	u64 mappable_end;		/* End offset that we can CPU map */

366 367 368 369 370 371 372 373
	/* Stolen memory is segmented in hardware with different portions
	 * offlimits to certain functions.
	 *
	 * The drm_mm is initialised to the total accessible range, as found
	 * from the PCI config. On Broadwell+, this is further restricted to
	 * avoid the first page! The upper end of stolen memory is reserved for
	 * hardware functions and similarly removed from the accessible range.
	 */
374 375 376 377
	u32 stolen_size;		/* Total size of stolen memory */
	u32 stolen_usable_size;	/* Total size minus reserved ranges */
	u32 stolen_reserved_base;
	u32 stolen_reserved_size;
378 379 380

	/** "Graphics Stolen Memory" holds the global PTEs */
	void __iomem *gsm;
381
	void (*invalidate)(struct drm_i915_private *dev_priv);
382 383 384 385

	bool do_idle_maps;

	int mtrr;
386 387

	struct drm_mm_node error_capture;
388 389 390 391 392 393
};

struct i915_hw_ppgtt {
	struct i915_address_space base;
	struct kref ref;
	struct drm_mm_node node;
394
	unsigned long pd_dirty_rings;
B
Ben Widawsky 已提交
395
	union {
396 397 398
		struct i915_pml4 pml4;		/* GEN8+ & 48b PPGTT */
		struct i915_page_directory_pointer pdp;	/* GEN8+ */
		struct i915_page_directory pd;		/* GEN6-7 */
B
Ben Widawsky 已提交
399
	};
400

401 402
	gen6_pte_t __iomem *pd_addr;

403
	int (*switch_mm)(struct i915_hw_ppgtt *ppgtt,
404
			 struct drm_i915_gem_request *req);
405 406 407
	void (*debug_dump)(struct i915_hw_ppgtt *ppgtt, struct seq_file *m);
};

408 409 410 411 412 413 414
/*
 * gen6_for_each_pde() iterates over every pde from start until start+length.
 * If start and start+length are not perfectly divisible, the macro will round
 * down and up as needed. Start=0 and length=2G effectively iterates over
 * every PDE in the system. The macro modifies ALL its parameters except 'pd',
 * so each of the other parameters should preferably be a simple variable, or
 * at most an lvalue with no side-effects!
415
 */
416 417 418 419 420 421 422 423 424 425 426 427 428
#define gen6_for_each_pde(pt, pd, start, length, iter)			\
	for (iter = gen6_pde_index(start);				\
	     length > 0 && iter < I915_PDES &&				\
		(pt = (pd)->page_table[iter], true);			\
	     ({ u32 temp = ALIGN(start+1, 1 << GEN6_PDE_SHIFT);		\
		    temp = min(temp - start, length);			\
		    start += temp, length -= temp; }), ++iter)

#define gen6_for_all_pdes(pt, pd, iter)					\
	for (iter = 0;							\
	     iter < I915_PDES &&					\
		(pt = (pd)->page_table[iter], true);			\
	     ++iter)
429

430
static inline u32 i915_pte_index(u64 address, unsigned int pde_shift)
431
{
432
	const u32 mask = NUM_PTE(pde_shift) - 1;
433 434 435 436 437 438 439 440

	return (address >> PAGE_SHIFT) & mask;
}

/* Helper to counts the number of PTEs within the given length. This count
 * does not cross a page table boundary, so the max value would be
 * GEN6_PTES for GEN6, and GEN8_PTES for GEN8.
*/
441
static inline u32 i915_pte_count(u64 addr, u64 length, unsigned int pde_shift)
442
{
443 444
	const u64 mask = ~((1ULL << pde_shift) - 1);
	u64 end;
445 446 447 448 449 450 451 452 453 454 455 456

	WARN_ON(length == 0);
	WARN_ON(offset_in_page(addr|length));

	end = addr + length;

	if ((addr & mask) != (end & mask))
		return NUM_PTE(pde_shift) - i915_pte_index(addr, pde_shift);

	return i915_pte_index(end, pde_shift) - i915_pte_index(addr, pde_shift);
}

457
static inline u32 i915_pde_index(u64 addr, u32 shift)
458 459 460 461
{
	return (addr >> shift) & I915_PDE_MASK;
}

462
static inline u32 gen6_pte_index(u32 addr)
463 464 465 466
{
	return i915_pte_index(addr, GEN6_PDE_SHIFT);
}

467
static inline u32 gen6_pte_count(u32 addr, u32 length)
468 469 470 471
{
	return i915_pte_count(addr, length, GEN6_PDE_SHIFT);
}

472
static inline u32 gen6_pde_index(u32 addr)
473 474 475 476
{
	return i915_pde_index(addr, GEN6_PDE_SHIFT);
}

477 478 479 480 481 482
static inline unsigned int
i915_pdpes_per_pdp(const struct i915_address_space *vm)
{
	if (i915_vm_is_48bit(vm))
		return GEN8_PML4ES_PER_PML4;

483
	return GEN8_3LVL_PDPES;
484 485
}

486 487 488 489
/* Equivalent to the gen6 version, For each pde iterates over every pde
 * between from start until start + length. On gen8+ it simply iterates
 * over every page directory entry in a page directory.
 */
490 491 492 493 494 495 496 497 498 499
#define gen8_for_each_pde(pt, pd, start, length, iter)			\
	for (iter = gen8_pde_index(start);				\
	     length > 0 && iter < I915_PDES &&				\
		(pt = (pd)->page_table[iter], true);			\
	     ({ u64 temp = ALIGN(start+1, 1 << GEN8_PDE_SHIFT);		\
		    temp = min(temp - start, length);			\
		    start += temp, length -= temp; }), ++iter)

#define gen8_for_each_pdpe(pd, pdp, start, length, iter)		\
	for (iter = gen8_pdpe_index(start);				\
500
	     length > 0 && iter < i915_pdpes_per_pdp(vm) &&		\
501 502 503 504 505 506 507 508 509 510 511 512
		(pd = (pdp)->page_directory[iter], true);		\
	     ({ u64 temp = ALIGN(start+1, 1 << GEN8_PDPE_SHIFT);	\
		    temp = min(temp - start, length);			\
		    start += temp, length -= temp; }), ++iter)

#define gen8_for_each_pml4e(pdp, pml4, start, length, iter)		\
	for (iter = gen8_pml4e_index(start);				\
	     length > 0 && iter < GEN8_PML4ES_PER_PML4 &&		\
		(pdp = (pml4)->pdps[iter], true);			\
	     ({ u64 temp = ALIGN(start+1, 1ULL << GEN8_PML4E_SHIFT);	\
		    temp = min(temp - start, length);			\
		    start += temp, length -= temp; }), ++iter)
513

514
static inline u32 gen8_pte_index(u64 address)
515 516 517 518
{
	return i915_pte_index(address, GEN8_PDE_SHIFT);
}

519
static inline u32 gen8_pde_index(u64 address)
520 521 522 523
{
	return i915_pde_index(address, GEN8_PDE_SHIFT);
}

524
static inline u32 gen8_pdpe_index(u64 address)
525 526 527 528
{
	return (address >> GEN8_PDPE_SHIFT) & GEN8_PDPE_MASK;
}

529
static inline u32 gen8_pml4e_index(u64 address)
530
{
531
	return (address >> GEN8_PML4E_SHIFT) & GEN8_PML4E_MASK;
532 533
}

534
static inline u64 gen8_pte_count(u64 address, u64 length)
535 536 537 538
{
	return i915_pte_count(address, length, GEN8_PDE_SHIFT);
}

539 540 541
static inline dma_addr_t
i915_page_dir_dma_addr(const struct i915_hw_ppgtt *ppgtt, const unsigned n)
{
542
	return px_dma(ppgtt->pdp.page_directory[n]);
543 544
}

J
Joonas Lahtinen 已提交
545 546 547 548 549 550 551
static inline struct i915_ggtt *
i915_vm_to_ggtt(struct i915_address_space *vm)
{
	GEM_BUG_ON(!i915_is_ggtt(vm));
	return container_of(vm, struct i915_ggtt, base);
}

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
#define INTEL_MAX_PPAT_ENTRIES 8
#define INTEL_PPAT_PERFECT_MATCH (~0U)

struct intel_ppat;

struct intel_ppat_entry {
	struct intel_ppat *ppat;
	struct kref ref;
	u8 value;
};

struct intel_ppat {
	struct intel_ppat_entry entries[INTEL_MAX_PPAT_ENTRIES];
	DECLARE_BITMAP(used, INTEL_MAX_PPAT_ENTRIES);
	DECLARE_BITMAP(dirty, INTEL_MAX_PPAT_ENTRIES);
	unsigned int max_entries;
	u8 clear_value;
	/*
	 * Return a score to show how two PPAT values match,
	 * a INTEL_PPAT_PERFECT_MATCH indicates a perfect match
	 */
	unsigned int (*match)(u8 src, u8 dst);
	void (*update_hw)(struct drm_i915_private *i915);

	struct drm_i915_private *i915;
};

const struct intel_ppat_entry *
intel_ppat_get(struct drm_i915_private *i915, u8 value);
void intel_ppat_put(const struct intel_ppat_entry *entry);

583 584 585
int i915_gem_init_aliasing_ppgtt(struct drm_i915_private *i915);
void i915_gem_fini_aliasing_ppgtt(struct drm_i915_private *i915);

586 587 588
int i915_ggtt_probe_hw(struct drm_i915_private *dev_priv);
int i915_ggtt_init_hw(struct drm_i915_private *dev_priv);
int i915_ggtt_enable_hw(struct drm_i915_private *dev_priv);
589 590
void i915_ggtt_enable_guc(struct drm_i915_private *i915);
void i915_ggtt_disable_guc(struct drm_i915_private *i915);
591
int i915_gem_init_ggtt(struct drm_i915_private *dev_priv);
592
void i915_ggtt_cleanup_hw(struct drm_i915_private *dev_priv);
593

594
int i915_ppgtt_init_hw(struct drm_i915_private *dev_priv);
595
void i915_ppgtt_release(struct kref *kref);
596
struct i915_hw_ppgtt *i915_ppgtt_create(struct drm_i915_private *dev_priv,
C
Chris Wilson 已提交
597 598
					struct drm_i915_file_private *fpriv,
					const char *name);
599
void i915_ppgtt_close(struct i915_address_space *vm);
600 601 602 603 604 605 606 607 608 609
static inline void i915_ppgtt_get(struct i915_hw_ppgtt *ppgtt)
{
	if (ppgtt)
		kref_get(&ppgtt->ref);
}
static inline void i915_ppgtt_put(struct i915_hw_ppgtt *ppgtt)
{
	if (ppgtt)
		kref_put(&ppgtt->ref, i915_ppgtt_release);
}
610

611
void i915_check_and_clear_faults(struct drm_i915_private *dev_priv);
612 613
void i915_gem_suspend_gtt_mappings(struct drm_i915_private *dev_priv);
void i915_gem_restore_gtt_mappings(struct drm_i915_private *dev_priv);
614

615 616 617 618
int __must_check i915_gem_gtt_prepare_pages(struct drm_i915_gem_object *obj,
					    struct sg_table *pages);
void i915_gem_gtt_finish_pages(struct drm_i915_gem_object *obj,
			       struct sg_table *pages);
619

620 621 622 623 624
int i915_gem_gtt_reserve(struct i915_address_space *vm,
			 struct drm_mm_node *node,
			 u64 size, u64 offset, unsigned long color,
			 unsigned int flags);

625 626 627 628 629
int i915_gem_gtt_insert(struct i915_address_space *vm,
			struct drm_mm_node *node,
			u64 size, u64 alignment, unsigned long color,
			u64 start, u64 end, unsigned int flags);

630
/* Flags used by pin/bind&friends. */
631 632 633
#define PIN_NONBLOCK		BIT(0)
#define PIN_MAPPABLE		BIT(1)
#define PIN_ZONE_4G		BIT(2)
634
#define PIN_NONFAULT		BIT(3)
635
#define PIN_NOEVICT		BIT(4)
636 637 638 639 640 641 642 643 644

#define PIN_MBZ			BIT(5) /* I915_VMA_PIN_OVERFLOW */
#define PIN_GLOBAL		BIT(6) /* I915_VMA_GLOBAL_BIND */
#define PIN_USER		BIT(7) /* I915_VMA_LOCAL_BIND */
#define PIN_UPDATE		BIT(8)

#define PIN_HIGH		BIT(9)
#define PIN_OFFSET_BIAS		BIT(10)
#define PIN_OFFSET_FIXED	BIT(11)
645
#define PIN_OFFSET_MASK		(-I915_GTT_PAGE_SIZE)
646

647
#endif