intel_reset.c 38.6 KB
Newer Older
1 2 3 4 5 6 7
/*
 * SPDX-License-Identifier: MIT
 *
 * Copyright © 2008-2018 Intel Corporation
 */

#include <linux/sched/mm.h>
8
#include <linux/stop_machine.h>
9

10 11
#include "gem/i915_gem_context.h"

12 13
#include "i915_drv.h"
#include "i915_gpu_error.h"
14
#include "i915_irq.h"
15 16
#include "intel_engine_pm.h"
#include "intel_gt_pm.h"
17
#include "intel_reset.h"
18 19

#include "intel_guc.h"
20
#include "intel_overlay.h"
21

22 23
#define RESET_MAX_RETRIES 3

24 25 26
/* XXX How to handle concurrent GGTT updates using tiling registers? */
#define RESET_UNDER_STOP_MACHINE 0

27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
static void rmw_set(struct intel_uncore *uncore, i915_reg_t reg, u32 set)
{
	intel_uncore_rmw(uncore, reg, 0, set);
}

static void rmw_clear(struct intel_uncore *uncore, i915_reg_t reg, u32 clr)
{
	intel_uncore_rmw(uncore, reg, clr, 0);
}

static void rmw_set_fw(struct intel_uncore *uncore, i915_reg_t reg, u32 set)
{
	intel_uncore_rmw_fw(uncore, reg, 0, set);
}

static void rmw_clear_fw(struct intel_uncore *uncore, i915_reg_t reg, u32 clr)
{
	intel_uncore_rmw_fw(uncore, reg, clr, 0);
}

47 48 49 50 51
static void engine_skip_context(struct i915_request *rq)
{
	struct intel_engine_cs *engine = rq->engine;
	struct i915_gem_context *hung_ctx = rq->gem_context;

52
	lockdep_assert_held(&engine->active.lock);
53

54 55
	if (!i915_request_is_active(rq))
		return;
56

57
	list_for_each_entry_continue(rq, &engine->active.requests, sched.link)
58 59
		if (rq->gem_context == hung_ctx)
			i915_request_skip(rq, -EIO);
60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85
}

static void client_mark_guilty(struct drm_i915_file_private *file_priv,
			       const struct i915_gem_context *ctx)
{
	unsigned int score;
	unsigned long prev_hang;

	if (i915_gem_context_is_banned(ctx))
		score = I915_CLIENT_SCORE_CONTEXT_BAN;
	else
		score = 0;

	prev_hang = xchg(&file_priv->hang_timestamp, jiffies);
	if (time_before(jiffies, prev_hang + I915_CLIENT_FAST_HANG_JIFFIES))
		score += I915_CLIENT_SCORE_HANG_FAST;

	if (score) {
		atomic_add(score, &file_priv->ban_score);

		DRM_DEBUG_DRIVER("client %s: gained %u ban score, now %u\n",
				 ctx->name, score,
				 atomic_read(&file_priv->ban_score));
	}
}

86
static bool context_mark_guilty(struct i915_gem_context *ctx)
87
{
88 89 90
	unsigned long prev_hang;
	bool banned;
	int i;
91 92 93

	atomic_inc(&ctx->guilty_count);

94 95
	/* Cool contexts are too cool to be banned! (Used for reset testing.) */
	if (!i915_gem_context_is_bannable(ctx))
96
		return false;
97

98 99 100 101 102 103 104 105 106 107
	/* Record the timestamp for the last N hangs */
	prev_hang = ctx->hang_timestamp[0];
	for (i = 0; i < ARRAY_SIZE(ctx->hang_timestamp) - 1; i++)
		ctx->hang_timestamp[i] = ctx->hang_timestamp[i + 1];
	ctx->hang_timestamp[i] = jiffies;

	/* If we have hung N+1 times in rapid succession, we ban the context! */
	banned = !i915_gem_context_is_recoverable(ctx);
	if (time_before(jiffies, prev_hang + CONTEXT_FAST_HANG_JIFFIES))
		banned = true;
108
	if (banned) {
109 110
		DRM_DEBUG_DRIVER("context %s: guilty %d, banned\n",
				 ctx->name, atomic_read(&ctx->guilty_count));
111 112 113 114 115
		i915_gem_context_set_banned(ctx);
	}

	if (!IS_ERR_OR_NULL(ctx->file_priv))
		client_mark_guilty(ctx->file_priv, ctx);
116 117

	return banned;
118 119 120 121 122 123 124
}

static void context_mark_innocent(struct i915_gem_context *ctx)
{
	atomic_inc(&ctx->active_count);
}

125 126
void i915_reset_request(struct i915_request *rq, bool guilty)
{
127 128 129 130 131 132
	GEM_TRACE("%s rq=%llx:%lld, guilty? %s\n",
		  rq->engine->name,
		  rq->fence.context,
		  rq->fence.seqno,
		  yesno(guilty));

133
	lockdep_assert_held(&rq->engine->active.lock);
134 135 136 137 138 139 140 141 142 143 144 145
	GEM_BUG_ON(i915_request_completed(rq));

	if (guilty) {
		i915_request_skip(rq, -EIO);
		if (context_mark_guilty(rq->gem_context))
			engine_skip_context(rq);
	} else {
		dma_fence_set_error(&rq->fence, -EAGAIN);
		context_mark_innocent(rq->gem_context);
	}
}

146 147
static void gen3_stop_engine(struct intel_engine_cs *engine)
{
148
	struct intel_uncore *uncore = engine->uncore;
149 150
	const u32 base = engine->mmio_base;

151 152
	GEM_TRACE("%s\n", engine->name);

153
	if (intel_engine_stop_cs(engine))
154
		GEM_TRACE("%s: timed out on STOP_RING\n", engine->name);
155

156 157 158 159
	intel_uncore_write_fw(uncore,
			      RING_HEAD(base),
			      intel_uncore_read_fw(uncore, RING_TAIL(base)));
	intel_uncore_posting_read_fw(uncore, RING_HEAD(base)); /* paranoia */
160

161 162 163
	intel_uncore_write_fw(uncore, RING_HEAD(base), 0);
	intel_uncore_write_fw(uncore, RING_TAIL(base), 0);
	intel_uncore_posting_read_fw(uncore, RING_TAIL(base));
164 165

	/* The ring must be empty before it is disabled */
166
	intel_uncore_write_fw(uncore, RING_CTL(base), 0);
167 168

	/* Check acts as a post */
169
	if (intel_uncore_read_fw(uncore, RING_HEAD(base)))
170
		GEM_TRACE("%s: ring head [%x] not parked\n",
171 172
			  engine->name,
			  intel_uncore_read_fw(uncore, RING_HEAD(base)));
173 174 175
}

static void i915_stop_engines(struct drm_i915_private *i915,
176
			      intel_engine_mask_t engine_mask)
177 178
{
	struct intel_engine_cs *engine;
179
	intel_engine_mask_t tmp;
180 181 182 183

	if (INTEL_GEN(i915) < 3)
		return;

184
	for_each_engine_masked(engine, i915, engine_mask, tmp)
185 186 187 188 189 190 191 192 193 194 195 196
		gen3_stop_engine(engine);
}

static bool i915_in_reset(struct pci_dev *pdev)
{
	u8 gdrst;

	pci_read_config_byte(pdev, I915_GDRST, &gdrst);
	return gdrst & GRDOM_RESET_STATUS;
}

static int i915_do_reset(struct drm_i915_private *i915,
197
			 intel_engine_mask_t engine_mask,
198 199 200 201 202 203 204
			 unsigned int retry)
{
	struct pci_dev *pdev = i915->drm.pdev;
	int err;

	/* Assert reset for at least 20 usec, and wait for acknowledgement. */
	pci_write_config_byte(pdev, I915_GDRST, GRDOM_RESET_ENABLE);
205 206
	udelay(50);
	err = wait_for_atomic(i915_in_reset(pdev), 50);
207 208 209

	/* Clear the reset request. */
	pci_write_config_byte(pdev, I915_GDRST, 0);
210
	udelay(50);
211
	if (!err)
212
		err = wait_for_atomic(!i915_in_reset(pdev), 50);
213 214 215 216 217 218 219 220 221 222 223 224 225

	return err;
}

static bool g4x_reset_complete(struct pci_dev *pdev)
{
	u8 gdrst;

	pci_read_config_byte(pdev, I915_GDRST, &gdrst);
	return (gdrst & GRDOM_RESET_ENABLE) == 0;
}

static int g33_do_reset(struct drm_i915_private *i915,
226
			intel_engine_mask_t engine_mask,
227 228 229 230 231
			unsigned int retry)
{
	struct pci_dev *pdev = i915->drm.pdev;

	pci_write_config_byte(pdev, I915_GDRST, GRDOM_RESET_ENABLE);
232
	return wait_for_atomic(g4x_reset_complete(pdev), 50);
233 234
}

235
static int g4x_do_reset(struct drm_i915_private *i915,
236
			intel_engine_mask_t engine_mask,
237 238
			unsigned int retry)
{
239 240
	struct pci_dev *pdev = i915->drm.pdev;
	struct intel_uncore *uncore = &i915->uncore;
241 242 243
	int ret;

	/* WaVcpClkGateDisableForMediaReset:ctg,elk */
244 245
	rmw_set_fw(uncore, VDECCLK_GATE_D, VCP_UNIT_CLOCK_GATE_DISABLE);
	intel_uncore_posting_read_fw(uncore, VDECCLK_GATE_D);
246 247 248

	pci_write_config_byte(pdev, I915_GDRST,
			      GRDOM_MEDIA | GRDOM_RESET_ENABLE);
249
	ret =  wait_for_atomic(g4x_reset_complete(pdev), 50);
250 251 252 253 254 255 256
	if (ret) {
		DRM_DEBUG_DRIVER("Wait for media reset failed\n");
		goto out;
	}

	pci_write_config_byte(pdev, I915_GDRST,
			      GRDOM_RENDER | GRDOM_RESET_ENABLE);
257
	ret =  wait_for_atomic(g4x_reset_complete(pdev), 50);
258 259 260 261 262 263 264 265
	if (ret) {
		DRM_DEBUG_DRIVER("Wait for render reset failed\n");
		goto out;
	}

out:
	pci_write_config_byte(pdev, I915_GDRST, 0);

266 267
	rmw_clear_fw(uncore, VDECCLK_GATE_D, VCP_UNIT_CLOCK_GATE_DISABLE);
	intel_uncore_posting_read_fw(uncore, VDECCLK_GATE_D);
268 269 270 271

	return ret;
}

272
static int ironlake_do_reset(struct drm_i915_private *i915,
273
			     intel_engine_mask_t engine_mask,
274 275
			     unsigned int retry)
{
276
	struct intel_uncore *uncore = &i915->uncore;
277 278
	int ret;

279 280 281
	intel_uncore_write_fw(uncore, ILK_GDSR,
			      ILK_GRDOM_RENDER | ILK_GRDOM_RESET_ENABLE);
	ret = __intel_wait_for_register_fw(uncore, ILK_GDSR,
282 283 284
					   ILK_GRDOM_RESET_ENABLE, 0,
					   5000, 0,
					   NULL);
285 286 287 288 289
	if (ret) {
		DRM_DEBUG_DRIVER("Wait for render reset failed\n");
		goto out;
	}

290 291 292
	intel_uncore_write_fw(uncore, ILK_GDSR,
			      ILK_GRDOM_MEDIA | ILK_GRDOM_RESET_ENABLE);
	ret = __intel_wait_for_register_fw(uncore, ILK_GDSR,
293 294 295
					   ILK_GRDOM_RESET_ENABLE, 0,
					   5000, 0,
					   NULL);
296 297 298 299 300 301
	if (ret) {
		DRM_DEBUG_DRIVER("Wait for media reset failed\n");
		goto out;
	}

out:
302 303
	intel_uncore_write_fw(uncore, ILK_GDSR, 0);
	intel_uncore_posting_read_fw(uncore, ILK_GDSR);
304 305 306 307
	return ret;
}

/* Reset the hardware domains (GENX_GRDOM_*) specified by mask */
308
static int gen6_hw_domain_reset(struct drm_i915_private *i915,
309 310
				u32 hw_domain_mask)
{
311
	struct intel_uncore *uncore = &i915->uncore;
312 313 314 315 316 317 318
	int err;

	/*
	 * 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
	 */
319
	intel_uncore_write_fw(uncore, GEN6_GDRST, hw_domain_mask);
320 321

	/* Wait for the device to ack the reset requests */
322
	err = __intel_wait_for_register_fw(uncore,
323 324 325 326 327 328 329 330 331 332 333
					   GEN6_GDRST, hw_domain_mask, 0,
					   500, 0,
					   NULL);
	if (err)
		DRM_DEBUG_DRIVER("Wait for 0x%08x engines reset failed\n",
				 hw_domain_mask);

	return err;
}

static int gen6_reset_engines(struct drm_i915_private *i915,
334
			      intel_engine_mask_t engine_mask,
335 336 337
			      unsigned int retry)
{
	struct intel_engine_cs *engine;
338 339 340 341 342 343
	const u32 hw_engine_mask[] = {
		[RCS0]  = GEN6_GRDOM_RENDER,
		[BCS0]  = GEN6_GRDOM_BLT,
		[VCS0]  = GEN6_GRDOM_MEDIA,
		[VCS1]  = GEN8_GRDOM_MEDIA2,
		[VECS0] = GEN6_GRDOM_VECS,
344 345 346 347 348 349
	};
	u32 hw_mask;

	if (engine_mask == ALL_ENGINES) {
		hw_mask = GEN6_GRDOM_FULL;
	} else {
350
		intel_engine_mask_t tmp;
351 352

		hw_mask = 0;
353 354
		for_each_engine_masked(engine, i915, engine_mask, tmp) {
			GEM_BUG_ON(engine->id >= ARRAY_SIZE(hw_engine_mask));
355
			hw_mask |= hw_engine_mask[engine->id];
356
		}
357 358 359 360 361
	}

	return gen6_hw_domain_reset(i915, hw_mask);
}

362
static u32 gen11_lock_sfc(struct intel_engine_cs *engine)
363
{
364 365
	struct intel_uncore *uncore = engine->uncore;
	u8 vdbox_sfc_access = RUNTIME_INFO(engine->i915)->vdbox_sfc_access;
366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
	i915_reg_t sfc_forced_lock, sfc_forced_lock_ack;
	u32 sfc_forced_lock_bit, sfc_forced_lock_ack_bit;
	i915_reg_t sfc_usage;
	u32 sfc_usage_bit;
	u32 sfc_reset_bit;

	switch (engine->class) {
	case VIDEO_DECODE_CLASS:
		if ((BIT(engine->instance) & vdbox_sfc_access) == 0)
			return 0;

		sfc_forced_lock = GEN11_VCS_SFC_FORCED_LOCK(engine);
		sfc_forced_lock_bit = GEN11_VCS_SFC_FORCED_LOCK_BIT;

		sfc_forced_lock_ack = GEN11_VCS_SFC_LOCK_STATUS(engine);
		sfc_forced_lock_ack_bit  = GEN11_VCS_SFC_LOCK_ACK_BIT;

		sfc_usage = GEN11_VCS_SFC_LOCK_STATUS(engine);
		sfc_usage_bit = GEN11_VCS_SFC_USAGE_BIT;
		sfc_reset_bit = GEN11_VCS_SFC_RESET_BIT(engine->instance);
		break;

	case VIDEO_ENHANCEMENT_CLASS:
		sfc_forced_lock = GEN11_VECS_SFC_FORCED_LOCK(engine);
		sfc_forced_lock_bit = GEN11_VECS_SFC_FORCED_LOCK_BIT;

		sfc_forced_lock_ack = GEN11_VECS_SFC_LOCK_ACK(engine);
		sfc_forced_lock_ack_bit  = GEN11_VECS_SFC_LOCK_ACK_BIT;

		sfc_usage = GEN11_VECS_SFC_USAGE(engine);
		sfc_usage_bit = GEN11_VECS_SFC_USAGE_BIT;
		sfc_reset_bit = GEN11_VECS_SFC_RESET_BIT(engine->instance);
		break;

	default:
		return 0;
	}

	/*
	 * Tell the engine that a software reset is going to happen. The engine
	 * will then try to force lock the SFC (if currently locked, it will
	 * remain so until we tell the engine it is safe to unlock; if currently
	 * unlocked, it will ignore this and all new lock requests). If SFC
	 * ends up being locked to the engine we want to reset, we have to reset
	 * it as well (we will unlock it once the reset sequence is completed).
	 */
412
	rmw_set_fw(uncore, sfc_forced_lock, sfc_forced_lock_bit);
413

414
	if (__intel_wait_for_register_fw(uncore,
415 416 417 418 419 420 421 422
					 sfc_forced_lock_ack,
					 sfc_forced_lock_ack_bit,
					 sfc_forced_lock_ack_bit,
					 1000, 0, NULL)) {
		DRM_DEBUG_DRIVER("Wait for SFC forced lock ack failed\n");
		return 0;
	}

423
	if (intel_uncore_read_fw(uncore, sfc_usage) & sfc_usage_bit)
424 425 426 427 428
		return sfc_reset_bit;

	return 0;
}

429
static void gen11_unlock_sfc(struct intel_engine_cs *engine)
430
{
431 432
	struct intel_uncore *uncore = engine->uncore;
	u8 vdbox_sfc_access = RUNTIME_INFO(engine->i915)->vdbox_sfc_access;
433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453
	i915_reg_t sfc_forced_lock;
	u32 sfc_forced_lock_bit;

	switch (engine->class) {
	case VIDEO_DECODE_CLASS:
		if ((BIT(engine->instance) & vdbox_sfc_access) == 0)
			return;

		sfc_forced_lock = GEN11_VCS_SFC_FORCED_LOCK(engine);
		sfc_forced_lock_bit = GEN11_VCS_SFC_FORCED_LOCK_BIT;
		break;

	case VIDEO_ENHANCEMENT_CLASS:
		sfc_forced_lock = GEN11_VECS_SFC_FORCED_LOCK(engine);
		sfc_forced_lock_bit = GEN11_VECS_SFC_FORCED_LOCK_BIT;
		break;

	default:
		return;
	}

454
	rmw_clear_fw(uncore, sfc_forced_lock, sfc_forced_lock_bit);
455 456 457
}

static int gen11_reset_engines(struct drm_i915_private *i915,
458
			       intel_engine_mask_t engine_mask,
459 460
			       unsigned int retry)
{
461 462 463 464 465 466 467 468 469
	const u32 hw_engine_mask[] = {
		[RCS0]  = GEN11_GRDOM_RENDER,
		[BCS0]  = GEN11_GRDOM_BLT,
		[VCS0]  = GEN11_GRDOM_MEDIA,
		[VCS1]  = GEN11_GRDOM_MEDIA2,
		[VCS2]  = GEN11_GRDOM_MEDIA3,
		[VCS3]  = GEN11_GRDOM_MEDIA4,
		[VECS0] = GEN11_GRDOM_VECS,
		[VECS1] = GEN11_GRDOM_VECS2,
470 471
	};
	struct intel_engine_cs *engine;
472
	intel_engine_mask_t tmp;
473 474 475 476 477 478 479 480
	u32 hw_mask;
	int ret;

	if (engine_mask == ALL_ENGINES) {
		hw_mask = GEN11_GRDOM_FULL;
	} else {
		hw_mask = 0;
		for_each_engine_masked(engine, i915, engine_mask, tmp) {
481
			GEM_BUG_ON(engine->id >= ARRAY_SIZE(hw_engine_mask));
482
			hw_mask |= hw_engine_mask[engine->id];
483
			hw_mask |= gen11_lock_sfc(engine);
484 485 486 487 488 489 490
		}
	}

	ret = gen6_hw_domain_reset(i915, hw_mask);

	if (engine_mask != ALL_ENGINES)
		for_each_engine_masked(engine, i915, engine_mask, tmp)
491
			gen11_unlock_sfc(engine);
492 493 494 495 496 497

	return ret;
}

static int gen8_engine_reset_prepare(struct intel_engine_cs *engine)
{
498
	struct intel_uncore *uncore = engine->uncore;
499 500
	const i915_reg_t reg = RING_RESET_CTL(engine->mmio_base);
	u32 request, mask, ack;
501 502
	int ret;

503
	ack = intel_uncore_read_fw(uncore, reg);
504 505 506 507 508 509 510 511 512 513 514
	if (ack & RESET_CTL_CAT_ERROR) {
		/*
		 * For catastrophic errors, ready-for-reset sequence
		 * needs to be bypassed: HAS#396813
		 */
		request = RESET_CTL_CAT_ERROR;
		mask = RESET_CTL_CAT_ERROR;

		/* Catastrophic errors need to be cleared by HW */
		ack = 0;
	} else if (!(ack & RESET_CTL_READY_TO_RESET)) {
515 516 517 518 519 520
		request = RESET_CTL_REQUEST_RESET;
		mask = RESET_CTL_READY_TO_RESET;
		ack = RESET_CTL_READY_TO_RESET;
	} else {
		return 0;
	}
521

522 523 524
	intel_uncore_write_fw(uncore, reg, _MASKED_BIT_ENABLE(request));
	ret = __intel_wait_for_register_fw(uncore, reg, mask, ack,
					   700, 0, NULL);
525
	if (ret)
526 527 528
		DRM_ERROR("%s reset request timed out: {request: %08x, RESET_CTL: %08x}\n",
			  engine->name, request,
			  intel_uncore_read_fw(uncore, reg));
529 530 531 532 533 534

	return ret;
}

static void gen8_engine_reset_cancel(struct intel_engine_cs *engine)
{
535 536 537
	intel_uncore_write_fw(engine->uncore,
			      RING_RESET_CTL(engine->mmio_base),
			      _MASKED_BIT_DISABLE(RESET_CTL_REQUEST_RESET));
538 539 540
}

static int gen8_reset_engines(struct drm_i915_private *i915,
541
			      intel_engine_mask_t engine_mask,
542 543 544 545
			      unsigned int retry)
{
	struct intel_engine_cs *engine;
	const bool reset_non_ready = retry >= 1;
546
	intel_engine_mask_t tmp;
547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581
	int ret;

	for_each_engine_masked(engine, i915, engine_mask, tmp) {
		ret = gen8_engine_reset_prepare(engine);
		if (ret && !reset_non_ready)
			goto skip_reset;

		/*
		 * If this is not the first failed attempt to prepare,
		 * we decide to proceed anyway.
		 *
		 * By doing so we risk context corruption and with
		 * some gens (kbl), possible system hang if reset
		 * happens during active bb execution.
		 *
		 * We rather take context corruption instead of
		 * failed reset with a wedged driver/gpu. And
		 * active bb execution case should be covered by
		 * i915_stop_engines we have before the reset.
		 */
	}

	if (INTEL_GEN(i915) >= 11)
		ret = gen11_reset_engines(i915, engine_mask, retry);
	else
		ret = gen6_reset_engines(i915, engine_mask, retry);

skip_reset:
	for_each_engine_masked(engine, i915, engine_mask, tmp)
		gen8_engine_reset_cancel(engine);

	return ret;
}

typedef int (*reset_func)(struct drm_i915_private *,
582
			  intel_engine_mask_t engine_mask,
583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602
			  unsigned int retry);

static reset_func intel_get_gpu_reset(struct drm_i915_private *i915)
{
	if (INTEL_GEN(i915) >= 8)
		return gen8_reset_engines;
	else if (INTEL_GEN(i915) >= 6)
		return gen6_reset_engines;
	else if (INTEL_GEN(i915) >= 5)
		return ironlake_do_reset;
	else if (IS_G4X(i915))
		return g4x_do_reset;
	else if (IS_G33(i915) || IS_PINEVIEW(i915))
		return g33_do_reset;
	else if (INTEL_GEN(i915) >= 3)
		return i915_do_reset;
	else
		return NULL;
}

603 604
int intel_gpu_reset(struct drm_i915_private *i915,
		    intel_engine_mask_t engine_mask)
605
{
606 607 608
	const int retries = engine_mask == ALL_ENGINES ? RESET_MAX_RETRIES : 1;
	reset_func reset;
	int ret = -ETIMEDOUT;
609 610
	int retry;

611 612 613
	reset = intel_get_gpu_reset(i915);
	if (!reset)
		return -ENODEV;
614 615 616 617 618

	/*
	 * If the power well sleeps during the reset, the reset
	 * request may be dropped and never completes (causing -EIO).
	 */
619
	intel_uncore_forcewake_get(&i915->uncore, FORCEWAKE_ALL);
620
	for (retry = 0; ret == -ETIMEDOUT && retry < retries; retry++) {
621 622 623 624 625 626 627 628 629 630 631 632 633 634
		/*
		 * We stop engines, otherwise we might get failed reset and a
		 * dead gpu (on elk). Also as modern gpu as kbl can suffer
		 * from system hang if batchbuffer is progressing when
		 * the reset is issued, regardless of READY_TO_RESET ack.
		 * Thus assume it is best to stop engines on all gens
		 * where we have a gpu reset.
		 *
		 * WaKBLVECSSemaphoreWaitPoll:kbl (on ALL_ENGINES)
		 *
		 * WaMediaResetMainRingCleanup:ctg,elk (presumably)
		 *
		 * FIXME: Wa for more modern gens needs to be validated
		 */
635 636
		if (retry)
			i915_stop_engines(i915, engine_mask);
637

638 639 640 641
		GEM_TRACE("engine_mask=%x\n", engine_mask);
		preempt_disable();
		ret = reset(i915, engine_mask, retry);
		preempt_enable();
642
	}
643
	intel_uncore_forcewake_put(&i915->uncore, FORCEWAKE_ALL);
644 645 646 647 648 649

	return ret;
}

bool intel_has_gpu_reset(struct drm_i915_private *i915)
{
650 651 652
	if (!i915_modparams.reset)
		return NULL;

653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668
	return intel_get_gpu_reset(i915);
}

bool intel_has_reset_engine(struct drm_i915_private *i915)
{
	return INTEL_INFO(i915)->has_reset_engine && i915_modparams.reset >= 2;
}

int intel_reset_guc(struct drm_i915_private *i915)
{
	u32 guc_domain =
		INTEL_GEN(i915) >= 11 ? GEN11_GRDOM_GUC : GEN9_GRDOM_GUC;
	int ret;

	GEM_BUG_ON(!HAS_GUC(i915));

669
	intel_uncore_forcewake_get(&i915->uncore, FORCEWAKE_ALL);
670
	ret = gen6_hw_domain_reset(i915, guc_domain);
671
	intel_uncore_forcewake_put(&i915->uncore, FORCEWAKE_ALL);
672 673 674 675 676 677 678 679

	return ret;
}

/*
 * Ensure irq handler finishes, and not run again.
 * Also return the active request so that we only search for it once.
 */
680
static void reset_prepare_engine(struct intel_engine_cs *engine)
681 682 683 684 685 686 687 688
{
	/*
	 * During the reset sequence, we must prevent the engine from
	 * entering RC6. As the context state is undefined until we restart
	 * the engine, if it does enter RC6 during the reset, the state
	 * written to the powercontext is undefined and so we may lose
	 * GPU state upon resume, i.e. fail to restart after a reset.
	 */
689
	intel_engine_pm_get(engine);
690
	intel_uncore_forcewake_get(engine->uncore, FORCEWAKE_ALL);
691
	engine->reset.prepare(engine);
692 693
}

694 695 696 697
static void revoke_mmaps(struct drm_i915_private *i915)
{
	int i;

698
	for (i = 0; i < i915->ggtt.num_fences; i++) {
699 700 701 702
		struct drm_vma_offset_node *node;
		struct i915_vma *vma;
		u64 vma_offset;

703
		vma = READ_ONCE(i915->ggtt.fence_regs[i].vma);
704 705 706 707 708 709
		if (!vma)
			continue;

		if (!i915_vma_has_userfault(vma))
			continue;

710
		GEM_BUG_ON(vma->fence != &i915->ggtt.fence_regs[i]);
711 712 713 714 715 716 717 718 719
		node = &vma->obj->base.vma_node;
		vma_offset = vma->ggtt_view.partial.offset << PAGE_SHIFT;
		unmap_mapping_range(i915->drm.anon_inode->i_mapping,
				    drm_vma_node_offset_addr(node) + vma_offset,
				    vma->size,
				    1);
	}
}

720
static void reset_prepare(struct drm_i915_private *i915)
721 722 723 724
{
	struct intel_engine_cs *engine;
	enum intel_engine_id id;

725
	intel_gt_pm_get(i915);
726 727
	for_each_engine(engine, i915, id)
		reset_prepare_engine(engine);
728

729
	intel_uc_reset_prepare(i915);
730 731 732 733
}

static void gt_revoke(struct drm_i915_private *i915)
{
734
	revoke_mmaps(i915);
735 736
}

737 738
static int gt_reset(struct drm_i915_private *i915,
		    intel_engine_mask_t stalled_mask)
739
{
740 741 742 743
	struct intel_engine_cs *engine;
	enum intel_engine_id id;
	int err;

744
	/*
745 746
	 * Everything depends on having the GTT running, so we need to start
	 * there.
747
	 */
748 749 750
	err = i915_ggtt_enable_hw(i915);
	if (err)
		return err;
751

752
	for_each_engine(engine, i915, id)
753
		intel_engine_reset(engine, stalled_mask & engine->mask);
754

755
	i915_gem_restore_fences(i915);
756

757
	return err;
758 759
}

760
static void reset_finish_engine(struct intel_engine_cs *engine)
761
{
762
	engine->reset.finish(engine);
763
	intel_engine_pm_put(engine);
764
	intel_uncore_forcewake_put(engine->uncore, FORCEWAKE_ALL);
765 766 767 768 769 770 771
}

static void reset_finish(struct drm_i915_private *i915)
{
	struct intel_engine_cs *engine;
	enum intel_engine_id id;

772
	for_each_engine(engine, i915, id) {
773
		reset_finish_engine(engine);
774 775
		intel_engine_signal_breadcrumbs(engine);
	}
776
	intel_gt_pm_put(i915);
777 778 779 780
}

static void nop_submit_request(struct i915_request *request)
{
781
	struct intel_engine_cs *engine = request->engine;
782 783 784
	unsigned long flags;

	GEM_TRACE("%s fence %llx:%lld -> -EIO\n",
785
		  engine->name, request->fence.context, request->fence.seqno);
786 787
	dma_fence_set_error(&request->fence, -EIO);

788
	spin_lock_irqsave(&engine->active.lock, flags);
789
	__i915_request_submit(request);
790
	i915_request_mark_complete(request);
791
	spin_unlock_irqrestore(&engine->active.lock, flags);
792 793

	intel_engine_queue_breadcrumbs(engine);
794 795
}

796
static void __i915_gem_set_wedged(struct drm_i915_private *i915)
797 798 799 800 801
{
	struct i915_gpu_error *error = &i915->gpu_error;
	struct intel_engine_cs *engine;
	enum intel_engine_id id;

802
	if (test_bit(I915_WEDGED, &error->flags))
803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818
		return;

	if (GEM_SHOW_DEBUG() && !intel_engines_are_idle(i915)) {
		struct drm_printer p = drm_debug_printer(__func__);

		for_each_engine(engine, i915, id)
			intel_engine_dump(engine, &p, "%s\n", engine->name);
	}

	GEM_TRACE("start\n");

	/*
	 * First, stop submission to hw, but do not yet complete requests by
	 * rolling the global seqno forward (since this would complete requests
	 * for which we haven't set the fence error to EIO yet).
	 */
819
	reset_prepare(i915);
820

821
	/* Even if the GPU reset fails, it should still stop the engines */
822
	if (!INTEL_INFO(i915)->gpu_reset_clobbers_display)
823 824 825 826 827 828 829 830 831 832 833 834 835
		intel_gpu_reset(i915, ALL_ENGINES);

	for_each_engine(engine, i915, id) {
		engine->submit_request = nop_submit_request;
		engine->schedule = NULL;
	}
	i915->caps.scheduler = 0;

	/*
	 * Make sure no request can slip through without getting completed by
	 * either this call here to intel_engine_write_global_seqno, or the one
	 * in nop_submit_request.
	 */
C
Chris Wilson 已提交
836
	synchronize_rcu_expedited();
837
	set_bit(I915_WEDGED, &error->flags);
838 839 840 841 842

	/* Mark all executing requests as skipped */
	for_each_engine(engine, i915, id)
		engine->cancel_requests(engine);

843
	reset_finish(i915);
844 845

	GEM_TRACE("end\n");
846
}
847

848 849 850
void i915_gem_set_wedged(struct drm_i915_private *i915)
{
	struct i915_gpu_error *error = &i915->gpu_error;
851
	intel_wakeref_t wakeref;
852 853

	mutex_lock(&error->wedge_mutex);
854
	with_intel_runtime_pm(&i915->runtime_pm, wakeref)
855
		__i915_gem_set_wedged(i915);
856
	mutex_unlock(&error->wedge_mutex);
857 858
}

859
static bool __i915_gem_unset_wedged(struct drm_i915_private *i915)
860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881
{
	struct i915_gpu_error *error = &i915->gpu_error;
	struct i915_timeline *tl;

	if (!test_bit(I915_WEDGED, &error->flags))
		return true;

	if (!i915->gt.scratch) /* Never full initialised, recovery impossible */
		return false;

	GEM_TRACE("start\n");

	/*
	 * Before unwedging, make sure that all pending operations
	 * are flushed and errored out - we may have requests waiting upon
	 * third party fences. We marked all inflight requests as EIO, and
	 * every execbuf since returned EIO, for consistency we want all
	 * the currently pending requests to also be marked as EIO, which
	 * is done inside our nop_submit_request - and so we must wait.
	 *
	 * No more can be submitted until we reset the wedged bit.
	 */
882
	mutex_lock(&i915->gt.timelines.mutex);
C
Chris Wilson 已提交
883
	list_for_each_entry(tl, &i915->gt.timelines.active_list, link) {
884 885
		struct i915_request *rq;

886
		rq = i915_active_request_get_unlocked(&tl->last_request);
887 888 889 890
		if (!rq)
			continue;

		/*
891 892 893 894 895
		 * All internal dependencies (i915_requests) will have
		 * been flushed by the set-wedge, but we may be stuck waiting
		 * for external fences. These should all be capped to 10s
		 * (I915_FENCE_TIMEOUT) so this wait should not be unbounded
		 * in the worst case.
896
		 */
897
		dma_fence_default_wait(&rq->fence, false, MAX_SCHEDULE_TIMEOUT);
898
		i915_request_put(rq);
899
	}
900
	mutex_unlock(&i915->gt.timelines.mutex);
901

902
	intel_gt_sanitize(i915, false);
903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918

	/*
	 * Undo nop_submit_request. We prevent all new i915 requests from
	 * being queued (by disallowing execbuf whilst wedged) so having
	 * waited for all active requests above, we know the system is idle
	 * and do not have to worry about a thread being inside
	 * engine->submit_request() as we swap over. So unlike installing
	 * the nop_submit_request on reset, we can do this from normal
	 * context and do not require stop_machine().
	 */
	intel_engines_reset_default_submission(i915);

	GEM_TRACE("end\n");

	smp_mb__before_atomic(); /* complete takeover before enabling execbuf */
	clear_bit(I915_WEDGED, &i915->gpu_error.flags);
919 920

	return true;
921 922
}

923 924 925 926 927 928 929 930 931 932 933 934
bool i915_gem_unset_wedged(struct drm_i915_private *i915)
{
	struct i915_gpu_error *error = &i915->gpu_error;
	bool result;

	mutex_lock(&error->wedge_mutex);
	result = __i915_gem_unset_wedged(i915);
	mutex_unlock(&error->wedge_mutex);

	return result;
}

935 936
static int do_reset(struct drm_i915_private *i915,
		    intel_engine_mask_t stalled_mask)
937 938 939
{
	int err, i;

940 941
	gt_revoke(i915);

942
	err = intel_gpu_reset(i915, ALL_ENGINES);
943
	for (i = 0; err && i < RESET_MAX_RETRIES; i++) {
944 945
		msleep(10 * (i + 1));
		err = intel_gpu_reset(i915, ALL_ENGINES);
946
	}
947 948
	if (err)
		return err;
949

950
	return gt_reset(i915, stalled_mask);
951 952
}

953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970
/**
 * i915_reset - reset chip after a hang
 * @i915: #drm_i915_private to reset
 * @stalled_mask: mask of the stalled engines with the guilty requests
 * @reason: user error message for why we are resetting
 *
 * Reset the chip.  Useful if a hang is detected. Marks the device as wedged
 * on failure.
 *
 * Procedure is fairly simple:
 *   - reset the chip using the reset reg
 *   - re-init context state
 *   - re-init hardware status page
 *   - re-init ring buffer
 *   - re-init interrupt state
 *   - re-init display
 */
void i915_reset(struct drm_i915_private *i915,
971
		intel_engine_mask_t stalled_mask,
972 973 974 975 976 977 978 979 980
		const char *reason)
{
	struct i915_gpu_error *error = &i915->gpu_error;
	int ret;

	GEM_TRACE("flags=%lx\n", error->flags);

	might_sleep();
	GEM_BUG_ON(!test_bit(I915_RESET_BACKOFF, &error->flags));
981
	mutex_lock(&error->wedge_mutex);
982 983

	/* Clear any previous failed attempts at recovery. Time to try again. */
984
	if (!__i915_gem_unset_wedged(i915))
985
		goto unlock;
986 987 988 989 990

	if (reason)
		dev_notice(i915->drm.dev, "Resetting chip for %s\n", reason);
	error->reset_count++;

991
	reset_prepare(i915);
992 993 994 995 996 997 998 999 1000

	if (!intel_has_gpu_reset(i915)) {
		if (i915_modparams.reset)
			dev_err(i915->drm.dev, "GPU reset not supported\n");
		else
			DRM_DEBUG_DRIVER("GPU reset disabled\n");
		goto error;
	}

1001 1002 1003
	if (INTEL_INFO(i915)->gpu_reset_clobbers_display)
		intel_runtime_pm_disable_interrupts(i915);

1004
	if (do_reset(i915, stalled_mask)) {
1005 1006 1007 1008
		dev_err(i915->drm.dev, "Failed to reset chip\n");
		goto taint;
	}

1009 1010 1011
	if (INTEL_INFO(i915)->gpu_reset_clobbers_display)
		intel_runtime_pm_enable_interrupts(i915);

1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032
	intel_overlay_reset(i915);

	/*
	 * Next we need to restore the context, but we don't use those
	 * yet either...
	 *
	 * Ring buffer needs to be re-initialized in the KMS case, or if X
	 * was running at the time of the reset (i.e. we weren't VT
	 * switched away).
	 */
	ret = i915_gem_init_hw(i915);
	if (ret) {
		DRM_ERROR("Failed to initialise HW following reset (%d)\n",
			  ret);
		goto error;
	}

	i915_queue_hangcheck(i915);

finish:
	reset_finish(i915);
1033
unlock:
1034
	mutex_unlock(&error->wedge_mutex);
1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049
	return;

taint:
	/*
	 * History tells us that if we cannot reset the GPU now, we
	 * never will. This then impacts everything that is run
	 * subsequently. On failing the reset, we mark the driver
	 * as wedged, preventing further execution on the GPU.
	 * We also want to go one step further and add a taint to the
	 * kernel so that any subsequent faults can be traced back to
	 * this failure. This is important for CI, where if the
	 * GPU/driver fails we would like to reboot and restart testing
	 * rather than continue on into oblivion. For everyone else,
	 * the system should still plod along, but they have been warned!
	 */
1050
	add_taint_for_CI(TAINT_WARN);
1051
error:
1052
	__i915_gem_set_wedged(i915);
1053 1054 1055 1056 1057 1058
	goto finish;
}

static inline int intel_gt_reset_engine(struct drm_i915_private *i915,
					struct intel_engine_cs *engine)
{
1059
	return intel_gpu_reset(i915, engine->mask);
1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082
}

/**
 * i915_reset_engine - reset GPU engine to recover from a hang
 * @engine: engine to reset
 * @msg: reason for GPU reset; or NULL for no dev_notice()
 *
 * Reset a specific GPU engine. Useful if a hang is detected.
 * Returns zero on successful reset or otherwise an error code.
 *
 * Procedure is:
 *  - identifies the request that caused the hang and it is dropped
 *  - reset engine (which will force the engine to idle)
 *  - re-init/configure engine
 */
int i915_reset_engine(struct intel_engine_cs *engine, const char *msg)
{
	struct i915_gpu_error *error = &engine->i915->gpu_error;
	int ret;

	GEM_TRACE("%s flags=%lx\n", engine->name, error->flags);
	GEM_BUG_ON(!test_bit(I915_RESET_ENGINE + engine->id, &error->flags));

1083 1084 1085
	if (!intel_wakeref_active(&engine->wakeref))
		return 0;

1086
	reset_prepare_engine(engine);
1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109

	if (msg)
		dev_notice(engine->i915->drm.dev,
			   "Resetting %s for %s\n", engine->name, msg);
	error->reset_engine_count[engine->id]++;

	if (!engine->i915->guc.execbuf_client)
		ret = intel_gt_reset_engine(engine->i915, engine);
	else
		ret = intel_guc_reset_engine(&engine->i915->guc, engine);
	if (ret) {
		/* If we fail here, we expect to fallback to a global reset */
		DRM_DEBUG_DRIVER("%sFailed to reset %s, ret=%d\n",
				 engine->i915->guc.execbuf_client ? "GuC " : "",
				 engine->name, ret);
		goto out;
	}

	/*
	 * The request that caused the hang is stuck on elsp, we know the
	 * active request and can drop it, adjust head to skip the offending
	 * request to resume executing remaining requests in the queue.
	 */
1110
	intel_engine_reset(engine, true);
1111 1112 1113 1114 1115 1116

	/*
	 * The engine and its registers (and workarounds in case of render)
	 * have been reset to their default values. Follow the init_ring
	 * process to program RING_MODE, HWSP and re-enable submission.
	 */
1117
	ret = engine->resume(engine);
1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146
	if (ret)
		goto out;

out:
	intel_engine_cancel_stop_cs(engine);
	reset_finish_engine(engine);
	return ret;
}

static void i915_reset_device(struct drm_i915_private *i915,
			      u32 engine_mask,
			      const char *reason)
{
	struct i915_gpu_error *error = &i915->gpu_error;
	struct kobject *kobj = &i915->drm.primary->kdev->kobj;
	char *error_event[] = { I915_ERROR_UEVENT "=1", NULL };
	char *reset_event[] = { I915_RESET_UEVENT "=1", NULL };
	char *reset_done_event[] = { I915_ERROR_UEVENT "=0", NULL };
	struct i915_wedge_me w;

	kobject_uevent_env(kobj, KOBJ_CHANGE, error_event);

	DRM_DEBUG_DRIVER("resetting chip\n");
	kobject_uevent_env(kobj, KOBJ_CHANGE, reset_event);

	/* Use a watchdog to ensure that our reset completes */
	i915_wedge_on_timeout(&w, i915, 5 * HZ) {
		intel_prepare_reset(i915);

1147
		/* Flush everyone using a resource about to be clobbered */
1148
		synchronize_srcu_expedited(&error->reset_backoff_srcu);
1149

1150
		i915_reset(i915, engine_mask, reason);
1151 1152 1153 1154 1155 1156 1157 1158

		intel_finish_reset(i915);
	}

	if (!test_bit(I915_WEDGED, &error->flags))
		kobject_uevent_env(kobj, KOBJ_CHANGE, reset_done_event);
}

1159
static void clear_register(struct intel_uncore *uncore, i915_reg_t reg)
1160
{
1161
	intel_uncore_rmw(uncore, reg, 0, 0);
1162 1163
}

1164 1165 1166 1167 1168 1169
static void gen8_clear_engine_error_register(struct intel_engine_cs *engine)
{
	GEN6_RING_FAULT_REG_RMW(engine, RING_FAULT_VALID, 0);
	GEN6_RING_FAULT_REG_POSTING_READ(engine);
}

1170 1171
static void clear_error_registers(struct drm_i915_private *i915,
				  intel_engine_mask_t engine_mask)
1172
{
1173
	struct intel_uncore *uncore = &i915->uncore;
1174 1175
	u32 eir;

1176 1177
	if (!IS_GEN(i915, 2))
		clear_register(uncore, PGTBL_ER);
1178

1179 1180
	if (INTEL_GEN(i915) < 4)
		clear_register(uncore, IPEIR(RENDER_RING_BASE));
1181
	else
1182
		clear_register(uncore, IPEIR_I965);
1183

1184 1185
	clear_register(uncore, EIR);
	eir = intel_uncore_read(uncore, EIR);
1186 1187 1188 1189 1190 1191
	if (eir) {
		/*
		 * some errors might have become stuck,
		 * mask them.
		 */
		DRM_DEBUG_DRIVER("EIR stuck: 0x%08x, masking\n", eir);
1192
		rmw_set(uncore, EMR, eir);
1193 1194
		intel_uncore_write(uncore, GEN2_IIR,
				   I915_MASTER_ERROR_INTERRUPT);
1195 1196
	}

1197 1198 1199 1200
	if (INTEL_GEN(i915) >= 8) {
		rmw_clear(uncore, GEN8_RING_FAULT_REG, RING_FAULT_VALID);
		intel_uncore_posting_read(uncore, GEN8_RING_FAULT_REG);
	} else if (INTEL_GEN(i915) >= 6) {
1201 1202 1203
		struct intel_engine_cs *engine;
		enum intel_engine_id id;

1204 1205
		for_each_engine_masked(engine, i915, engine_mask, id)
			gen8_clear_engine_error_register(engine);
1206 1207 1208
	}
}

1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271
static void gen6_check_faults(struct drm_i915_private *dev_priv)
{
	struct intel_engine_cs *engine;
	enum intel_engine_id id;
	u32 fault;

	for_each_engine(engine, dev_priv, id) {
		fault = GEN6_RING_FAULT_REG_READ(engine);
		if (fault & RING_FAULT_VALID) {
			DRM_DEBUG_DRIVER("Unexpected fault\n"
					 "\tAddr: 0x%08lx\n"
					 "\tAddress space: %s\n"
					 "\tSource ID: %d\n"
					 "\tType: %d\n",
					 fault & PAGE_MASK,
					 fault & RING_FAULT_GTTSEL_MASK ? "GGTT" : "PPGTT",
					 RING_FAULT_SRCID(fault),
					 RING_FAULT_FAULT_TYPE(fault));
		}
	}
}

static void gen8_check_faults(struct drm_i915_private *dev_priv)
{
	u32 fault = I915_READ(GEN8_RING_FAULT_REG);

	if (fault & RING_FAULT_VALID) {
		u32 fault_data0, fault_data1;
		u64 fault_addr;

		fault_data0 = I915_READ(GEN8_FAULT_TLB_DATA0);
		fault_data1 = I915_READ(GEN8_FAULT_TLB_DATA1);
		fault_addr = ((u64)(fault_data1 & FAULT_VA_HIGH_BITS) << 44) |
			     ((u64)fault_data0 << 12);

		DRM_DEBUG_DRIVER("Unexpected fault\n"
				 "\tAddr: 0x%08x_%08x\n"
				 "\tAddress space: %s\n"
				 "\tEngine ID: %d\n"
				 "\tSource ID: %d\n"
				 "\tType: %d\n",
				 upper_32_bits(fault_addr),
				 lower_32_bits(fault_addr),
				 fault_data1 & FAULT_GTT_SEL ? "GGTT" : "PPGTT",
				 GEN8_RING_FAULT_ENGINE_ID(fault),
				 RING_FAULT_SRCID(fault),
				 RING_FAULT_FAULT_TYPE(fault));
	}
}

void i915_check_and_clear_faults(struct drm_i915_private *i915)
{
	/* From GEN8 onwards we only have one 'All Engine Fault Register' */
	if (INTEL_GEN(i915) >= 8)
		gen8_check_faults(i915);
	else if (INTEL_GEN(i915) >= 6)
		gen6_check_faults(i915);
	else
		return;

	clear_error_registers(i915, ALL_ENGINES);
}

1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285
/**
 * i915_handle_error - handle a gpu error
 * @i915: i915 device private
 * @engine_mask: mask representing engines that are hung
 * @flags: control flags
 * @fmt: Error message format string
 *
 * Do some basic checking of register state at error time and
 * dump it to the syslog.  Also call i915_capture_error_state() to make
 * sure we get a record and make it available in debugfs.  Fire a uevent
 * so userspace knows something bad happened (should trigger collection
 * of a ring dump etc.).
 */
void i915_handle_error(struct drm_i915_private *i915,
1286
		       intel_engine_mask_t engine_mask,
1287 1288 1289
		       unsigned long flags,
		       const char *fmt, ...)
{
1290
	struct i915_gpu_error *error = &i915->gpu_error;
1291 1292
	struct intel_engine_cs *engine;
	intel_wakeref_t wakeref;
1293
	intel_engine_mask_t tmp;
1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313
	char error_msg[80];
	char *msg = NULL;

	if (fmt) {
		va_list args;

		va_start(args, fmt);
		vscnprintf(error_msg, sizeof(error_msg), fmt, args);
		va_end(args);

		msg = error_msg;
	}

	/*
	 * In most cases it's guaranteed that we get here with an RPM
	 * reference held, for example because there is a pending GPU
	 * request that won't finish until the reset is done. This
	 * isn't the case at least when we get here by doing a
	 * simulated reset via debugfs, so get an RPM reference.
	 */
1314
	wakeref = intel_runtime_pm_get(&i915->runtime_pm);
1315

1316
	engine_mask &= INTEL_INFO(i915)->engine_mask;
1317 1318 1319

	if (flags & I915_ERROR_CAPTURE) {
		i915_capture_error_state(i915, engine_mask, msg);
1320
		clear_error_registers(i915, engine_mask);
1321 1322 1323 1324 1325 1326
	}

	/*
	 * Try engine reset when available. We fall back to full reset if
	 * single reset fails.
	 */
1327
	if (intel_has_reset_engine(i915) && !__i915_wedged(error)) {
1328 1329 1330
		for_each_engine_masked(engine, i915, engine_mask, tmp) {
			BUILD_BUG_ON(I915_RESET_MODESET >= I915_RESET_ENGINE);
			if (test_and_set_bit(I915_RESET_ENGINE + engine->id,
1331
					     &error->flags))
1332 1333 1334
				continue;

			if (i915_reset_engine(engine, msg) == 0)
1335
				engine_mask &= ~engine->mask;
1336 1337

			clear_bit(I915_RESET_ENGINE + engine->id,
1338 1339
				  &error->flags);
			wake_up_bit(&error->flags,
1340 1341 1342 1343 1344 1345 1346 1347
				    I915_RESET_ENGINE + engine->id);
		}
	}

	if (!engine_mask)
		goto out;

	/* Full reset needs the mutex, stop any other user trying to do so. */
1348 1349 1350
	if (test_and_set_bit(I915_RESET_BACKOFF, &error->flags)) {
		wait_event(error->reset_queue,
			   !test_bit(I915_RESET_BACKOFF, &error->flags));
1351
		goto out; /* piggy-back on the other reset */
1352 1353
	}

1354 1355 1356
	/* Make sure i915_reset_trylock() sees the I915_RESET_BACKOFF */
	synchronize_rcu_expedited();

1357 1358 1359
	/* Prevent any other reset-engine attempt. */
	for_each_engine(engine, i915, tmp) {
		while (test_and_set_bit(I915_RESET_ENGINE + engine->id,
1360 1361
					&error->flags))
			wait_on_bit(&error->flags,
1362 1363 1364 1365 1366 1367 1368 1369
				    I915_RESET_ENGINE + engine->id,
				    TASK_UNINTERRUPTIBLE);
	}

	i915_reset_device(i915, engine_mask, msg);

	for_each_engine(engine, i915, tmp) {
		clear_bit(I915_RESET_ENGINE + engine->id,
1370
			  &error->flags);
1371 1372
	}

1373 1374
	clear_bit(I915_RESET_BACKOFF, &error->flags);
	wake_up_all(&error->reset_queue);
1375 1376

out:
1377
	intel_runtime_pm_put(&i915->runtime_pm, wakeref);
1378 1379
}

1380 1381 1382 1383 1384
int i915_reset_trylock(struct drm_i915_private *i915)
{
	struct i915_gpu_error *error = &i915->gpu_error;
	int srcu;

1385 1386 1387
	might_lock(&error->reset_backoff_srcu);
	might_sleep();

1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412
	rcu_read_lock();
	while (test_bit(I915_RESET_BACKOFF, &error->flags)) {
		rcu_read_unlock();

		if (wait_event_interruptible(error->reset_queue,
					     !test_bit(I915_RESET_BACKOFF,
						       &error->flags)))
			return -EINTR;

		rcu_read_lock();
	}
	srcu = srcu_read_lock(&error->reset_backoff_srcu);
	rcu_read_unlock();

	return srcu;
}

void i915_reset_unlock(struct drm_i915_private *i915, int tag)
__releases(&i915->gpu_error.reset_backoff_srcu)
{
	struct i915_gpu_error *error = &i915->gpu_error;

	srcu_read_unlock(&error->reset_backoff_srcu, tag);
}

1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437
int i915_terminally_wedged(struct drm_i915_private *i915)
{
	struct i915_gpu_error *error = &i915->gpu_error;

	might_sleep();

	if (!__i915_wedged(error))
		return 0;

	/* Reset still in progress? Maybe we will recover? */
	if (!test_bit(I915_RESET_BACKOFF, &error->flags))
		return -EIO;

	/* XXX intel_reset_finish() still takes struct_mutex!!! */
	if (mutex_is_locked(&i915->drm.struct_mutex))
		return -EAGAIN;

	if (wait_event_interruptible(error->reset_queue,
				     !test_bit(I915_RESET_BACKOFF,
					       &error->flags)))
		return -EINTR;

	return __i915_wedged(error) ? -EIO : 0;
}

1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465
static void i915_wedge_me(struct work_struct *work)
{
	struct i915_wedge_me *w = container_of(work, typeof(*w), work.work);

	dev_err(w->i915->drm.dev,
		"%s timed out, cancelling all in-flight rendering.\n",
		w->name);
	i915_gem_set_wedged(w->i915);
}

void __i915_init_wedge(struct i915_wedge_me *w,
		       struct drm_i915_private *i915,
		       long timeout,
		       const char *name)
{
	w->i915 = i915;
	w->name = name;

	INIT_DELAYED_WORK_ONSTACK(&w->work, i915_wedge_me);
	schedule_delayed_work(&w->work, timeout);
}

void __i915_fini_wedge(struct i915_wedge_me *w)
{
	cancel_delayed_work_sync(&w->work);
	destroy_delayed_work_on_stack(&w->work);
	w->i915 = NULL;
}
1466 1467 1468 1469

#if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
#include "selftest_reset.c"
#endif