drm/i915: Preallocate next seqno before touching the ring

Based on the work by Mika Kuoppala, we realised that we need to handle
seqno wraparound prior to committing our changes to the ring. The most
obvious point then is to grab the seqno inside intel_ring_begin(), and
then to reuse that seqno for all ring operations until the next request.
As intel_ring_begin() can fail, the callers must already be prepared to
handle such failure and so we can safely add further checks.

This patch looks like it should be split up into the interface
changes and the tweaks to move seqno wrapping from the execbuffer into
the core seqno increment. However, I found no easy way to break it into
incremental steps without introducing further broken behaviour.

v2: Mika found a silly mistake and a subtle error in the existing code;
inside i915_gem_retire_requests() we were resetting the sync_seqno of
the target ring based on the seqno from this ring - which are only
related by the order of their allocation, not retirement. Hence we were
applying the optimisation that the rings were synchronised too early,
fortunately the only real casualty there is the handling of seqno
wrapping.

v3: Do not forget to reset the sync_seqno upon module reinitialisation,
ala resume.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=863861
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com> [v2]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

1 files changed, 47 insertions, 27 deletions

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 9be450e7c54f..3b9b250ceac4 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -1857,11 +1857,11 @@ i915_gem_object_get_pages(struct drm_i915_gem_object *obj)
 
 void
 i915_gem_object_move_to_active(struct drm_i915_gem_object *obj,
-			       struct intel_ring_buffer *ring,
-			       u32 seqno)
+			       struct intel_ring_buffer *ring)
 {
 	struct drm_device *dev = obj->base.dev;
 	struct drm_i915_private *dev_priv = dev->dev_private;
+	u32 seqno = intel_ring_get_seqno(ring);
 
 	BUG_ON(ring == NULL);
 	obj->ring = ring;
@@ -1922,26 +1922,54 @@ i915_gem_object_move_to_inactive(struct drm_i915_gem_object *obj)
 	WARN_ON(i915_verify_lists(dev));
 }
 
-static u32
-i915_gem_get_seqno(struct drm_device *dev)
+static int
+i915_gem_handle_seqno_wrap(struct drm_device *dev)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
-	u32 seqno = dev_priv->next_seqno;
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct intel_ring_buffer *ring;
+	int ret, i, j;
 
-	/* reserve 0 for non-seqno */
-	if (++dev_priv->next_seqno == 0)
-		dev_priv->next_seqno = 1;
+	/* The hardware uses various monotonic 32-bit counters, if we
+	 * detect that they will wraparound we need to idle the GPU
+	 * and reset those counters.
+	 */
+	ret = 0;
+	for_each_ring(ring, dev_priv, i) {
+		for (j = 0; j < ARRAY_SIZE(ring->sync_seqno); j++)
+			ret |= ring->sync_seqno[j] != 0;
+	}
+	if (ret == 0)
+		return ret;
 
-	return seqno;
+	ret = i915_gpu_idle(dev);
+	if (ret)
+		return ret;
+
+	i915_gem_retire_requests(dev);
+	for_each_ring(ring, dev_priv, i) {
+		for (j = 0; j < ARRAY_SIZE(ring->sync_seqno); j++)
+			ring->sync_seqno[j] = 0;
+	}
+
+	return 0;
 }
 
-u32
-i915_gem_next_request_seqno(struct intel_ring_buffer *ring)
+int
+i915_gem_get_seqno(struct drm_device *dev, u32 *seqno)
 {
-	if (ring->outstanding_lazy_request == 0)
-		ring->outstanding_lazy_request = i915_gem_get_seqno(ring->dev);
+	struct drm_i915_private *dev_priv = dev->dev_private;
+
+	/* reserve 0 for non-seqno */
+	if (dev_priv->next_seqno == 0) {
+		int ret = i915_gem_handle_seqno_wrap(dev);
+		if (ret)
+			return ret;
 
-	return ring->outstanding_lazy_request;
+		dev_priv->next_seqno = 1;
+	}
+
+	*seqno = dev_priv->next_seqno++;
+	return 0;
 }
 
 int
@@ -1952,7 +1980,6 @@ i915_add_request(struct intel_ring_buffer *ring,
 	drm_i915_private_t *dev_priv = ring->dev->dev_private;
 	struct drm_i915_gem_request *request;
 	u32 request_ring_position;
-	u32 seqno;
 	int was_empty;
 	int ret;
 
@@ -1971,7 +1998,6 @@ i915_add_request(struct intel_ring_buffer *ring,
 	if (request == NULL)
 		return -ENOMEM;
 
-	seqno = i915_gem_next_request_seqno(ring);
 
 	/* Record the position of the start of the request so that
 	 * should we detect the updated seqno part-way through the
@@ -1980,15 +2006,13 @@ i915_add_request(struct intel_ring_buffer *ring,
 	 */
 	request_ring_position = intel_ring_get_tail(ring);
 
-	ret = ring->add_request(ring, &seqno);
+	ret = ring->add_request(ring);
 	if (ret) {
 		kfree(request);
 		return ret;
 	}
 
-	trace_i915_gem_request_add(ring, seqno);
-
-	request->seqno = seqno;
+	request->seqno = intel_ring_get_seqno(ring);
 	request->ring = ring;
 	request->tail = request_ring_position;
 	request->emitted_jiffies = jiffies;
@@ -2006,6 +2030,7 @@ i915_add_request(struct intel_ring_buffer *ring,
 		spin_unlock(&file_priv->mm.lock);
 	}
 
+	trace_i915_gem_request_add(ring, request->seqno);
 	ring->outstanding_lazy_request = 0;
 
 	if (!dev_priv->mm.suspended) {
@@ -2022,7 +2047,7 @@ i915_add_request(struct intel_ring_buffer *ring,
 	}
 
 	if (out_seqno)
-		*out_seqno = seqno;
+		*out_seqno = request->seqno;
 	return 0;
 }
 
@@ -2120,7 +2145,6 @@ void
 i915_gem_retire_requests_ring(struct intel_ring_buffer *ring)
 {
 	uint32_t seqno;
-	int i;
 
 	if (list_empty(&ring->request_list))
 		return;
@@ -2129,10 +2153,6 @@ i915_gem_retire_requests_ring(struct intel_ring_buffer *ring)
 
 	seqno = ring->get_seqno(ring, true);
 
-	for (i = 0; i < ARRAY_SIZE(ring->sync_seqno); i++)
-		if (seqno >= ring->sync_seqno[i])
-			ring->sync_seqno[i] = 0;
-
 	while (!list_empty(&ring->request_list)) {
 		struct drm_i915_gem_request *request;