drm/i915/hangcheck: Prevent long walks across full-ppgtt

mikuint · mikuint · commit 24a65e624bcd · 2016-03-04T15:17:14.000+02:00
With full-ppgtt, it takes the GPU an eon to traverse the entire 256PiB address space, causing a loop to be detected. Under the current scheme, if ACTHD walks off the end of a batch buffer and into an empty address space, we "never" detect the hang. If we always increment the score as the ACTHD is progressing then we will eventually timeout (after ~46.5s (31 * 1.5s) without advancing onto a new batch). To counter act this, increase the amount we reduce the score for good batches, so that only a series of almost-bad batches trigger a full reset. DoS detection suffers slightly but series of long running shader tests will benefit. Based on a patch from Chris Wilson. Testcase: igt/drv_hangman/hangcheck-unterminated Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/1456930109-21532-1-git-send-email-mika.kuoppala@intel.com
diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -1367,8 +1367,6 @@ static int i915_hangcheck_info(struct seq_file *m, void *unused)
 		seq_printf(m, "\tACTHD = 0x%08llx [current 0x%08llx]\n",
 			   (long long)ring->hangcheck.acthd,
 			   (long long)acthd[i]);
-		seq_printf(m, "\tmax ACTHD = 0x%08llx\n",
-			   (long long)ring->hangcheck.max_acthd);
 		seq_printf(m, "\tscore = %d\n", ring->hangcheck.score);
 		seq_printf(m, "\taction = %d\n", ring->hangcheck.action);
 
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c
@@ -230,8 +230,6 @@ static const char *hangcheck_action_to_str(enum intel_ring_hangcheck_action a)
 		return "wait";
 	case HANGCHECK_ACTIVE:
 		return "active";
-	case HANGCHECK_ACTIVE_LOOP:
-		return "active (loop)";
 	case HANGCHECK_KICK:
 		return "kick";
 	case HANGCHECK_HUNG:
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
@@ -3001,12 +3001,7 @@ head_stuck(struct intel_engine_cs *ring, u64 acthd)
 		memset(ring->hangcheck.instdone, 0,
 		       sizeof(ring->hangcheck.instdone));
 
-		if (acthd > ring->hangcheck.max_acthd) {
-			ring->hangcheck.max_acthd = acthd;
-			return HANGCHECK_ACTIVE;
-		}
-
-		return HANGCHECK_ACTIVE_LOOP;
+		return HANGCHECK_ACTIVE;
 	}
 
 	if (!subunits_stuck(ring))
@@ -3083,6 +3078,7 @@ static void i915_hangcheck_elapsed(struct work_struct *work)
 #define BUSY 1
 #define KICK 5
 #define HUNG 20
+#define ACTIVE_DECAY 15
 
 	if (!i915.enable_hangcheck)
 		return;
@@ -3151,9 +3147,8 @@ static void i915_hangcheck_elapsed(struct work_struct *work)
 				switch (ring->hangcheck.action) {
 				case HANGCHECK_IDLE:
 				case HANGCHECK_WAIT:
-				case HANGCHECK_ACTIVE:
 					break;
-				case HANGCHECK_ACTIVE_LOOP:
+				case HANGCHECK_ACTIVE:
 					ring->hangcheck.score += BUSY;
 					break;
 				case HANGCHECK_KICK:
@@ -3172,10 +3167,12 @@ static void i915_hangcheck_elapsed(struct work_struct *work)
 			 * attempts across multiple batches.
 			 */
 			if (ring->hangcheck.score > 0)
-				ring->hangcheck.score--;
+				ring->hangcheck.score -= ACTIVE_DECAY;
+			if (ring->hangcheck.score < 0)
+				ring->hangcheck.score = 0;
 
 			/* Clear head and subunit states on seqno movement */
-			ring->hangcheck.acthd = ring->hangcheck.max_acthd = 0;
+			ring->hangcheck.acthd = 0;
 
 			memset(ring->hangcheck.instdone, 0,
 			       sizeof(ring->hangcheck.instdone));
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -79,7 +79,6 @@ enum intel_ring_hangcheck_action {
 	HANGCHECK_IDLE = 0,
 	HANGCHECK_WAIT,
 	HANGCHECK_ACTIVE,
-	HANGCHECK_ACTIVE_LOOP,
 	HANGCHECK_KICK,
 	HANGCHECK_HUNG,
 };
@@ -88,7 +87,6 @@ enum intel_ring_hangcheck_action {
 
 struct intel_ring_hangcheck {
 	u64 acthd;
-	u64 max_acthd;
 	u32 seqno;
 	int score;
 	enum intel_ring_hangcheck_action action;
