intel_uncore.c 41.6 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
/*
 * Copyright © 2013 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.
 */

#include "i915_drv.h"
#include "intel_drv.h"
26
#include "i915_vgpu.h"
27

28 29
#include <linux/pm_runtime.h>

30
#define FORCEWAKE_ACK_TIMEOUT_MS 50
31

32
#define __raw_posting_read(dev_priv__, reg__) (void)__raw_i915_read32((dev_priv__), (reg__))
33

34 35 36 37 38 39 40
static const char * const forcewake_domain_names[] = {
	"render",
	"blitter",
	"media",
};

const char *
41
intel_uncore_forcewake_domain_to_str(const enum forcewake_domain_id id)
42
{
43
	BUILD_BUG_ON(ARRAY_SIZE(forcewake_domain_names) != FW_DOMAIN_ID_COUNT);
44 45 46 47 48 49 50 51 52

	if (id >= 0 && id < FW_DOMAIN_ID_COUNT)
		return forcewake_domain_names[id];

	WARN_ON(id);

	return "unknown";
}

53 54 55
static void
assert_device_not_suspended(struct drm_i915_private *dev_priv)
{
56 57
	WARN_ONCE(HAS_RUNTIME_PM(dev_priv->dev) && dev_priv->pm.suspended,
		  "Device suspended\n");
58
}
59

60 61
static inline void
fw_domain_reset(const struct intel_uncore_forcewake_domain *d)
62
{
63
	WARN_ON(d->reg_set == 0);
64
	__raw_i915_write32(d->i915, d->reg_set, d->val_reset);
65 66
}

67 68
static inline void
fw_domain_arm_timer(struct intel_uncore_forcewake_domain *d)
69
{
70
	mod_timer_pinned(&d->timer, jiffies + 1);
71 72
}

73 74
static inline void
fw_domain_wait_ack_clear(const struct intel_uncore_forcewake_domain *d)
75
{
76 77
	if (wait_for_atomic((__raw_i915_read32(d->i915, d->reg_ack) &
			     FORCEWAKE_KERNEL) == 0,
78
			    FORCEWAKE_ACK_TIMEOUT_MS))
79 80 81
		DRM_ERROR("%s: timed out waiting for forcewake ack to clear.\n",
			  intel_uncore_forcewake_domain_to_str(d->id));
}
82

83 84 85 86 87
static inline void
fw_domain_get(const struct intel_uncore_forcewake_domain *d)
{
	__raw_i915_write32(d->i915, d->reg_set, d->val_set);
}
88

89 90 91 92 93
static inline void
fw_domain_wait_ack(const struct intel_uncore_forcewake_domain *d)
{
	if (wait_for_atomic((__raw_i915_read32(d->i915, d->reg_ack) &
			     FORCEWAKE_KERNEL),
94
			    FORCEWAKE_ACK_TIMEOUT_MS))
95 96 97
		DRM_ERROR("%s: timed out waiting for forcewake ack request.\n",
			  intel_uncore_forcewake_domain_to_str(d->id));
}
98

99 100 101 102
static inline void
fw_domain_put(const struct intel_uncore_forcewake_domain *d)
{
	__raw_i915_write32(d->i915, d->reg_set, d->val_clear);
103 104
}

105 106
static inline void
fw_domain_posting_read(const struct intel_uncore_forcewake_domain *d)
107
{
108 109 110
	/* something from same cacheline, but not from the set register */
	if (d->reg_post)
		__raw_posting_read(d->i915, d->reg_post);
111 112
}

113
static void
114
fw_domains_get(struct drm_i915_private *dev_priv, enum forcewake_domains fw_domains)
115
{
116
	struct intel_uncore_forcewake_domain *d;
117
	enum forcewake_domain_id id;
118

119 120 121 122 123 124
	for_each_fw_domain_mask(d, fw_domains, dev_priv, id) {
		fw_domain_wait_ack_clear(d);
		fw_domain_get(d);
		fw_domain_wait_ack(d);
	}
}
125

126
static void
127
fw_domains_put(struct drm_i915_private *dev_priv, enum forcewake_domains fw_domains)
128 129
{
	struct intel_uncore_forcewake_domain *d;
130
	enum forcewake_domain_id id;
131

132 133 134 135 136
	for_each_fw_domain_mask(d, fw_domains, dev_priv, id) {
		fw_domain_put(d);
		fw_domain_posting_read(d);
	}
}
137

138 139 140 141
static void
fw_domains_posting_read(struct drm_i915_private *dev_priv)
{
	struct intel_uncore_forcewake_domain *d;
142
	enum forcewake_domain_id id;
143 144 145 146 147 148 149 150 151

	/* No need to do for all, just do for first found */
	for_each_fw_domain(d, dev_priv, id) {
		fw_domain_posting_read(d);
		break;
	}
}

static void
152
fw_domains_reset(struct drm_i915_private *dev_priv, enum forcewake_domains fw_domains)
153 154
{
	struct intel_uncore_forcewake_domain *d;
155
	enum forcewake_domain_id id;
156

157 158
	if (dev_priv->uncore.fw_domains == 0)
		return;
159

160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176
	for_each_fw_domain_mask(d, fw_domains, dev_priv, id)
		fw_domain_reset(d);

	fw_domains_posting_read(dev_priv);
}

static void __gen6_gt_wait_for_thread_c0(struct drm_i915_private *dev_priv)
{
	/* w/a for a sporadic read returning 0 by waiting for the GT
	 * thread to wake up.
	 */
	if (wait_for_atomic_us((__raw_i915_read32(dev_priv, GEN6_GT_THREAD_STATUS_REG) &
				GEN6_GT_THREAD_STATUS_CORE_MASK) == 0, 500))
		DRM_ERROR("GT thread status wait timed out\n");
}

static void fw_domains_get_with_thread_status(struct drm_i915_private *dev_priv,
177
					      enum forcewake_domains fw_domains)
178 179
{
	fw_domains_get(dev_priv, fw_domains);
180

181
	/* WaRsForcewakeWaitTC0:snb,ivb,hsw,bdw,vlv */
182
	__gen6_gt_wait_for_thread_c0(dev_priv);
183 184 185 186 187
}

static void gen6_gt_check_fifodbg(struct drm_i915_private *dev_priv)
{
	u32 gtfifodbg;
188 189

	gtfifodbg = __raw_i915_read32(dev_priv, GTFIFODBG);
190 191
	if (WARN(gtfifodbg, "GT wake FIFO error 0x%x\n", gtfifodbg))
		__raw_i915_write32(dev_priv, GTFIFODBG, gtfifodbg);
192 193
}

194
static void fw_domains_put_with_fifo(struct drm_i915_private *dev_priv,
195
				     enum forcewake_domains fw_domains)
196
{
197
	fw_domains_put(dev_priv, fw_domains);
198 199 200
	gen6_gt_check_fifodbg(dev_priv);
}

201 202 203 204 205 206 207
static inline u32 fifo_free_entries(struct drm_i915_private *dev_priv)
{
	u32 count = __raw_i915_read32(dev_priv, GTFIFOCTL);

	return count & GT_FIFO_FREE_ENTRIES_MASK;
}

208 209 210 211
static int __gen6_gt_wait_for_fifo(struct drm_i915_private *dev_priv)
{
	int ret = 0;

212 213 214
	/* On VLV, FIFO will be shared by both SW and HW.
	 * So, we need to read the FREE_ENTRIES everytime */
	if (IS_VALLEYVIEW(dev_priv->dev))
215
		dev_priv->uncore.fifo_count = fifo_free_entries(dev_priv);
216

217 218
	if (dev_priv->uncore.fifo_count < GT_FIFO_NUM_RESERVED_ENTRIES) {
		int loop = 500;
219 220
		u32 fifo = fifo_free_entries(dev_priv);

221 222
		while (fifo <= GT_FIFO_NUM_RESERVED_ENTRIES && loop--) {
			udelay(10);
223
			fifo = fifo_free_entries(dev_priv);
224 225 226 227 228 229 230 231 232 233
		}
		if (WARN_ON(loop < 0 && fifo <= GT_FIFO_NUM_RESERVED_ENTRIES))
			++ret;
		dev_priv->uncore.fifo_count = fifo;
	}
	dev_priv->uncore.fifo_count--;

	return ret;
}

234
static void intel_uncore_fw_release_timer(unsigned long arg)
Z
Zhe Wang 已提交
235
{
236 237
	struct intel_uncore_forcewake_domain *domain = (void *)arg;
	unsigned long irqflags;
Z
Zhe Wang 已提交
238

239
	assert_device_not_suspended(domain->i915);
Z
Zhe Wang 已提交
240

241 242 243 244 245 246 247 248 249
	spin_lock_irqsave(&domain->i915->uncore.lock, irqflags);
	if (WARN_ON(domain->wake_count == 0))
		domain->wake_count++;

	if (--domain->wake_count == 0)
		domain->i915->uncore.funcs.force_wake_put(domain->i915,
							  1 << domain->id);

	spin_unlock_irqrestore(&domain->i915->uncore.lock, irqflags);
Z
Zhe Wang 已提交
250 251
}

252
void intel_uncore_forcewake_reset(struct drm_device *dev, bool restore)
Z
Zhe Wang 已提交
253
{
254
	struct drm_i915_private *dev_priv = dev->dev_private;
255
	unsigned long irqflags;
256
	struct intel_uncore_forcewake_domain *domain;
257 258 259
	int retry_count = 100;
	enum forcewake_domain_id id;
	enum forcewake_domains fw = 0, active_domains;
Z
Zhe Wang 已提交
260

261 262 263 264 265 266
	/* Hold uncore.lock across reset to prevent any register access
	 * with forcewake not set correctly. Wait until all pending
	 * timers are run before holding.
	 */
	while (1) {
		active_domains = 0;
Z
Zhe Wang 已提交
267

268 269 270
		for_each_fw_domain(domain, dev_priv, id) {
			if (del_timer_sync(&domain->timer) == 0)
				continue;
Z
Zhe Wang 已提交
271

272
			intel_uncore_fw_release_timer((unsigned long)domain);
273
		}
274

275
		spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
276

277 278 279 280
		for_each_fw_domain(domain, dev_priv, id) {
			if (timer_pending(&domain->timer))
				active_domains |= (1 << id);
		}
281

282 283
		if (active_domains == 0)
			break;
284

285 286 287 288
		if (--retry_count == 0) {
			DRM_ERROR("Timed out waiting for forcewake timers to finish\n");
			break;
		}
289

290 291 292
		spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
		cond_resched();
	}
293

294 295 296 297 298 299 300 301
	WARN_ON(active_domains);

	for_each_fw_domain(domain, dev_priv, id)
		if (domain->wake_count)
			fw |= 1 << id;

	if (fw)
		dev_priv->uncore.funcs.force_wake_put(dev_priv, fw);
302

303
	fw_domains_reset(dev_priv, FORCEWAKE_ALL);
Z
Zhe Wang 已提交
304

305 306 307 308 309 310
	if (restore) { /* If reset with a user forcewake, try to restore */
		if (fw)
			dev_priv->uncore.funcs.force_wake_get(dev_priv, fw);

		if (IS_GEN6(dev) || IS_GEN7(dev))
			dev_priv->uncore.fifo_count =
311
				fifo_free_entries(dev_priv);
312 313
	}

314
	if (!restore)
315
		assert_forcewakes_inactive(dev_priv);
316

317
	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
318 319
}

320
static void intel_uncore_ellc_detect(struct drm_device *dev)
321 322 323
{
	struct drm_i915_private *dev_priv = dev->dev_private;

324 325
	if ((IS_HASWELL(dev) || IS_BROADWELL(dev) ||
	     INTEL_INFO(dev)->gen >= 9) &&
326
	    (__raw_i915_read32(dev_priv, HSW_EDRAM_PRESENT) & EDRAM_ENABLED)) {
327 328 329 330 331 332 333 334
		/* The docs do not explain exactly how the calculation can be
		 * made. It is somewhat guessable, but for now, it's always
		 * 128MB.
		 * NB: We can't write IDICR yet because we do not have gt funcs
		 * set up */
		dev_priv->ellc_size = 128;
		DRM_INFO("Found %zuMB of eLLC\n", dev_priv->ellc_size);
	}
335 336 337 338 339 340 341 342 343
}

static void __intel_uncore_early_sanitize(struct drm_device *dev,
					  bool restore_forcewake)
{
	struct drm_i915_private *dev_priv = dev->dev_private;

	if (HAS_FPGA_DBG_UNCLAIMED(dev))
		__raw_i915_write32(dev_priv, FPGA_DBG, FPGA_DBG_RM_NOCLAIM);
344

345 346 347 348 349
	/* clear out old GT FIFO errors */
	if (IS_GEN6(dev) || IS_GEN7(dev))
		__raw_i915_write32(dev_priv, GTFIFODBG,
				   __raw_i915_read32(dev_priv, GTFIFODBG));

350 351 352 353 354 355 356 357
	/* WaDisableShadowRegForCpd:chv */
	if (IS_CHERRYVIEW(dev)) {
		__raw_i915_write32(dev_priv, GTFIFOCTL,
				   __raw_i915_read32(dev_priv, GTFIFOCTL) |
				   GT_FIFO_CTL_BLOCK_ALL_POLICY_STALL |
				   GT_FIFO_CTL_RC6_POLICY_STALL);
	}

358
	intel_uncore_forcewake_reset(dev, restore_forcewake);
359 360
}

361 362 363 364 365 366
void intel_uncore_early_sanitize(struct drm_device *dev, bool restore_forcewake)
{
	__intel_uncore_early_sanitize(dev, restore_forcewake);
	i915_check_and_clear_faults(dev);
}

367 368
void intel_uncore_sanitize(struct drm_device *dev)
{
369 370 371 372
	/* BIOS often leaves RC6 enabled, but disable it for hw init */
	intel_disable_gt_powersave(dev);
}

373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392
static void __intel_uncore_forcewake_get(struct drm_i915_private *dev_priv,
					 enum forcewake_domains fw_domains)
{
	struct intel_uncore_forcewake_domain *domain;
	enum forcewake_domain_id id;

	if (!dev_priv->uncore.funcs.force_wake_get)
		return;

	fw_domains &= dev_priv->uncore.fw_domains;

	for_each_fw_domain_mask(domain, fw_domains, dev_priv, id) {
		if (domain->wake_count++)
			fw_domains &= ~(1 << id);
	}

	if (fw_domains)
		dev_priv->uncore.funcs.force_wake_get(dev_priv, fw_domains);
}

393 394 395 396 397 398 399 400 401 402 403 404
/**
 * intel_uncore_forcewake_get - grab forcewake domain references
 * @dev_priv: i915 device instance
 * @fw_domains: forcewake domains to get reference on
 *
 * This function can be used get GT's forcewake domain references.
 * Normal register access will handle the forcewake domains automatically.
 * However if some sequence requires the GT to not power down a particular
 * forcewake domains this function should be called at the beginning of the
 * sequence. And subsequently the reference should be dropped by symmetric
 * call to intel_unforce_forcewake_put(). Usually caller wants all the domains
 * to be kept awake so the @fw_domains would be then FORCEWAKE_ALL.
405
 */
406
void intel_uncore_forcewake_get(struct drm_i915_private *dev_priv,
407
				enum forcewake_domains fw_domains)
408 409 410
{
	unsigned long irqflags;

411 412 413
	if (!dev_priv->uncore.funcs.force_wake_get)
		return;

414
	WARN_ON(dev_priv->pm.suspended);
415

416
	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
417
	__intel_uncore_forcewake_get(dev_priv, fw_domains);
418 419 420
	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
}

421
/**
422
 * intel_uncore_forcewake_get__locked - grab forcewake domain references
423
 * @dev_priv: i915 device instance
424
 * @fw_domains: forcewake domains to get reference on
425
 *
426 427
 * See intel_uncore_forcewake_get(). This variant places the onus
 * on the caller to explicitly handle the dev_priv->uncore.lock spinlock.
428
 */
429 430 431 432 433 434 435 436 437 438 439 440 441
void intel_uncore_forcewake_get__locked(struct drm_i915_private *dev_priv,
					enum forcewake_domains fw_domains)
{
	assert_spin_locked(&dev_priv->uncore.lock);

	if (!dev_priv->uncore.funcs.force_wake_get)
		return;

	__intel_uncore_forcewake_get(dev_priv, fw_domains);
}

static void __intel_uncore_forcewake_put(struct drm_i915_private *dev_priv,
					 enum forcewake_domains fw_domains)
442
{
443
	struct intel_uncore_forcewake_domain *domain;
444
	enum forcewake_domain_id id;
445

446 447 448
	if (!dev_priv->uncore.funcs.force_wake_put)
		return;

449 450 451 452 453 454 455 456 457 458
	fw_domains &= dev_priv->uncore.fw_domains;

	for_each_fw_domain_mask(domain, fw_domains, dev_priv, id) {
		if (WARN_ON(domain->wake_count == 0))
			continue;

		if (--domain->wake_count)
			continue;

		domain->wake_count++;
459
		fw_domain_arm_timer(domain);
460
	}
461
}
462

463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480
/**
 * intel_uncore_forcewake_put - release a forcewake domain reference
 * @dev_priv: i915 device instance
 * @fw_domains: forcewake domains to put references
 *
 * This function drops the device-level forcewakes for specified
 * domains obtained by intel_uncore_forcewake_get().
 */
void intel_uncore_forcewake_put(struct drm_i915_private *dev_priv,
				enum forcewake_domains fw_domains)
{
	unsigned long irqflags;

	if (!dev_priv->uncore.funcs.force_wake_put)
		return;

	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
	__intel_uncore_forcewake_put(dev_priv, fw_domains);
481 482 483
	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
}

484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502
/**
 * intel_uncore_forcewake_put__locked - grab forcewake domain references
 * @dev_priv: i915 device instance
 * @fw_domains: forcewake domains to get reference on
 *
 * See intel_uncore_forcewake_put(). This variant places the onus
 * on the caller to explicitly handle the dev_priv->uncore.lock spinlock.
 */
void intel_uncore_forcewake_put__locked(struct drm_i915_private *dev_priv,
					enum forcewake_domains fw_domains)
{
	assert_spin_locked(&dev_priv->uncore.lock);

	if (!dev_priv->uncore.funcs.force_wake_put)
		return;

	__intel_uncore_forcewake_put(dev_priv, fw_domains);
}

503
void assert_forcewakes_inactive(struct drm_i915_private *dev_priv)
504
{
505
	struct intel_uncore_forcewake_domain *domain;
506
	enum forcewake_domain_id id;
507

508 509 510
	if (!dev_priv->uncore.funcs.force_wake_get)
		return;

511
	for_each_fw_domain(domain, dev_priv, id)
512
		WARN_ON(domain->wake_count);
513 514
}

515
/* We give fast paths for the really cool registers */
516
#define NEEDS_FORCE_WAKE(reg) ((reg) < 0x40000)
517

518
#define REG_RANGE(reg, start, end) ((reg) >= (start) && (reg) < (end))
519

520 521 522 523 524 525 526 527 528 529 530 531 532
#define FORCEWAKE_VLV_RENDER_RANGE_OFFSET(reg) \
	(REG_RANGE((reg), 0x2000, 0x4000) || \
	 REG_RANGE((reg), 0x5000, 0x8000) || \
	 REG_RANGE((reg), 0xB000, 0x12000) || \
	 REG_RANGE((reg), 0x2E000, 0x30000))

#define FORCEWAKE_VLV_MEDIA_RANGE_OFFSET(reg) \
	(REG_RANGE((reg), 0x12000, 0x14000) || \
	 REG_RANGE((reg), 0x22000, 0x24000) || \
	 REG_RANGE((reg), 0x30000, 0x40000))

#define FORCEWAKE_CHV_RENDER_RANGE_OFFSET(reg) \
	(REG_RANGE((reg), 0x2000, 0x4000) || \
533
	 REG_RANGE((reg), 0x5200, 0x8000) || \
534
	 REG_RANGE((reg), 0x8300, 0x8500) || \
535
	 REG_RANGE((reg), 0xB000, 0xB480) || \
536 537 538 539 540 541 542 543
	 REG_RANGE((reg), 0xE000, 0xE800))

#define FORCEWAKE_CHV_MEDIA_RANGE_OFFSET(reg) \
	(REG_RANGE((reg), 0x8800, 0x8900) || \
	 REG_RANGE((reg), 0xD000, 0xD800) || \
	 REG_RANGE((reg), 0x12000, 0x14000) || \
	 REG_RANGE((reg), 0x1A000, 0x1C000) || \
	 REG_RANGE((reg), 0x1E800, 0x1EA00) || \
544
	 REG_RANGE((reg), 0x30000, 0x38000))
545 546 547 548 549 550

#define FORCEWAKE_CHV_COMMON_RANGE_OFFSET(reg) \
	(REG_RANGE((reg), 0x4000, 0x5000) || \
	 REG_RANGE((reg), 0x8000, 0x8300) || \
	 REG_RANGE((reg), 0x8500, 0x8600) || \
	 REG_RANGE((reg), 0x9000, 0xB000) || \
551
	 REG_RANGE((reg), 0xF000, 0x10000))
552

553
#define FORCEWAKE_GEN9_UNCORE_RANGE_OFFSET(reg) \
554
	REG_RANGE((reg), 0xB00,  0x2000)
555 556

#define FORCEWAKE_GEN9_RENDER_RANGE_OFFSET(reg) \
557 558
	(REG_RANGE((reg), 0x2000, 0x2700) || \
	 REG_RANGE((reg), 0x3000, 0x4000) || \
559
	 REG_RANGE((reg), 0x5200, 0x8000) || \
560
	 REG_RANGE((reg), 0x8140, 0x8160) || \
561 562 563
	 REG_RANGE((reg), 0x8300, 0x8500) || \
	 REG_RANGE((reg), 0x8C00, 0x8D00) || \
	 REG_RANGE((reg), 0xB000, 0xB480) || \
564 565
	 REG_RANGE((reg), 0xE000, 0xE900) || \
	 REG_RANGE((reg), 0x24400, 0x24800))
566 567

#define FORCEWAKE_GEN9_MEDIA_RANGE_OFFSET(reg) \
568 569
	(REG_RANGE((reg), 0x8130, 0x8140) || \
	 REG_RANGE((reg), 0x8800, 0x8A00) || \
570 571 572 573 574 575 576 577 578
	 REG_RANGE((reg), 0xD000, 0xD800) || \
	 REG_RANGE((reg), 0x12000, 0x14000) || \
	 REG_RANGE((reg), 0x1A000, 0x1EA00) || \
	 REG_RANGE((reg), 0x30000, 0x40000))

#define FORCEWAKE_GEN9_COMMON_RANGE_OFFSET(reg) \
	REG_RANGE((reg), 0x9400, 0x9800)

#define FORCEWAKE_GEN9_BLITTER_RANGE_OFFSET(reg) \
579
	((reg) < 0x40000 && \
580 581 582 583 584
	 !FORCEWAKE_GEN9_UNCORE_RANGE_OFFSET(reg) && \
	 !FORCEWAKE_GEN9_RENDER_RANGE_OFFSET(reg) && \
	 !FORCEWAKE_GEN9_MEDIA_RANGE_OFFSET(reg) && \
	 !FORCEWAKE_GEN9_COMMON_RANGE_OFFSET(reg))

585 586 587 588 589 590
static void
ilk_dummy_write(struct drm_i915_private *dev_priv)
{
	/* WaIssueDummyWriteToWakeupFromRC6:ilk Issue a dummy write to wake up
	 * the chip from rc6 before touching it for real. MI_MODE is masked,
	 * hence harmless to write 0 into. */
591
	__raw_i915_write32(dev_priv, MI_MODE, 0);
592 593 594
}

static void
595 596
hsw_unclaimed_reg_debug(struct drm_i915_private *dev_priv, u32 reg, bool read,
			bool before)
597
{
598 599 600 601 602 603
	const char *op = read ? "reading" : "writing to";
	const char *when = before ? "before" : "after";

	if (!i915.mmio_debug)
		return;

604
	if (__raw_i915_read32(dev_priv, FPGA_DBG) & FPGA_DBG_RM_NOCLAIM) {
605 606
		WARN(1, "Unclaimed register detected %s %s register 0x%x\n",
		     when, op, reg);
607
		__raw_i915_write32(dev_priv, FPGA_DBG, FPGA_DBG_RM_NOCLAIM);
608
		i915.mmio_debug--; /* Only report the first N failures */
609 610 611 612
	}
}

static void
613
hsw_unclaimed_reg_detect(struct drm_i915_private *dev_priv)
614
{
615 616 617
	static bool mmio_debug_once = true;

	if (i915.mmio_debug || !mmio_debug_once)
618 619
		return;

620
	if (__raw_i915_read32(dev_priv, FPGA_DBG) & FPGA_DBG_RM_NOCLAIM) {
621 622 623
		DRM_DEBUG("Unclaimed register detected, "
			  "enabling oneshot unclaimed register reporting. "
			  "Please use i915.mmio_debug=N for more information.\n");
624
		__raw_i915_write32(dev_priv, FPGA_DBG, FPGA_DBG_RM_NOCLAIM);
625
		i915.mmio_debug = mmio_debug_once--;
626 627 628
	}
}

629
#define GEN2_READ_HEADER(x) \
B
Ben Widawsky 已提交
630
	u##x val = 0; \
631
	assert_device_not_suspended(dev_priv);
B
Ben Widawsky 已提交
632

633
#define GEN2_READ_FOOTER \
B
Ben Widawsky 已提交
634 635 636
	trace_i915_reg_rw(false, reg, val, sizeof(val), trace); \
	return val

637
#define __gen2_read(x) \
638
static u##x \
639 640
gen2_read##x(struct drm_i915_private *dev_priv, off_t reg, bool trace) { \
	GEN2_READ_HEADER(x); \
641
	val = __raw_i915_read##x(dev_priv, reg); \
642
	GEN2_READ_FOOTER; \
643 644 645 646 647
}

#define __gen5_read(x) \
static u##x \
gen5_read##x(struct drm_i915_private *dev_priv, off_t reg, bool trace) { \
648
	GEN2_READ_HEADER(x); \
649 650
	ilk_dummy_write(dev_priv); \
	val = __raw_i915_read##x(dev_priv, reg); \
651
	GEN2_READ_FOOTER; \
652 653
}

654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679
__gen5_read(8)
__gen5_read(16)
__gen5_read(32)
__gen5_read(64)
__gen2_read(8)
__gen2_read(16)
__gen2_read(32)
__gen2_read(64)

#undef __gen5_read
#undef __gen2_read

#undef GEN2_READ_FOOTER
#undef GEN2_READ_HEADER

#define GEN6_READ_HEADER(x) \
	unsigned long irqflags; \
	u##x val = 0; \
	assert_device_not_suspended(dev_priv); \
	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags)

#define GEN6_READ_FOOTER \
	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags); \
	trace_i915_reg_rw(false, reg, val, sizeof(val), trace); \
	return val

680
static inline void __force_wake_get(struct drm_i915_private *dev_priv,
681
				    enum forcewake_domains fw_domains)
682 683
{
	struct intel_uncore_forcewake_domain *domain;
684
	enum forcewake_domain_id id;
685 686 687 688 689

	if (WARN_ON(!fw_domains))
		return;

	/* Ideally GCC would be constant-fold and eliminate this loop */
690
	for_each_fw_domain_mask(domain, fw_domains, dev_priv, id) {
691
		if (domain->wake_count) {
692
			fw_domains &= ~(1 << id);
693 694 695 696
			continue;
		}

		domain->wake_count++;
697
		fw_domain_arm_timer(domain);
698 699 700 701 702 703
	}

	if (fw_domains)
		dev_priv->uncore.funcs.force_wake_get(dev_priv, fw_domains);
}

704 705 706 707 708 709 710 711
#define __vgpu_read(x) \
static u##x \
vgpu_read##x(struct drm_i915_private *dev_priv, off_t reg, bool trace) { \
	GEN6_READ_HEADER(x); \
	val = __raw_i915_read##x(dev_priv, reg); \
	GEN6_READ_FOOTER; \
}

712 713 714
#define __gen6_read(x) \
static u##x \
gen6_read##x(struct drm_i915_private *dev_priv, off_t reg, bool trace) { \
715
	GEN6_READ_HEADER(x); \
716
	hsw_unclaimed_reg_debug(dev_priv, reg, true, true); \
717
	if (NEEDS_FORCE_WAKE(reg)) \
718
		__force_wake_get(dev_priv, FORCEWAKE_RENDER); \
719
	val = __raw_i915_read##x(dev_priv, reg); \
720
	hsw_unclaimed_reg_debug(dev_priv, reg, true, false); \
721
	GEN6_READ_FOOTER; \
722 723
}

724 725 726
#define __vlv_read(x) \
static u##x \
vlv_read##x(struct drm_i915_private *dev_priv, off_t reg, bool trace) { \
727
	enum forcewake_domains fw_engine = 0; \
728
	GEN6_READ_HEADER(x); \
729 730 731
	if (!NEEDS_FORCE_WAKE(reg)) \
		fw_engine = 0; \
	else if (FORCEWAKE_VLV_RENDER_RANGE_OFFSET(reg)) \
732
		fw_engine = FORCEWAKE_RENDER; \
733
	else if (FORCEWAKE_VLV_MEDIA_RANGE_OFFSET(reg)) \
734 735 736
		fw_engine = FORCEWAKE_MEDIA; \
	if (fw_engine) \
		__force_wake_get(dev_priv, fw_engine); \
737
	val = __raw_i915_read##x(dev_priv, reg); \
738
	GEN6_READ_FOOTER; \
739 740
}

741 742 743
#define __chv_read(x) \
static u##x \
chv_read##x(struct drm_i915_private *dev_priv, off_t reg, bool trace) { \
744
	enum forcewake_domains fw_engine = 0; \
745
	GEN6_READ_HEADER(x); \
746 747 748
	if (!NEEDS_FORCE_WAKE(reg)) \
		fw_engine = 0; \
	else if (FORCEWAKE_CHV_RENDER_RANGE_OFFSET(reg)) \
749
		fw_engine = FORCEWAKE_RENDER; \
750
	else if (FORCEWAKE_CHV_MEDIA_RANGE_OFFSET(reg)) \
751
		fw_engine = FORCEWAKE_MEDIA; \
752
	else if (FORCEWAKE_CHV_COMMON_RANGE_OFFSET(reg)) \
753 754 755
		fw_engine = FORCEWAKE_RENDER | FORCEWAKE_MEDIA; \
	if (fw_engine) \
		__force_wake_get(dev_priv, fw_engine); \
756
	val = __raw_i915_read##x(dev_priv, reg); \
757
	GEN6_READ_FOOTER; \
758
}
759

760
#define SKL_NEEDS_FORCE_WAKE(reg) \
761
	((reg) < 0x40000 && !FORCEWAKE_GEN9_UNCORE_RANGE_OFFSET(reg))
762 763 764 765

#define __gen9_read(x) \
static u##x \
gen9_read##x(struct drm_i915_private *dev_priv, off_t reg, bool trace) { \
766
	enum forcewake_domains fw_engine; \
767
	GEN6_READ_HEADER(x); \
768
	hsw_unclaimed_reg_debug(dev_priv, reg, true, true); \
769
	if (!SKL_NEEDS_FORCE_WAKE(reg)) \
770
		fw_engine = 0; \
771
	else if (FORCEWAKE_GEN9_RENDER_RANGE_OFFSET(reg)) \
772 773 774 775 776 777 778 779 780 781
		fw_engine = FORCEWAKE_RENDER; \
	else if (FORCEWAKE_GEN9_MEDIA_RANGE_OFFSET(reg)) \
		fw_engine = FORCEWAKE_MEDIA; \
	else if (FORCEWAKE_GEN9_COMMON_RANGE_OFFSET(reg)) \
		fw_engine = FORCEWAKE_RENDER | FORCEWAKE_MEDIA; \
	else \
		fw_engine = FORCEWAKE_BLITTER; \
	if (fw_engine) \
		__force_wake_get(dev_priv, fw_engine); \
	val = __raw_i915_read##x(dev_priv, reg); \
782
	hsw_unclaimed_reg_debug(dev_priv, reg, true, false); \
783
	GEN6_READ_FOOTER; \
784 785
}

786 787 788 789
__vgpu_read(8)
__vgpu_read(16)
__vgpu_read(32)
__vgpu_read(64)
790 791 792 793
__gen9_read(8)
__gen9_read(16)
__gen9_read(32)
__gen9_read(64)
794 795 796 797
__chv_read(8)
__chv_read(16)
__chv_read(32)
__chv_read(64)
798 799 800 801
__vlv_read(8)
__vlv_read(16)
__vlv_read(32)
__vlv_read(64)
802 803 804 805 806
__gen6_read(8)
__gen6_read(16)
__gen6_read(32)
__gen6_read(64)

807
#undef __gen9_read
808
#undef __chv_read
809
#undef __vlv_read
810
#undef __gen6_read
811
#undef __vgpu_read
812 813
#undef GEN6_READ_FOOTER
#undef GEN6_READ_HEADER
B
Ben Widawsky 已提交
814

815
#define GEN2_WRITE_HEADER \
B
Ben Widawsky 已提交
816
	trace_i915_reg_rw(true, reg, val, sizeof(val), trace); \
817
	assert_device_not_suspended(dev_priv); \
818

819
#define GEN2_WRITE_FOOTER
V
Ville Syrjälä 已提交
820

821
#define __gen2_write(x) \
822
static void \
823 824
gen2_write##x(struct drm_i915_private *dev_priv, off_t reg, u##x val, bool trace) { \
	GEN2_WRITE_HEADER; \
825
	__raw_i915_write##x(dev_priv, reg, val); \
826
	GEN2_WRITE_FOOTER; \
827 828 829 830 831
}

#define __gen5_write(x) \
static void \
gen5_write##x(struct drm_i915_private *dev_priv, off_t reg, u##x val, bool trace) { \
832
	GEN2_WRITE_HEADER; \
833 834
	ilk_dummy_write(dev_priv); \
	__raw_i915_write##x(dev_priv, reg, val); \
835
	GEN2_WRITE_FOOTER; \
836 837
}

838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861
__gen5_write(8)
__gen5_write(16)
__gen5_write(32)
__gen5_write(64)
__gen2_write(8)
__gen2_write(16)
__gen2_write(32)
__gen2_write(64)

#undef __gen5_write
#undef __gen2_write

#undef GEN2_WRITE_FOOTER
#undef GEN2_WRITE_HEADER

#define GEN6_WRITE_HEADER \
	unsigned long irqflags; \
	trace_i915_reg_rw(true, reg, val, sizeof(val), trace); \
	assert_device_not_suspended(dev_priv); \
	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags)

#define GEN6_WRITE_FOOTER \
	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags)

862 863 864 865
#define __gen6_write(x) \
static void \
gen6_write##x(struct drm_i915_private *dev_priv, off_t reg, u##x val, bool trace) { \
	u32 __fifo_ret = 0; \
866
	GEN6_WRITE_HEADER; \
867
	if (NEEDS_FORCE_WAKE(reg)) { \
868 869 870 871 872 873
		__fifo_ret = __gen6_gt_wait_for_fifo(dev_priv); \
	} \
	__raw_i915_write##x(dev_priv, reg, val); \
	if (unlikely(__fifo_ret)) { \
		gen6_gt_check_fifodbg(dev_priv); \
	} \
874
	GEN6_WRITE_FOOTER; \
875 876 877 878 879
}

#define __hsw_write(x) \
static void \
hsw_write##x(struct drm_i915_private *dev_priv, off_t reg, u##x val, bool trace) { \
880
	u32 __fifo_ret = 0; \
881
	GEN6_WRITE_HEADER; \
882
	if (NEEDS_FORCE_WAKE(reg)) { \
883 884
		__fifo_ret = __gen6_gt_wait_for_fifo(dev_priv); \
	} \
885
	hsw_unclaimed_reg_debug(dev_priv, reg, false, true); \
886
	__raw_i915_write##x(dev_priv, reg, val); \
887 888 889
	if (unlikely(__fifo_ret)) { \
		gen6_gt_check_fifodbg(dev_priv); \
	} \
890 891
	hsw_unclaimed_reg_debug(dev_priv, reg, false, false); \
	hsw_unclaimed_reg_detect(dev_priv); \
892
	GEN6_WRITE_FOOTER; \
893
}
894

895 896 897 898 899 900 901 902
#define __vgpu_write(x) \
static void vgpu_write##x(struct drm_i915_private *dev_priv, \
			  off_t reg, u##x val, bool trace) { \
	GEN6_WRITE_HEADER; \
	__raw_i915_write##x(dev_priv, reg, val); \
	GEN6_WRITE_FOOTER; \
}

903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926
static const u32 gen8_shadowed_regs[] = {
	FORCEWAKE_MT,
	GEN6_RPNSWREQ,
	GEN6_RC_VIDEO_FREQ,
	RING_TAIL(RENDER_RING_BASE),
	RING_TAIL(GEN6_BSD_RING_BASE),
	RING_TAIL(VEBOX_RING_BASE),
	RING_TAIL(BLT_RING_BASE),
	/* TODO: Other registers are not yet used */
};

static bool is_gen8_shadowed(struct drm_i915_private *dev_priv, u32 reg)
{
	int i;
	for (i = 0; i < ARRAY_SIZE(gen8_shadowed_regs); i++)
		if (reg == gen8_shadowed_regs[i])
			return true;

	return false;
}

#define __gen8_write(x) \
static void \
gen8_write##x(struct drm_i915_private *dev_priv, off_t reg, u##x val, bool trace) { \
927
	GEN6_WRITE_HEADER; \
928
	hsw_unclaimed_reg_debug(dev_priv, reg, false, true); \
929
	if (NEEDS_FORCE_WAKE(reg) && !is_gen8_shadowed(dev_priv, reg)) \
930 931
		__force_wake_get(dev_priv, FORCEWAKE_RENDER); \
	__raw_i915_write##x(dev_priv, reg, val); \
932 933
	hsw_unclaimed_reg_debug(dev_priv, reg, false, false); \
	hsw_unclaimed_reg_detect(dev_priv); \
934
	GEN6_WRITE_FOOTER; \
935 936
}

937 938 939
#define __chv_write(x) \
static void \
chv_write##x(struct drm_i915_private *dev_priv, off_t reg, u##x val, bool trace) { \
940
	enum forcewake_domains fw_engine = 0; \
941
	GEN6_WRITE_HEADER; \
942 943
	if (!NEEDS_FORCE_WAKE(reg) || \
	    is_gen8_shadowed(dev_priv, reg)) \
944 945 946 947 948 949 950 951 952
		fw_engine = 0; \
	else if (FORCEWAKE_CHV_RENDER_RANGE_OFFSET(reg)) \
		fw_engine = FORCEWAKE_RENDER; \
	else if (FORCEWAKE_CHV_MEDIA_RANGE_OFFSET(reg)) \
		fw_engine = FORCEWAKE_MEDIA; \
	else if (FORCEWAKE_CHV_COMMON_RANGE_OFFSET(reg)) \
		fw_engine = FORCEWAKE_RENDER | FORCEWAKE_MEDIA; \
	if (fw_engine) \
		__force_wake_get(dev_priv, fw_engine); \
953
	__raw_i915_write##x(dev_priv, reg, val); \
954
	GEN6_WRITE_FOOTER; \
955 956
}

Z
Zhe Wang 已提交
957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979
static const u32 gen9_shadowed_regs[] = {
	RING_TAIL(RENDER_RING_BASE),
	RING_TAIL(GEN6_BSD_RING_BASE),
	RING_TAIL(VEBOX_RING_BASE),
	RING_TAIL(BLT_RING_BASE),
	FORCEWAKE_BLITTER_GEN9,
	FORCEWAKE_RENDER_GEN9,
	FORCEWAKE_MEDIA_GEN9,
	GEN6_RPNSWREQ,
	GEN6_RC_VIDEO_FREQ,
	/* TODO: Other registers are not yet used */
};

static bool is_gen9_shadowed(struct drm_i915_private *dev_priv, u32 reg)
{
	int i;
	for (i = 0; i < ARRAY_SIZE(gen9_shadowed_regs); i++)
		if (reg == gen9_shadowed_regs[i])
			return true;

	return false;
}

980 981 982 983
#define __gen9_write(x) \
static void \
gen9_write##x(struct drm_i915_private *dev_priv, off_t reg, u##x val, \
		bool trace) { \
984
	enum forcewake_domains fw_engine; \
985
	GEN6_WRITE_HEADER; \
986
	hsw_unclaimed_reg_debug(dev_priv, reg, false, true); \
987
	if (!SKL_NEEDS_FORCE_WAKE(reg) || \
988 989 990 991 992 993 994 995 996 997 998 999 1000
	    is_gen9_shadowed(dev_priv, reg)) \
		fw_engine = 0; \
	else if (FORCEWAKE_GEN9_RENDER_RANGE_OFFSET(reg)) \
		fw_engine = FORCEWAKE_RENDER; \
	else if (FORCEWAKE_GEN9_MEDIA_RANGE_OFFSET(reg)) \
		fw_engine = FORCEWAKE_MEDIA; \
	else if (FORCEWAKE_GEN9_COMMON_RANGE_OFFSET(reg)) \
		fw_engine = FORCEWAKE_RENDER | FORCEWAKE_MEDIA; \
	else \
		fw_engine = FORCEWAKE_BLITTER; \
	if (fw_engine) \
		__force_wake_get(dev_priv, fw_engine); \
	__raw_i915_write##x(dev_priv, reg, val); \
1001 1002
	hsw_unclaimed_reg_debug(dev_priv, reg, false, false); \
	hsw_unclaimed_reg_detect(dev_priv); \
1003
	GEN6_WRITE_FOOTER; \
1004 1005 1006 1007 1008 1009
}

__gen9_write(8)
__gen9_write(16)
__gen9_write(32)
__gen9_write(64)
1010 1011 1012 1013
__chv_write(8)
__chv_write(16)
__chv_write(32)
__chv_write(64)
1014 1015 1016 1017
__gen8_write(8)
__gen8_write(16)
__gen8_write(32)
__gen8_write(64)
1018 1019 1020 1021 1022 1023 1024 1025
__hsw_write(8)
__hsw_write(16)
__hsw_write(32)
__hsw_write(64)
__gen6_write(8)
__gen6_write(16)
__gen6_write(32)
__gen6_write(64)
1026 1027 1028 1029
__vgpu_write(8)
__vgpu_write(16)
__vgpu_write(32)
__vgpu_write(64)
1030

1031
#undef __gen9_write
1032
#undef __chv_write
1033
#undef __gen8_write
1034 1035
#undef __hsw_write
#undef __gen6_write
1036
#undef __vgpu_write
1037 1038
#undef GEN6_WRITE_FOOTER
#undef GEN6_WRITE_HEADER
1039

1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
#define ASSIGN_WRITE_MMIO_VFUNCS(x) \
do { \
	dev_priv->uncore.funcs.mmio_writeb = x##_write8; \
	dev_priv->uncore.funcs.mmio_writew = x##_write16; \
	dev_priv->uncore.funcs.mmio_writel = x##_write32; \
	dev_priv->uncore.funcs.mmio_writeq = x##_write64; \
} while (0)

#define ASSIGN_READ_MMIO_VFUNCS(x) \
do { \
	dev_priv->uncore.funcs.mmio_readb = x##_read8; \
	dev_priv->uncore.funcs.mmio_readw = x##_read16; \
	dev_priv->uncore.funcs.mmio_readl = x##_read32; \
	dev_priv->uncore.funcs.mmio_readq = x##_read64; \
} while (0)

1056 1057

static void fw_domain_init(struct drm_i915_private *dev_priv,
1058 1059
			   enum forcewake_domain_id domain_id,
			   u32 reg_set, u32 reg_ack)
1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078
{
	struct intel_uncore_forcewake_domain *d;

	if (WARN_ON(domain_id >= FW_DOMAIN_ID_COUNT))
		return;

	d = &dev_priv->uncore.fw_domain[domain_id];

	WARN_ON(d->wake_count);

	d->wake_count = 0;
	d->reg_set = reg_set;
	d->reg_ack = reg_ack;

	if (IS_GEN6(dev_priv)) {
		d->val_reset = 0;
		d->val_set = FORCEWAKE_KERNEL;
		d->val_clear = 0;
	} else {
1079
		/* WaRsClearFWBitsAtReset:bdw,skl */
1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094
		d->val_reset = _MASKED_BIT_DISABLE(0xffff);
		d->val_set = _MASKED_BIT_ENABLE(FORCEWAKE_KERNEL);
		d->val_clear = _MASKED_BIT_DISABLE(FORCEWAKE_KERNEL);
	}

	if (IS_VALLEYVIEW(dev_priv))
		d->reg_post = FORCEWAKE_ACK_VLV;
	else if (IS_GEN6(dev_priv) || IS_GEN7(dev_priv) || IS_GEN8(dev_priv))
		d->reg_post = ECOBUS;
	else
		d->reg_post = 0;

	d->i915 = dev_priv;
	d->id = domain_id;

1095
	setup_timer(&d->timer, intel_uncore_fw_release_timer, (unsigned long)d);
1096 1097

	dev_priv->uncore.fw_domains |= (1 << domain_id);
1098 1099

	fw_domain_reset(d);
1100 1101
}

1102
static void intel_uncore_fw_domains_init(struct drm_device *dev)
1103 1104 1105
{
	struct drm_i915_private *dev_priv = dev->dev_private;

1106 1107 1108
	if (INTEL_INFO(dev_priv->dev)->gen <= 5)
		return;

Z
Zhe Wang 已提交
1109
	if (IS_GEN9(dev)) {
1110 1111 1112 1113 1114 1115 1116 1117 1118 1119
		dev_priv->uncore.funcs.force_wake_get = fw_domains_get;
		dev_priv->uncore.funcs.force_wake_put = fw_domains_put;
		fw_domain_init(dev_priv, FW_DOMAIN_ID_RENDER,
			       FORCEWAKE_RENDER_GEN9,
			       FORCEWAKE_ACK_RENDER_GEN9);
		fw_domain_init(dev_priv, FW_DOMAIN_ID_BLITTER,
			       FORCEWAKE_BLITTER_GEN9,
			       FORCEWAKE_ACK_BLITTER_GEN9);
		fw_domain_init(dev_priv, FW_DOMAIN_ID_MEDIA,
			       FORCEWAKE_MEDIA_GEN9, FORCEWAKE_ACK_MEDIA_GEN9);
Z
Zhe Wang 已提交
1120
	} else if (IS_VALLEYVIEW(dev)) {
1121
		dev_priv->uncore.funcs.force_wake_get = fw_domains_get;
1122 1123 1124 1125 1126
		if (!IS_CHERRYVIEW(dev))
			dev_priv->uncore.funcs.force_wake_put =
				fw_domains_put_with_fifo;
		else
			dev_priv->uncore.funcs.force_wake_put = fw_domains_put;
1127 1128 1129 1130
		fw_domain_init(dev_priv, FW_DOMAIN_ID_RENDER,
			       FORCEWAKE_VLV, FORCEWAKE_ACK_VLV);
		fw_domain_init(dev_priv, FW_DOMAIN_ID_MEDIA,
			       FORCEWAKE_MEDIA_VLV, FORCEWAKE_ACK_MEDIA_VLV);
1131
	} else if (IS_HASWELL(dev) || IS_BROADWELL(dev)) {
1132 1133 1134 1135 1136
		dev_priv->uncore.funcs.force_wake_get =
			fw_domains_get_with_thread_status;
		dev_priv->uncore.funcs.force_wake_put = fw_domains_put;
		fw_domain_init(dev_priv, FW_DOMAIN_ID_RENDER,
			       FORCEWAKE_MT, FORCEWAKE_ACK_HSW);
1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148
	} else if (IS_IVYBRIDGE(dev)) {
		u32 ecobus;

		/* IVB configs may use multi-threaded forcewake */

		/* A small trick here - if the bios hasn't configured
		 * MT forcewake, and if the device is in RC6, then
		 * force_wake_mt_get will not wake the device and the
		 * ECOBUS read will return zero. Which will be
		 * (correctly) interpreted by the test below as MT
		 * forcewake being disabled.
		 */
1149 1150 1151 1152 1153
		dev_priv->uncore.funcs.force_wake_get =
			fw_domains_get_with_thread_status;
		dev_priv->uncore.funcs.force_wake_put =
			fw_domains_put_with_fifo;

1154 1155
		/* We need to init first for ECOBUS access and then
		 * determine later if we want to reinit, in case of MT access is
1156 1157 1158
		 * not working. In this stage we don't know which flavour this
		 * ivb is, so it is better to reset also the gen6 fw registers
		 * before the ecobus check.
1159
		 */
1160 1161 1162 1163

		__raw_i915_write32(dev_priv, FORCEWAKE, 0);
		__raw_posting_read(dev_priv, ECOBUS);

1164 1165
		fw_domain_init(dev_priv, FW_DOMAIN_ID_RENDER,
			       FORCEWAKE_MT, FORCEWAKE_MT_ACK);
1166

1167
		mutex_lock(&dev->struct_mutex);
1168
		fw_domains_get_with_thread_status(dev_priv, FORCEWAKE_ALL);
1169
		ecobus = __raw_i915_read32(dev_priv, ECOBUS);
1170
		fw_domains_put_with_fifo(dev_priv, FORCEWAKE_ALL);
1171 1172
		mutex_unlock(&dev->struct_mutex);

1173
		if (!(ecobus & FORCEWAKE_MT_ENABLE)) {
1174 1175
			DRM_INFO("No MT forcewake available on Ivybridge, this can result in issues\n");
			DRM_INFO("when using vblank-synced partial screen updates.\n");
1176 1177
			fw_domain_init(dev_priv, FW_DOMAIN_ID_RENDER,
				       FORCEWAKE, FORCEWAKE_ACK);
1178 1179 1180
		}
	} else if (IS_GEN6(dev)) {
		dev_priv->uncore.funcs.force_wake_get =
1181
			fw_domains_get_with_thread_status;
1182
		dev_priv->uncore.funcs.force_wake_put =
1183 1184 1185
			fw_domains_put_with_fifo;
		fw_domain_init(dev_priv, FW_DOMAIN_ID_RENDER,
			       FORCEWAKE, FORCEWAKE_ACK);
1186
	}
1187 1188 1189

	/* All future platforms are expected to require complex power gating */
	WARN_ON(dev_priv->uncore.fw_domains == 0);
1190 1191 1192 1193 1194 1195
}

void intel_uncore_init(struct drm_device *dev)
{
	struct drm_i915_private *dev_priv = dev->dev_private;

1196 1197
	i915_check_vgpu(dev);

1198 1199 1200
	intel_uncore_ellc_detect(dev);
	intel_uncore_fw_domains_init(dev);
	__intel_uncore_early_sanitize(dev, false);
1201

1202
	switch (INTEL_INFO(dev)->gen) {
1203
	default:
1204 1205 1206 1207 1208
	case 9:
		ASSIGN_WRITE_MMIO_VFUNCS(gen9);
		ASSIGN_READ_MMIO_VFUNCS(gen9);
		break;
	case 8:
1209
		if (IS_CHERRYVIEW(dev)) {
1210 1211
			ASSIGN_WRITE_MMIO_VFUNCS(chv);
			ASSIGN_READ_MMIO_VFUNCS(chv);
1212 1213

		} else {
1214 1215
			ASSIGN_WRITE_MMIO_VFUNCS(gen8);
			ASSIGN_READ_MMIO_VFUNCS(gen6);
1216
		}
1217
		break;
1218 1219
	case 7:
	case 6:
1220
		if (IS_HASWELL(dev)) {
1221
			ASSIGN_WRITE_MMIO_VFUNCS(hsw);
1222
		} else {
1223
			ASSIGN_WRITE_MMIO_VFUNCS(gen6);
1224
		}
1225 1226

		if (IS_VALLEYVIEW(dev)) {
1227
			ASSIGN_READ_MMIO_VFUNCS(vlv);
1228
		} else {
1229
			ASSIGN_READ_MMIO_VFUNCS(gen6);
1230
		}
1231 1232
		break;
	case 5:
1233 1234
		ASSIGN_WRITE_MMIO_VFUNCS(gen5);
		ASSIGN_READ_MMIO_VFUNCS(gen5);
1235 1236 1237 1238
		break;
	case 4:
	case 3:
	case 2:
1239 1240
		ASSIGN_WRITE_MMIO_VFUNCS(gen2);
		ASSIGN_READ_MMIO_VFUNCS(gen2);
1241 1242
		break;
	}
1243

1244 1245 1246 1247 1248
	if (intel_vgpu_active(dev)) {
		ASSIGN_WRITE_MMIO_VFUNCS(vgpu);
		ASSIGN_READ_MMIO_VFUNCS(vgpu);
	}

1249
	i915_check_and_clear_faults(dev);
1250
}
1251 1252
#undef ASSIGN_WRITE_MMIO_VFUNCS
#undef ASSIGN_READ_MMIO_VFUNCS
1253 1254 1255 1256 1257

void intel_uncore_fini(struct drm_device *dev)
{
	/* Paranoia: make sure we have disabled everything before we exit. */
	intel_uncore_sanitize(dev);
1258
	intel_uncore_forcewake_reset(dev, false);
1259 1260
}

1261 1262
#define GEN_RANGE(l, h) GENMASK(h, l)

1263 1264 1265
static const struct register_whitelist {
	uint64_t offset;
	uint32_t size;
1266 1267
	/* supported gens, 0x10 for 4, 0x30 for 4 and 5, etc. */
	uint32_t gen_bitmask;
1268
} whitelist[] = {
1269
	{ RING_TIMESTAMP(RENDER_RING_BASE), 8, GEN_RANGE(4, 9) },
1270 1271 1272 1273 1274 1275 1276 1277
};

int i915_reg_read_ioctl(struct drm_device *dev,
			void *data, struct drm_file *file)
{
	struct drm_i915_private *dev_priv = dev->dev_private;
	struct drm_i915_reg_read *reg = data;
	struct register_whitelist const *entry = whitelist;
1278 1279
	unsigned size;
	u64 offset;
1280
	int i, ret = 0;
1281 1282

	for (i = 0; i < ARRAY_SIZE(whitelist); i++, entry++) {
1283
		if (entry->offset == (reg->offset & -entry->size) &&
1284 1285 1286 1287 1288 1289 1290
		    (1 << INTEL_INFO(dev)->gen & entry->gen_bitmask))
			break;
	}

	if (i == ARRAY_SIZE(whitelist))
		return -EINVAL;

1291 1292 1293 1294 1295 1296 1297 1298
	/* We use the low bits to encode extra flags as the register should
	 * be naturally aligned (and those that are not so aligned merely
	 * limit the available flags for that register).
	 */
	offset = entry->offset;
	size = entry->size;
	size |= reg->offset ^ offset;

1299 1300
	intel_runtime_pm_get(dev_priv);

1301 1302 1303 1304
	switch (size) {
	case 8 | 1:
		reg->val = I915_READ64_2x32(offset, offset+4);
		break;
1305
	case 8:
1306
		reg->val = I915_READ64(offset);
1307 1308
		break;
	case 4:
1309
		reg->val = I915_READ(offset);
1310 1311
		break;
	case 2:
1312
		reg->val = I915_READ16(offset);
1313 1314
		break;
	case 1:
1315
		reg->val = I915_READ8(offset);
1316 1317
		break;
	default:
1318 1319
		ret = -EINVAL;
		goto out;
1320 1321
	}

1322 1323 1324
out:
	intel_runtime_pm_put(dev_priv);
	return ret;
1325 1326
}

1327 1328 1329 1330 1331 1332
int i915_get_reset_stats_ioctl(struct drm_device *dev,
			       void *data, struct drm_file *file)
{
	struct drm_i915_private *dev_priv = dev->dev_private;
	struct drm_i915_reset_stats *args = data;
	struct i915_ctx_hang_stats *hs;
1333
	struct intel_context *ctx;
1334 1335
	int ret;

1336 1337 1338
	if (args->flags || args->pad)
		return -EINVAL;

1339
	if (args->ctx_id == DEFAULT_CONTEXT_HANDLE && !capable(CAP_SYS_ADMIN))
1340 1341 1342 1343 1344 1345
		return -EPERM;

	ret = mutex_lock_interruptible(&dev->struct_mutex);
	if (ret)
		return ret;

1346 1347
	ctx = i915_gem_context_get(file->driver_priv, args->ctx_id);
	if (IS_ERR(ctx)) {
1348
		mutex_unlock(&dev->struct_mutex);
1349
		return PTR_ERR(ctx);
1350
	}
1351
	hs = &ctx->hang_stats;
1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365

	if (capable(CAP_SYS_ADMIN))
		args->reset_count = i915_reset_count(&dev_priv->gpu_error);
	else
		args->reset_count = 0;

	args->batch_active = hs->batch_active;
	args->batch_pending = hs->batch_pending;

	mutex_unlock(&dev->struct_mutex);

	return 0;
}

1366
static int i915_reset_complete(struct drm_device *dev)
1367 1368
{
	u8 gdrst;
1369
	pci_read_config_byte(dev->pdev, I915_GDRST, &gdrst);
V
Ville Syrjälä 已提交
1370
	return (gdrst & GRDOM_RESET_STATUS) == 0;
1371 1372
}

1373
static int i915_do_reset(struct drm_device *dev)
1374
{
V
Ville Syrjälä 已提交
1375
	/* assert reset for at least 20 usec */
1376
	pci_write_config_byte(dev->pdev, I915_GDRST, GRDOM_RESET_ENABLE);
V
Ville Syrjälä 已提交
1377
	udelay(20);
1378
	pci_write_config_byte(dev->pdev, I915_GDRST, 0);
1379

1380
	return wait_for(i915_reset_complete(dev), 500);
V
Ville Syrjälä 已提交
1381 1382 1383 1384 1385
}

static int g4x_reset_complete(struct drm_device *dev)
{
	u8 gdrst;
1386
	pci_read_config_byte(dev->pdev, I915_GDRST, &gdrst);
V
Ville Syrjälä 已提交
1387
	return (gdrst & GRDOM_RESET_ENABLE) == 0;
1388 1389
}

1390 1391 1392 1393 1394 1395
static int g33_do_reset(struct drm_device *dev)
{
	pci_write_config_byte(dev->pdev, I915_GDRST, GRDOM_RESET_ENABLE);
	return wait_for(g4x_reset_complete(dev), 500);
}

1396 1397 1398 1399 1400
static int g4x_do_reset(struct drm_device *dev)
{
	struct drm_i915_private *dev_priv = dev->dev_private;
	int ret;

1401
	pci_write_config_byte(dev->pdev, I915_GDRST,
1402
			      GRDOM_RENDER | GRDOM_RESET_ENABLE);
V
Ville Syrjälä 已提交
1403
	ret =  wait_for(g4x_reset_complete(dev), 500);
1404 1405 1406 1407 1408 1409 1410
	if (ret)
		return ret;

	/* WaVcpClkGateDisableForMediaReset:ctg,elk */
	I915_WRITE(VDECCLK_GATE_D, I915_READ(VDECCLK_GATE_D) | VCP_UNIT_CLOCK_GATE_DISABLE);
	POSTING_READ(VDECCLK_GATE_D);

1411
	pci_write_config_byte(dev->pdev, I915_GDRST,
1412
			      GRDOM_MEDIA | GRDOM_RESET_ENABLE);
V
Ville Syrjälä 已提交
1413
	ret =  wait_for(g4x_reset_complete(dev), 500);
1414 1415 1416 1417 1418 1419 1420
	if (ret)
		return ret;

	/* WaVcpClkGateDisableForMediaReset:ctg,elk */
	I915_WRITE(VDECCLK_GATE_D, I915_READ(VDECCLK_GATE_D) & ~VCP_UNIT_CLOCK_GATE_DISABLE);
	POSTING_READ(VDECCLK_GATE_D);

1421
	pci_write_config_byte(dev->pdev, I915_GDRST, 0);
1422 1423 1424 1425

	return 0;
}

1426 1427 1428 1429 1430
static int ironlake_do_reset(struct drm_device *dev)
{
	struct drm_i915_private *dev_priv = dev->dev_private;
	int ret;

1431
	I915_WRITE(ILK_GDSR,
1432
		   ILK_GRDOM_RENDER | ILK_GRDOM_RESET_ENABLE);
1433
	ret = wait_for((I915_READ(ILK_GDSR) &
1434
			ILK_GRDOM_RESET_ENABLE) == 0, 500);
1435 1436 1437
	if (ret)
		return ret;

1438
	I915_WRITE(ILK_GDSR,
1439
		   ILK_GRDOM_MEDIA | ILK_GRDOM_RESET_ENABLE);
1440
	ret = wait_for((I915_READ(ILK_GDSR) &
1441 1442 1443 1444
			ILK_GRDOM_RESET_ENABLE) == 0, 500);
	if (ret)
		return ret;

1445
	I915_WRITE(ILK_GDSR, 0);
1446 1447

	return 0;
1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460
}

static int gen6_do_reset(struct drm_device *dev)
{
	struct drm_i915_private *dev_priv = dev->dev_private;
	int	ret;

	/* Reset the chip */

	/* GEN6_GDRST is not in the gt power well, no need to check
	 * for fifo space for the write or forcewake the chip for
	 * the read
	 */
1461
	__raw_i915_write32(dev_priv, GEN6_GDRST, GEN6_GRDOM_FULL);
1462 1463

	/* Spin waiting for the device to ack the reset request */
1464
	ret = wait_for((__raw_i915_read32(dev_priv, GEN6_GDRST) & GEN6_GRDOM_FULL) == 0, 500);
1465

1466
	intel_uncore_forcewake_reset(dev, true);
1467

1468 1469 1470
	return ret;
}

1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
static int wait_for_register(struct drm_i915_private *dev_priv,
			     const u32 reg,
			     const u32 mask,
			     const u32 value,
			     const unsigned long timeout_ms)
{
	return wait_for((I915_READ(reg) & mask) == value, timeout_ms);
}

static int gen8_do_reset(struct drm_device *dev)
{
	struct drm_i915_private *dev_priv = dev->dev_private;
	struct intel_engine_cs *engine;
	int i;

	for_each_ring(engine, dev_priv, i) {
		I915_WRITE(RING_RESET_CTL(engine->mmio_base),
			   _MASKED_BIT_ENABLE(RESET_CTL_REQUEST_RESET));

		if (wait_for_register(dev_priv,
				      RING_RESET_CTL(engine->mmio_base),
				      RESET_CTL_READY_TO_RESET,
				      RESET_CTL_READY_TO_RESET,
				      700)) {
			DRM_ERROR("%s: reset request timeout\n", engine->name);
			goto not_ready;
		}
	}

	return gen6_do_reset(dev);

not_ready:
	for_each_ring(engine, dev_priv, i)
		I915_WRITE(RING_RESET_CTL(engine->mmio_base),
			   _MASKED_BIT_DISABLE(RESET_CTL_REQUEST_RESET));

	return -EIO;
}

1510
static int (*intel_get_gpu_reset(struct drm_device *dev))(struct drm_device *)
1511
{
1512 1513 1514
	if (!i915.reset)
		return NULL;

1515 1516 1517
	if (INTEL_INFO(dev)->gen >= 8)
		return gen8_do_reset;
	else if (INTEL_INFO(dev)->gen >= 6)
1518
		return gen6_do_reset;
1519
	else if (IS_GEN5(dev))
1520
		return ironlake_do_reset;
1521
	else if (IS_G4X(dev))
1522
		return g4x_do_reset;
1523
	else if (IS_G33(dev))
1524
		return g33_do_reset;
1525
	else if (INTEL_INFO(dev)->gen >= 3)
1526
		return i915_do_reset;
1527
	else
1528 1529 1530 1531 1532 1533 1534 1535 1536
		return NULL;
}

int intel_gpu_reset(struct drm_device *dev)
{
	int (*reset)(struct drm_device *);

	reset = intel_get_gpu_reset(dev);
	if (reset == NULL)
1537
		return -ENODEV;
1538 1539 1540 1541 1542 1543 1544

	return reset(dev);
}

bool intel_has_gpu_reset(struct drm_device *dev)
{
	return intel_get_gpu_reset(dev) != NULL;
1545 1546 1547 1548 1549 1550 1551
}

void intel_uncore_check_errors(struct drm_device *dev)
{
	struct drm_i915_private *dev_priv = dev->dev_private;

	if (HAS_FPGA_DBG_UNCLAIMED(dev) &&
1552
	    (__raw_i915_read32(dev_priv, FPGA_DBG) & FPGA_DBG_RM_NOCLAIM)) {
1553
		DRM_ERROR("Unclaimed register before interrupt\n");
1554
		__raw_i915_write32(dev_priv, FPGA_DBG, FPGA_DBG_RM_NOCLAIM);
1555 1556
	}
}