[Intel-gfx] [PATCH 1/2] drm/i915: Trace GEM steps between submit and wedging

[Chris Wilson]

We still have an odd race with wedging/unwedging as shown by igt/gem_eio
that defies expectations. Add some more trace_printks to try and
visualize the flow over the precipice.

Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala at linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
---
 drivers/gpu/drm/i915/i915_gem.c     | 14 ++++++++++++++
 drivers/gpu/drm/i915/i915_request.c | 23 +++++++++++++++++++++++
 2 files changed, 37 insertions(+)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 13d4b0e74641..2fbd622bba30 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3193,6 +3193,9 @@ void i915_gem_reset_finish(struct drm_i915_private *dev_priv)
 
 static void nop_submit_request(struct i915_request *request)
 {
+	GEM_TRACE("%s fence %llx:%d -> -EIO\n",
+		  request->engine->name,
+		  request->fence.context, request->fence.seqno);
 	dma_fence_set_error(&request->fence, -EIO);
 
 	i915_request_submit(request);
@@ -3202,6 +3205,9 @@ static void nop_complete_submit_request(struct i915_request *request)
 {
 	unsigned long flags;
 
+	GEM_TRACE("%s fence %llx:%d -> -EIO\n",
+		  request->engine->name,
+		  request->fence.context, request->fence.seqno);
 	dma_fence_set_error(&request->fence, -EIO);
 
 	spin_lock_irqsave(&request->engine->timeline->lock, flags);
@@ -3215,6 +3221,8 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
 
+	GEM_TRACE("start\n");
+
 	if (drm_debug & DRM_UT_DRIVER) {
 		struct drm_printer p = drm_debug_printer(__func__);
 
@@ -3279,6 +3287,8 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 		i915_gem_reset_finish_engine(engine);
 	}
 
+	GEM_TRACE("end\n");
+
 	wake_up_all(&i915->gpu_error.reset_queue);
 }
 
@@ -3291,6 +3301,8 @@ bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 	if (!test_bit(I915_WEDGED, &i915->gpu_error.flags))
 		return true;
 
+	GEM_TRACE("start\n");
+
 	/*
 	 * Before unwedging, make sure that all pending operations
 	 * are flushed and errored out - we may have requests waiting upon
@@ -3341,6 +3353,8 @@ bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 	intel_engines_reset_default_submission(i915);
 	i915_gem_contexts_lost(i915);
 
+	GEM_TRACE("end\n");
+
 	smp_mb__before_atomic(); /* complete takeover before enabling execbuf */
 	clear_bit(I915_WEDGED, &i915->gpu_error.flags);
 
diff --git a/drivers/gpu/drm/i915/i915_request.c b/drivers/gpu/drm/i915/i915_request.c
index 1810fa1b81cb..43c7134a9b93 100644
--- a/drivers/gpu/drm/i915/i915_request.c
+++ b/drivers/gpu/drm/i915/i915_request.c
@@ -207,11 +207,16 @@ static int reset_all_global_seqno(struct drm_i915_private *i915, u32 seqno)
 	if (ret)
 		return ret;
 
+	GEM_BUG_ON(i915->gt.active_requests);
+
 	/* If the seqno wraps around, we need to clear the breadcrumb rbtree */
 	for_each_engine(engine, i915, id) {
 		struct i915_gem_timeline *timeline;
 		struct intel_timeline *tl = engine->timeline;
 
+		GEM_TRACE("%s seqno %d -> %d\n",
+			  engine->name, tl->seqno, seqno);
+
 		if (!i915_seqno_passed(seqno, tl->seqno)) {
 			/* Flush any waiters before we reuse the seqno */
 			intel_engine_disarm_breadcrumbs(engine);
@@ -381,6 +386,11 @@ static void i915_request_retire(struct i915_request *request)
 	struct intel_engine_cs *engine = request->engine;
 	struct i915_gem_active *active, *next;
 
+	GEM_TRACE("%s(%d) fence %llx:%d, global_seqno %d\n",
+		  engine->name, intel_engine_get_seqno(engine),
+		  request->fence.context, request->fence.seqno,
+		  request->global_seqno);
+
 	lockdep_assert_held(&request->i915->drm.struct_mutex);
 	GEM_BUG_ON(!i915_sw_fence_signaled(&request->submit));
 	GEM_BUG_ON(!i915_request_completed(request));
@@ -488,6 +498,11 @@ void __i915_request_submit(struct i915_request *request)
 	struct intel_timeline *timeline;
 	u32 seqno;
 
+	GEM_TRACE("%s fence %llx:%d -> global_seqno %d\n",
+		  request->engine->name,
+		  request->fence.context, request->fence.seqno,
+		  engine->timeline->seqno);
+
 	GEM_BUG_ON(!irqs_disabled());
 	lockdep_assert_held(&engine->timeline->lock);
 
@@ -537,6 +552,11 @@ void __i915_request_unsubmit(struct i915_request *request)
 	struct intel_engine_cs *engine = request->engine;
 	struct intel_timeline *timeline;
 
+	GEM_TRACE("%s fence %llx:%d <- global_seqno %d\n",
+		  request->engine->name,
+		  request->fence.context, request->fence.seqno,
+		  request->global_seqno);
+
 	GEM_BUG_ON(!irqs_disabled());
 	lockdep_assert_held(&engine->timeline->lock);
 
@@ -996,6 +1016,9 @@ void __i915_request_add(struct i915_request *request, bool flush_caches)
 	u32 *cs;
 	int err;
 
+	GEM_TRACE("%s fence %llx:%d\n",
+		  engine->name, request->fence.context, request->fence.seqno);
+
 	lockdep_assert_held(&request->i915->drm.struct_mutex);
 	trace_i915_request_add(request);
 
-- 
2.16.2