drm/i915/gt: Replace hangcheck by heartbeats

Replace sampling the engine state every so often with a periodic
heartbeat request to measure the health of an engine. This is coupled
with the forced-preemption to allow long running requests to survive so
long as they do not block other users.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Jon Bloomfield <jon.bloomfield@intel.com>
Reviewed-by: Jon Bloomfield <jon.bloomfield@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191023133108.21401-5-chris@chris-wilson.co.uk

25 files changed, 379 insertions, 555 deletions

diff --git a/drivers/gpu/drm/i915/Kconfig.profile b/drivers/gpu/drm/i915/Kconfig.profile
index b071b6024152..8ab7af5eb311 100644
--- a/drivers/gpu/drm/i915/Kconfig.profile
+++ b/drivers/gpu/drm/i915/Kconfig.profile
@@ -12,6 +12,17 @@ config DRM_I915_USERFAULT_AUTOSUSPEND
 	  May be 0 to disable the extra delay and solely use the device level
 	  runtime pm autosuspend delay tunable.
 
+config DRM_I915_HEARTBEAT_INTERVAL
+	int "Interval between heartbeat pulses (ms)"
+	default 2500 # milliseconds
+	help
+	  The driver sends a periodic heartbeat down all active engines to
+	  check the health of the GPU and undertake regular house-keeping of
+	  internal driver state.
+
+	  May be 0 to disable heartbeats and therefore disable automatic GPU
+	  hang detection.
+
 config DRM_I915_PREEMPT_TIMEOUT
 	int "Preempt timeout (ms, jiffy granularity)"
 	default 100 # milliseconds
diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile
index 2fd4bed188e5..21601bb27b22 100644
--- a/drivers/gpu/drm/i915/Makefile
+++ b/drivers/gpu/drm/i915/Makefile
@@ -87,7 +87,6 @@ gt-y += \
 	gt/intel_gt_pm.o \
 	gt/intel_gt_pm_irq.o \
 	gt/intel_gt_requests.o \
-	gt/intel_hangcheck.o \
 	gt/intel_llc.o \
 	gt/intel_lrc.o \
 	gt/intel_rc6.o \
diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
index 236fdf122e47..579655675b08 100644
--- a/drivers/gpu/drm/i915/display/intel_display.c
+++ b/drivers/gpu/drm/i915/display/intel_display.c
@@ -14863,7 +14863,7 @@ static void intel_plane_unpin_fb(struct intel_plane_state *old_plane_state)
 static void fb_obj_bump_render_priority(struct drm_i915_gem_object *obj)
 {
 	struct i915_sched_attr attr = {
-		.priority = I915_PRIORITY_DISPLAY,
+		.priority = I915_USER_PRIORITY(I915_PRIORITY_DISPLAY),
 	};
 
 	i915_gem_object_wait_priority(obj, 0, &attr);
diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object.h b/drivers/gpu/drm/i915/gem/i915_gem_object.h
index aead7e6725f9..458cd51331f1 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_object.h
+++ b/drivers/gpu/drm/i915/gem/i915_gem_object.h
@@ -462,6 +462,5 @@ int i915_gem_object_wait(struct drm_i915_gem_object *obj,
 int i915_gem_object_wait_priority(struct drm_i915_gem_object *obj,
 				  unsigned int flags,
 				  const struct i915_sched_attr *attr);
-#define I915_PRIORITY_DISPLAY I915_USER_PRIORITY(I915_PRIORITY_MAX)
 
 #endif
diff --git a/drivers/gpu/drm/i915/gem/i915_gem_pm.c b/drivers/gpu/drm/i915/gem/i915_gem_pm.c
index 9bc0cf3139e3..c99bb94fe41e 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_pm.c
+++ b/drivers/gpu/drm/i915/gem/i915_gem_pm.c
@@ -76,8 +76,6 @@ void i915_gem_suspend(struct drm_i915_private *i915)
 	intel_gt_suspend(&i915->gt);
 	intel_uc_suspend(&i915->gt.uc);
 
-	cancel_delayed_work_sync(&i915->gt.hangcheck.work);
-
 	i915_gem_drain_freed_objects(i915);
 }
 
diff --git a/drivers/gpu/drm/i915/gt/intel_engine.h b/drivers/gpu/drm/i915/gt/intel_engine.h
index 9409b7856299..ee47444a6ad4 100644
--- a/drivers/gpu/drm/i915/gt/intel_engine.h
+++ b/drivers/gpu/drm/i915/gt/intel_engine.h
@@ -91,38 +91,6 @@ struct intel_gt;
 /* seqno size is actually only a uint32, but since we plan to use MI_FLUSH_DW to
  * do the writes, and that must have qw aligned offsets, simply pretend it's 8b.
  */
-enum intel_engine_hangcheck_action {
-	ENGINE_IDLE = 0,
-	ENGINE_WAIT,
-	ENGINE_ACTIVE_SEQNO,
-	ENGINE_ACTIVE_HEAD,
-	ENGINE_ACTIVE_SUBUNITS,
-	ENGINE_WAIT_KICK,
-	ENGINE_DEAD,
-};
-
-static inline const char *
-hangcheck_action_to_str(const enum intel_engine_hangcheck_action a)
-{
-	switch (a) {
-	case ENGINE_IDLE:
-		return "idle";
-	case ENGINE_WAIT:
-		return "wait";
-	case ENGINE_ACTIVE_SEQNO:
-		return "active seqno";
-	case ENGINE_ACTIVE_HEAD:
-		return "active head";
-	case ENGINE_ACTIVE_SUBUNITS:
-		return "active subunits";
-	case ENGINE_WAIT_KICK:
-		return "wait kick";
-	case ENGINE_DEAD:
-		return "dead";
-	}
-
-	return "unknown";
-}
 
 static inline unsigned int
 execlists_num_ports(const struct intel_engine_execlists * const execlists)
diff --git a/drivers/gpu/drm/i915/gt/intel_engine_cs.c b/drivers/gpu/drm/i915/gt/intel_engine_cs.c
index b91ea07f4819..8b5129e0b49d 100644
--- a/drivers/gpu/drm/i915/gt/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/gt/intel_engine_cs.c
@@ -308,6 +308,8 @@ static int intel_engine_setup(struct intel_gt *gt, enum intel_engine_id id)
 	engine->instance = info->instance;
 	__sprint_engine_name(engine);
 
+	engine->props.heartbeat_interval_ms =
+		CONFIG_DRM_I915_HEARTBEAT_INTERVAL;
 	engine->props.preempt_timeout_ms =
 		CONFIG_DRM_I915_PREEMPT_TIMEOUT;
 	engine->props.stop_timeout_ms =
@@ -609,7 +611,6 @@ static int intel_engine_setup_common(struct intel_engine_cs *engine)
 	intel_engine_init_active(engine, ENGINE_PHYSICAL);
 	intel_engine_init_breadcrumbs(engine);
 	intel_engine_init_execlists(engine);
-	intel_engine_init_hangcheck(engine);
 	intel_engine_init_cmd_parser(engine);
 	intel_engine_init__pm(engine);
 
@@ -1470,8 +1471,13 @@ void intel_engine_dump(struct intel_engine_cs *engine,
 		drm_printf(m, "*** WEDGED ***\n");
 
 	drm_printf(m, "\tAwake? %d\n", atomic_read(&engine->wakeref.count));
-	drm_printf(m, "\tHangcheck: %d ms ago\n",
-		   jiffies_to_msecs(jiffies - engine->hangcheck.action_timestamp));
+
+	rcu_read_lock();
+	rq = READ_ONCE(engine->heartbeat.systole);
+	if (rq)
+		drm_printf(m, "\tHeartbeat: %d ms ago\n",
+			   jiffies_to_msecs(jiffies - rq->emitted_jiffies));
+	rcu_read_unlock();
 	drm_printf(m, "\tReset count: %d (global %d)\n",
 		   i915_reset_engine_count(error, engine),
 		   i915_reset_count(error));
diff --git a/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.c b/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.c
index 4b9ab7813d54..9977f59f6b53 100644
--- a/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.c
+++ b/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.c
@@ -11,6 +11,30 @@
 #include "intel_engine_pm.h"
 #include "intel_engine.h"
 #include "intel_gt.h"
+#include "intel_reset.h"
+
+/*
+ * While the engine is active, we send a periodic pulse along the engine
+ * to check on its health and to flush any idle-barriers. If that request
+ * is stuck, and we fail to preempt it, we declare the engine hung and
+ * issue a reset -- in the hope that restores progress.
+ */
+
+static bool next_heartbeat(struct intel_engine_cs *engine)
+{
+	long delay;
+
+	delay = READ_ONCE(engine->props.heartbeat_interval_ms);
+	if (!delay)
+		return false;
+
+	delay = msecs_to_jiffies_timeout(delay);
+	if (delay >= HZ)
+		delay = round_jiffies_up_relative(delay);
+	schedule_delayed_work(&engine->heartbeat.work, delay);
+
+	return true;
+}
 
 static void idle_pulse(struct intel_engine_cs *engine, struct i915_request *rq)
 {
@@ -18,6 +42,139 @@ static void idle_pulse(struct intel_engine_cs *engine, struct i915_request *rq)
 	i915_request_add_active_barriers(rq);
 }
 
+static void show_heartbeat(const struct i915_request *rq,
+			   struct intel_engine_cs *engine)
+{
+	struct drm_printer p = drm_debug_printer("heartbeat");
+
+	intel_engine_dump(engine, &p,
+			  "%s heartbeat {prio:%d} not ticking\n",
+			  engine->name,
+			  rq->sched.attr.priority);
+}
+
+static void heartbeat(struct work_struct *wrk)
+{
+	struct i915_sched_attr attr = {
+		.priority = I915_USER_PRIORITY(I915_PRIORITY_MIN),
+	};
+	struct intel_engine_cs *engine =
+		container_of(wrk, typeof(*engine), heartbeat.work.work);
+	struct intel_context *ce = engine->kernel_context;
+	struct i915_request *rq;
+
+	if (!intel_engine_pm_get_if_awake(engine))
+		return;
+
+	rq = engine->heartbeat.systole;
+	if (rq && i915_request_completed(rq)) {
+		i915_request_put(rq);
+		engine->heartbeat.systole = NULL;
+	}
+
+	if (intel_gt_is_wedged(engine->gt))
+		goto out;
+
+	if (engine->heartbeat.systole) {
+		if (engine->schedule &&
+		    rq->sched.attr.priority < I915_PRIORITY_BARRIER) {
+			/*
+			 * Gradually raise the priority of the heartbeat to
+			 * give high priority work [which presumably desires
+			 * low latency and no jitter] the chance to naturally
+			 * complete before being preempted.
+			 */
+			attr.priority = I915_PRIORITY_MASK;
+			if (rq->sched.attr.priority >= attr.priority)
+				attr.priority |= I915_USER_PRIORITY(I915_PRIORITY_HEARTBEAT);
+			if (rq->sched.attr.priority >= attr.priority)
+				attr.priority = I915_PRIORITY_BARRIER;
+
+			local_bh_disable();
+			engine->schedule(rq, &attr);
+			local_bh_enable();
+		} else {
+			if (IS_ENABLED(CONFIG_DRM_I915_DEBUG_GEM))
+				show_heartbeat(rq, engine);
+
+			intel_gt_handle_error(engine->gt, engine->mask,
+					      I915_ERROR_CAPTURE,
+					      "stopped heartbeat on %s",
+					      engine->name);
+		}
+		goto out;
+	}
+
+	if (engine->wakeref_serial == engine->serial)
+		goto out;
+
+	mutex_lock(&ce->timeline->mutex);
+
+	intel_context_enter(ce);
+	rq = __i915_request_create(ce, GFP_NOWAIT | __GFP_NOWARN);
+	intel_context_exit(ce);
+	if (IS_ERR(rq))
+		goto unlock;
+
+	idle_pulse(engine, rq);
+	if (i915_modparams.enable_hangcheck)
+		engine->heartbeat.systole = i915_request_get(rq);
+
+	__i915_request_commit(rq);
+	__i915_request_queue(rq, &attr);
+
+unlock:
+	mutex_unlock(&ce->timeline->mutex);
+out:
+	if (!next_heartbeat(engine))
+		i915_request_put(fetch_and_zero(&engine->heartbeat.systole));
+	intel_engine_pm_put(engine);
+}
+
+void intel_engine_unpark_heartbeat(struct intel_engine_cs *engine)
+{
+	if (!CONFIG_DRM_I915_HEARTBEAT_INTERVAL)
+		return;
+
+	next_heartbeat(engine);
+}
+
+void intel_engine_park_heartbeat(struct intel_engine_cs *engine)
+{
+	cancel_delayed_work(&engine->heartbeat.work);
+	i915_request_put(fetch_and_zero(&engine->heartbeat.systole));
+}
+
+void intel_engine_init_heartbeat(struct intel_engine_cs *engine)
+{
+	INIT_DELAYED_WORK(&engine->heartbeat.work, heartbeat);
+}
+
+int intel_engine_set_heartbeat(struct intel_engine_cs *engine,
+			       unsigned long delay)
+{
+	int err;
+
+	/* Send one last pulse before to cleanup persistent hogs */
+	if (!delay && CONFIG_DRM_I915_PREEMPT_TIMEOUT) {
+		err = intel_engine_pulse(engine);
+		if (err)
+			return err;
+	}
+
+	WRITE_ONCE(engine->props.heartbeat_interval_ms, delay);
+
+	if (intel_engine_pm_get_if_awake(engine)) {
+		if (delay)
+			intel_engine_unpark_heartbeat(engine);
+		else
+			intel_engine_park_heartbeat(engine);
+		intel_engine_pm_put(engine);
+	}
+
+	return 0;
+}
+
 int intel_engine_pulse(struct intel_engine_cs *engine)
 {
 	struct i915_sched_attr attr = { .priority = I915_PRIORITY_BARRIER };
diff --git a/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.h b/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.h
index b334e5aaf78d..a7b8c0f9e005 100644
--- a/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.h
+++ b/drivers/gpu/drm/i915/gt/intel_engine_heartbeat.h
@@ -9,6 +9,14 @@
 
 struct intel_engine_cs;
 
+void intel_engine_init_heartbeat(struct intel_engine_cs *engine);
+
+int intel_engine_set_heartbeat(struct intel_engine_cs *engine,
+			       unsigned long delay);
+
+void intel_engine_park_heartbeat(struct intel_engine_cs *engine);
+void intel_engine_unpark_heartbeat(struct intel_engine_cs *engine);
+
 int intel_engine_pulse(struct intel_engine_cs *engine);
 int intel_engine_flush_barriers(struct intel_engine_cs *engine);
 
diff --git a/drivers/gpu/drm/i915/gt/intel_engine_pm.c b/drivers/gpu/drm/i915/gt/intel_engine_pm.c
index 7d76611d9df1..6fbfa2162e54 100644
--- a/drivers/gpu/drm/i915/gt/intel_engine_pm.c
+++ b/drivers/gpu/drm/i915/gt/intel_engine_pm.c
@@ -7,6 +7,7 @@
 #include "i915_drv.h"
 
 #include "intel_engine.h"
+#include "intel_engine_heartbeat.h"
 #include "intel_engine_pm.h"
 #include "intel_engine_pool.h"
 #include "intel_gt.h"
@@ -34,7 +35,7 @@ static int __engine_unpark(struct intel_wakeref *wf)
 	if (engine->unpark)
 		engine->unpark(engine);
 
-	intel_engine_init_hangcheck(engine);
+	intel_engine_unpark_heartbeat(engine);
 	return 0;
 }
 
@@ -158,6 +159,7 @@ static int __engine_park(struct intel_wakeref *wf)
 
 	call_idle_barriers(engine); /* cleanup after wedging */
 
+	intel_engine_park_heartbeat(engine);
 	intel_engine_disarm_breadcrumbs(engine);
 	intel_engine_pool_park(&engine->pool);
 
@@ -188,6 +190,7 @@ void intel_engine_init__pm(struct intel_engine_cs *engine)
 	struct intel_runtime_pm *rpm = engine->uncore->rpm;
 
 	intel_wakeref_init(&engine->wakeref, rpm, &wf_ops);
+	intel_engine_init_heartbeat(engine);
 }
 
 #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
diff --git a/drivers/gpu/drm/i915/gt/intel_engine_types.h b/drivers/gpu/drm/i915/gt/intel_engine_types.h
index 1251dac91f31..26998901feff 100644
--- a/drivers/gpu/drm/i915/gt/intel_engine_types.h
+++ b/drivers/gpu/drm/i915/gt/intel_engine_types.h
@@ -15,6 +15,7 @@
 #include <linux/rbtree.h>
 #include <linux/timer.h>
 #include <linux/types.h>
+#include <linux/workqueue.h>
 
 #include "i915_gem.h"
 #include "i915_pmu.h"
@@ -76,14 +77,6 @@ struct intel_instdone {
 	u32 row[I915_MAX_SLICES][I915_MAX_SUBSLICES];
 };
 
-struct intel_engine_hangcheck {
-	u64 acthd;
-	u32 last_ring;
-	u32 last_head;
-	unsigned long action_timestamp;
-	struct intel_instdone instdone;
-};
-
 struct intel_ring {
 	struct kref ref;
 	struct i915_vma *vma;
@@ -331,6 +324,11 @@ struct intel_engine_cs {
 
 	intel_engine_mask_t saturated; /* submitting semaphores too late? */
 
+	struct {
+		struct delayed_work work;
+		struct i915_request *systole;
+	} heartbeat;
+
 	unsigned long serial;
 
 	unsigned long wakeref_serial;
@@ -481,8 +479,6 @@ struct intel_engine_cs {
 	/* status_notifier: list of callbacks for context-switch changes */
 	struct atomic_notifier_head context_status_notifier;
 
-	struct intel_engine_hangcheck hangcheck;
-
 #define I915_ENGINE_NEEDS_CMD_PARSER BIT(0)
 #define I915_ENGINE_SUPPORTS_STATS   BIT(1)
 #define I915_ENGINE_HAS_PREEMPTION   BIT(2)
@@ -549,6 +545,7 @@ struct intel_engine_cs {
 	} stats;
 
 	struct {
+		unsigned long heartbeat_interval_ms;
 		unsigned long preempt_timeout_ms;
 		unsigned long stop_timeout_ms;
 	} props;
diff --git a/drivers/gpu/drm/i915/gt/intel_gt.c b/drivers/gpu/drm/i915/gt/intel_gt.c
index 1c4b6c9642ad..35127699591d 100644
--- a/drivers/gpu/drm/i915/gt/intel_gt.c
+++ b/drivers/gpu/drm/i915/gt/intel_gt.c
@@ -22,7 +22,6 @@ void intel_gt_init_early(struct intel_gt *gt, struct drm_i915_private *i915)
 	INIT_LIST_HEAD(&gt->closed_vma);
 	spin_lock_init(&gt->closed_lock);
 
-	intel_gt_init_hangcheck(gt);
 	intel_gt_init_reset(gt);
 	intel_gt_init_requests(gt);
 	intel_gt_pm_init_early(gt);
diff --git a/drivers/gpu/drm/i915/gt/intel_gt.h b/drivers/gpu/drm/i915/gt/intel_gt.h
index e6ab0bff0efb..5b6effed3713 100644
--- a/drivers/gpu/drm/i915/gt/intel_gt.h
+++ b/drivers/gpu/drm/i915/gt/intel_gt.h
@@ -46,8 +46,6 @@ void intel_gt_clear_error_registers(struct intel_gt *gt,
 void intel_gt_flush_ggtt_writes(struct intel_gt *gt);
 void intel_gt_chipset_flush(struct intel_gt *gt);
 
-void intel_gt_init_hangcheck(struct intel_gt *gt);
-
 static inline u32 intel_gt_scratch_offset(const struct intel_gt *gt,
 					  enum intel_gt_scratch_field field)
 {
@@ -59,6 +57,4 @@ static inline bool intel_gt_is_wedged(struct intel_gt *gt)
 	return __intel_reset_failed(&gt->reset);
 }
 
-void intel_gt_queue_hangcheck(struct intel_gt *gt);
-
 #endif /* __INTEL_GT_H__ */
diff --git a/drivers/gpu/drm/i915/gt/intel_gt_pm.c b/drivers/gpu/drm/i915/gt/intel_gt_pm.c
index 06e73d56cfcf..aff4eda47819 100644
--- a/drivers/gpu/drm/i915/gt/intel_gt_pm.c
+++ b/drivers/gpu/drm/i915/gt/intel_gt_pm.c
@@ -47,7 +47,6 @@ static int __gt_unpark(struct intel_wakeref *wf)
 
 	i915_pmu_gt_unparked(i915);
 
-	intel_gt_queue_hangcheck(gt);
 	intel_gt_unpark_requests(gt);
 
 	return 0;
diff --git a/drivers/gpu/drm/i915/gt/intel_gt_types.h b/drivers/gpu/drm/i915/gt/intel_gt_types.h
index 980973e66e7f..fcb3a6c9421c 100644
--- a/drivers/gpu/drm/i915/gt/intel_gt_types.h
+++ b/drivers/gpu/drm/i915/gt/intel_gt_types.h
@@ -27,14 +27,6 @@ struct i915_ggtt;
 struct intel_engine_cs;
 struct intel_uncore;
 
-struct intel_hangcheck {
-	/* For hangcheck timer */
-#define DRM_I915_HANGCHECK_PERIOD 1500 /* in ms */
-#define DRM_I915_HANGCHECK_JIFFIES msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD)
-
-	struct delayed_work work;
-};
-
 struct intel_gt {
 	struct drm_i915_private *i915;
 	struct intel_uncore *uncore;
@@ -68,7 +60,6 @@ struct intel_gt {
 	struct list_head closed_vma;
 	spinlock_t closed_lock; /* guards the list of closed_vma */
 
-	struct intel_hangcheck hangcheck;
 	struct intel_reset reset;
 
 	/**
diff --git a/drivers/gpu/drm/i915/gt/intel_hangcheck.c b/drivers/gpu/drm/i915/gt/intel_hangcheck.c
deleted file mode 100644
index 0fdef00af9e4..000000000000
--- a/drivers/gpu/drm/i915/gt/intel_hangcheck.c
+++ /dev/null
@@ -1,361 +0,0 @@
-/*
- * Copyright © 2016 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_engine.h"
-#include "intel_gt.h"
-#include "intel_reset.h"
-
-struct hangcheck {
-	u64 acthd;
-	u32 ring;
-	u32 head;
-	enum intel_engine_hangcheck_action action;
-	unsigned long action_timestamp;
-	int deadlock;
-	struct intel_instdone instdone;
-	bool wedged:1;
-	bool stalled:1;
-};
-
-static bool instdone_unchanged(u32 current_instdone, u32 *old_instdone)
-{
-	u32 tmp = current_instdone | *old_instdone;
-	bool unchanged;
-
-	unchanged = tmp == *old_instdone;
-	*old_instdone |= tmp;
-
-	return unchanged;
-}
-
-static bool subunits_stuck(struct intel_engine_cs *engine)
-{
-	struct drm_i915_private *dev_priv = engine->i915;
-	const struct sseu_dev_info *sseu = &RUNTIME_INFO(dev_priv)->sseu;
-	struct intel_instdone instdone;
-	struct intel_instdone *accu_instdone = &engine->hangcheck.instdone;
-	bool stuck;
-	int slice;
-	int subslice;
-
-	intel_engine_get_instdone(engine, &instdone);
-
-	/* There might be unstable subunit states even when
-	 * actual head is not moving. Filter out the unstable ones by
-	 * accumulating the undone -> done transitions and only
-	 * consider those as progress.
-	 */
-	stuck = instdone_unchanged(instdone.instdone,
-				   &accu_instdone->instdone);
-	stuck &= instdone_unchanged(instdone.slice_common,
-				    &accu_instdone->slice_common);
-
-	for_each_instdone_slice_subslice(dev_priv, sseu, slice, subslice) {
-		stuck &= instdone_unchanged(instdone.sampler[slice][subslice],
-					    &accu_instdone->sampler[slice][subslice]);
-		stuck &= instdone_unchanged(instdone.row[slice][subslice],
-					    &accu_instdone->row[slice][subslice]);
-	}
-
-	return stuck;
-}
-
-static enum intel_engine_hangcheck_action
-head_stuck(struct intel_engine_cs *engine, u64 acthd)
-{
-	if (acthd != engine->hangcheck.acthd) {
-
-		/* Clear subunit states on head movement */
-		memset(&engine->hangcheck.instdone, 0,
-		       sizeof(engine->hangcheck.instdone));
-
-		return ENGINE_ACTIVE_HEAD;
-	}
-
-	if (!subunits_stuck(engine))
-		return ENGINE_ACTIVE_SUBUNITS;
-
-	return ENGINE_DEAD;
-}
-
-static enum intel_engine_hangcheck_action
-engine_stuck(struct intel_engine_cs *engine, u64 acthd)
-{
-	enum intel_engine_hangcheck_action ha;
-	u32 tmp;
-
-	ha = head_stuck(engine, acthd);
-	if (ha != ENGINE_DEAD)
-		return ha;
-
-	if (IS_GEN(engine->i915, 2))
-		return ENGINE_DEAD;
-
-	/* Is the chip hanging on a WAIT_FOR_EVENT?
-	 * If so we can simply poke the RB_WAIT bit
-	 * and break the hang. This should work on
-	 * all but the second generation chipsets.
-	 */
-	tmp = ENGINE_READ(engine, RING_CTL);
-	if (tmp & RING_WAIT) {
-		intel_gt_handle_error(engine->gt, engine->mask, 0,
-				      "stuck wait on %s", engine->name);
-		ENGINE_WRITE(engine, RING_CTL, tmp);
-		return ENGINE_WAIT_KICK;
-	}
-
-	return ENGINE_DEAD;
-}
-
-static void hangcheck_load_sample(struct intel_engine_cs *engine,
-				  struct hangcheck *hc)
-{
-	hc->acthd = intel_engine_get_active_head(engine);
-	hc->ring = ENGINE_READ(engine, RING_START);
-	hc->head = ENGINE_READ(engine, RING_HEAD);
-}
-
-static void hangcheck_store_sample(struct intel_engine_cs *engine,
-				   const struct hangcheck *hc)
-{
-	engine->hangcheck.acthd = hc->acthd;
-	engine->hangcheck.last_ring = hc->ring;
-	engine->hangcheck.last_head = hc->head;
-}
-
-static enum intel_engine_hangcheck_action
-hangcheck_get_action(struct intel_engine_cs *engine,
-		     const struct hangcheck *hc)
-{
-	if (intel_engine_is_idle(engine))
-		return ENGINE_IDLE;
-
-	if (engine->hangcheck.last_ring != hc->ring)
-		return ENGINE_ACTIVE_SEQNO;
-
-	if (engine->hangcheck.last_head != hc->head)
-		return ENGINE_ACTIVE_SEQNO;
-
-	return engine_stuck(engine, hc->acthd);
-}
-
-static void hangcheck_accumulate_sample(struct intel_engine_cs *engine,
-					struct hangcheck *hc)
-{
-	unsigned long timeout = I915_ENGINE_DEAD_TIMEOUT;
-
-	hc->action = hangcheck_get_action(engine, hc);
-
-	/* We always increment the progress
-	 * if the engine is busy and still processing
-	 * the same request, so that no single request
-	 * can run indefinitely (such as a chain of
-	 * batches). The only time we do not increment
-	 * the hangcheck score on this ring, if this
-	 * engine is in a legitimate wait for another
-	 * engine. In that case the waiting engine is a
-	 * victim and we want to be sure we catch the
-	 * right culprit. Then every time we do kick
-	 * the ring, make it as a progress as the seqno
-	 * advancement might ensure and if not, it
-	 * will catch the hanging engine.
-	 */
-
-	switch (hc->action) {
-	case ENGINE_IDLE:
-	case ENGINE_ACTIVE_SEQNO:
-		/* Clear head and subunit states on seqno movement */
-		hc->acthd = 0;
-
-		memset(&engine->hangcheck.instdone, 0,
-		       sizeof(engine->hangcheck.instdone));
-
-		/* Intentional fall through */
-	case ENGINE_WAIT_KICK:
-	case ENGINE_WAIT:
-		engine->hangcheck.action_timestamp = jiffies;
-		break;
-
-	case ENGINE_ACTIVE_HEAD:
-	case ENGINE_ACTIVE_SUBUNITS:
-		/*
-		 * Seqno stuck with still active engine gets leeway,
-		 * in hopes that it is just a long shader.
-		 */
-		timeout = I915_SEQNO_DEAD_TIMEOUT;
-		break;
-
-	case ENGINE_DEAD:
-		break;
-
-	default:
-		MISSING_CASE(hc->action);
-	}
-
-	hc->stalled = time_after(jiffies,
-				 engine->hangcheck.action_timestamp + timeout);
-	hc->wedged = time_after(jiffies,
-				 engine->hangcheck.action_timestamp +
-				 I915_ENGINE_WEDGED_TIMEOUT);
-}
-
-static void hangcheck_declare_hang(struct intel_gt *gt,
-				   intel_engine_mask_t hung,
-				   intel_engine_mask_t stuck)
-{
-	struct intel_engine_cs *engine;
-	intel_engine_mask_t tmp;
-	char msg[80];
-	int len;
-
-	/* If some rings hung but others were still busy, only
-	 * blame the hanging rings in the synopsis.
-	 */
-	if (stuck != hung)
-		hung &= ~stuck;
-	len = scnprintf(msg, sizeof(msg),
-			"%s on ", stuck == hung ? "no progress" : "hang");
-	for_each_engine_masked(engine, gt, hung, tmp)
-		len += scnprintf(msg + len, sizeof(msg) - len,
-				 "%s, ", engine->name);
-	msg[len-2] = '\0';
-
-	return intel_gt_handle_error(gt, hung, I915_ERROR_CAPTURE, "%s", msg);
-}
-
-/*
- * This is called when the chip hasn't reported back with completed
- * batchbuffers in a long time. We keep track per ring seqno progress and
- * if there are no progress, hangcheck score for that ring is increased.
- * Further, acthd is inspected to see if the ring is stuck. On stuck case
- * we kick the ring. If we see no progress on three subsequent calls
- * we assume chip is wedged and try to fix it by resetting the chip.
- */
-static void hangcheck_elapsed(struct work_struct *work)
-{
-	struct intel_gt *gt =
-		container_of(work, typeof(*gt), hangcheck.work.work);
-	intel_engine_mask_t hung = 0, stuck = 0, wedged = 0;
-	struct intel_engine_cs *engine;
-	enum intel_engine_id id;
-	intel_wakeref_t wakeref;
-
-	if (!i915_modparams.enable_hangcheck)
-		return;
-
-	if (!READ_ONCE(gt->awake))
-		return;
-
-	if (intel_gt_is_wedged(gt))
-		return;
-
-	wakeref = intel_runtime_pm_get_if_in_use(gt->uncore->rpm);
-	if (!wakeref)
-		return;
-
-	/* As enabling the GPU requires fairly extensive mmio access,
-	 * periodically arm the mmio checker to see if we are triggering
-	 * any invalid access.
-	 */
-	intel_uncore_arm_unclaimed_mmio_detection(gt->uncore);
-
-	for_each_engine(engine, gt, id) {
-		struct hangcheck hc;
-
-		intel_engine_breadcrumbs_irq(engine);
-
-		hangcheck_load_sample(engine, &hc);
-		hangcheck_accumulate_sample(engine, &hc);
-		hangcheck_store_sample(engine, &hc);
-
-		if (hc.stalled) {
-			hung |= engine->mask;
-			if (hc.action != ENGINE_DEAD)
-				stuck |= engine->mask;
-		}
-
-		if (hc.wedged)
-			wedged |= engine->mask;
-	}
-
-	if (GEM_SHOW_DEBUG() && (hung | stuck)) {
-		struct drm_printer p = drm_debug_printer("hangcheck");
-
-		for_each_engine(engine, gt, id) {
-			if (intel_engine_is_idle(engine))
-				continue;
-
-			intel_engine_dump(engine, &p, "%