drm/i915: Inspect subunit states on hangcheck

If head seems stuck and engine in question is rcs,
inspect subunit state transitions from undone to done,
before deciding that this really is a hang instead of limited
progress. Only account the transitions of subunits from
undone to done once, to prevent unstable subunit states
to keep us falsely active.

As this adds one extra steps to hangcheck heuristics,
before hang is declared, it adds 1500ms to to detect hang
for render ring to a total of 7500ms. We could sample
the subunit states on first head stuck condition but
decide not to do so only in order to mimic old behaviour. This
way the check order of promotion from seqno > atchd > instdone
is consistently done.

v2: Deal with unstable done states (Arun)
    Clear instdone progress on head and seqno movement (Chris)
    Report raw and accumulated instdone's in in debugfs (Chris)
    Return HANGCHECK_ACTIVE on undone->done

References: https://bugs.freedesktop.org/show_bug.cgi?id=93029
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Dave Gordon <david.s.gordon@intel.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Arun Siluvery <arun.siluvery@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1448985372-19535-1-git-send-email-mika.kuoppala@intel.com

drivers/gpu/drm/i915/i915_debugfs.c

@@ -1331,7 +1331,8 @@ static int i915_hangcheck_info(struct seq_file *m, void *unused)
 	struct intel_engine_cs *ring;
 	u64 acthd[I915_NUM_RINGS];
 	u32 seqno[I915_NUM_RINGS];
-	int i;
+	u32 instdone[I915_NUM_INSTDONE_REG];
+	int i, j;
 
 	if (!i915.enable_hangcheck) {
 		seq_printf(m, "Hangcheck disabled\n");
@@ -1345,6 +1346,8 @@ static int i915_hangcheck_info(struct seq_file *m, void *unused)
 		acthd[i] = intel_ring_get_active_head(ring);
 	}
 
+	i915_get_extra_instdone(dev, instdone);
+
 	intel_runtime_pm_put(dev_priv);
 
 	if (delayed_work_pending(&dev_priv->gpu_error.hangcheck_work)) {
@@ -1365,6 +1368,21 @@ static int i915_hangcheck_info(struct seq_file *m, void *unused)
 			   (long long)ring->hangcheck.max_acthd);
 		seq_printf(m, "\tscore = %d\n", ring->hangcheck.score);
 		seq_printf(m, "\taction = %d\n", ring->hangcheck.action);
+
+		if (ring->id == RCS) {
+			seq_puts(m, "\tinstdone read =");
+
+			for (j = 0; j < I915_NUM_INSTDONE_REG; j++)
+				seq_printf(m, " 0x%08x", instdone[j]);
+
+			seq_puts(m, "\n\tinstdone accu =");
+
+			for (j = 0; j < I915_NUM_INSTDONE_REG; j++)
+				seq_printf(m, " 0x%08x",
+					   ring->hangcheck.instdone[j]);
+
+			seq_puts(m, "\n");
+		}
 	}
 
 	return 0;

drivers/gpu/drm/i915/i915_irq.c

@@ -2949,14 +2949,44 @@ static void semaphore_clear_deadlocks(struct drm_i915_private *dev_priv)
 		ring->hangcheck.deadlock = 0;
 }
 
-static enum intel_ring_hangcheck_action
-ring_stuck(struct intel_engine_cs *ring, u64 acthd)
+static bool subunits_stuck(struct intel_engine_cs *ring)
 {
-	struct drm_device *dev = ring->dev;
-	struct drm_i915_private *dev_priv = dev->dev_private;
-	u32 tmp;
+	u32 instdone[I915_NUM_INSTDONE_REG];
+	bool stuck;
+	int i;
 
+	if (ring->id != RCS)
+		return true;
+
+	i915_get_extra_instdone(ring->dev, 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 = true;
+	for (i = 0; i < I915_NUM_INSTDONE_REG; i++) {
+		const u32 tmp = instdone[i] | ring->hangcheck.instdone[i];
+
+		if (tmp != ring->hangcheck.instdone[i])
+			stuck = false;
+
+		ring->hangcheck.instdone[i] |= tmp;
+	}
+
+	return stuck;
+}
+
+static enum intel_ring_hangcheck_action
+head_stuck(struct intel_engine_cs *ring, u64 acthd)
+{
 	if (acthd != ring->hangcheck.acthd) {
+
+		/* Clear subunit states on head movement */
+		memset(ring->hangcheck.instdone, 0,
+		       sizeof(ring->hangcheck.instdone));
+
 		if (acthd > ring->hangcheck.max_acthd) {
 			ring->hangcheck.max_acthd = acthd;
 			return HANGCHECK_ACTIVE;
@@ -2965,6 +2995,24 @@ ring_stuck(struct intel_engine_cs *ring, u64 acthd)
 		return HANGCHECK_ACTIVE_LOOP;
 	}
 
+	if (!subunits_stuck(ring))
+		return HANGCHECK_ACTIVE;
+
+	return HANGCHECK_HUNG;
+}
+
+static enum intel_ring_hangcheck_action
+ring_stuck(struct intel_engine_cs *ring, u64 acthd)
+{
+	struct drm_device *dev = ring->dev;
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	enum intel_ring_hangcheck_action ha;
+	u32 tmp;
+
+	ha = head_stuck(ring, acthd);
+	if (ha != HANGCHECK_HUNG)
+		return ha;
+
 	if (IS_GEN2(dev))
 		return HANGCHECK_HUNG;
 
@@ -3106,7 +3154,11 @@ static void i915_hangcheck_elapsed(struct work_struct *work)
 			if (ring->hangcheck.score > 0)
 				ring->hangcheck.score--;
 
+			/* Clear head and subunit states on seqno movement */
 			ring->hangcheck.acthd = ring->hangcheck.max_acthd = 0;
+
+			memset(ring->hangcheck.instdone, 0,
+			       sizeof(ring->hangcheck.instdone));
 		}
 
 		ring->hangcheck.seqno = seqno;

drivers/gpu/drm/i915/intel_ringbuffer.h

@@ -93,6 +93,7 @@ struct intel_ring_hangcheck {
 	int score;
 	enum intel_ring_hangcheck_action action;
 	int deadlock;
+	u32 instdone[I915_NUM_INSTDONE_REG];
 };
 
 struct intel_ringbuffer {