[Intel-gfx] [PATCH] drm/i915: untangle page flip completion

[Jesse Barnes]

When a new page flip is requested, we need to both queue an unpin for
the current framebuffer, and also increment the flip pending count on
the newly submitted buffer.

At flip finish time, we need to unpin the old fb and decrement the flip
pending count on the new buffer.

The old code was conflating the two, and led to hangs when new direct
rendered apps were started, replacing the existing frame buffer.  This
patch splits out the buffers and prevents the hangs.

Signed-off-by: Jesse Barnes <jbarnes at virtuousgeek.org>

--

This one needs careful review.  I could be way off here, but it really
seems to fix the hangs (for real this time!).

diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 5f14dfb..00f2b10 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -3950,7 +3950,8 @@ static void intel_crtc_destroy(struct drm_crtc *crtc)
 struct intel_unpin_work {
 	struct work_struct work;
 	struct drm_device *dev;
-	struct drm_gem_object *obj;
+	struct drm_gem_object *old_fb_obj;
+	struct drm_gem_object *pending_flip_obj;
 	struct drm_pending_vblank_event *event;
 	int pending;
 };
@@ -3961,8 +3962,8 @@ static void intel_unpin_work_fn(struct work_struct *__work)
 		container_of(__work, struct intel_unpin_work, work);
 
 	mutex_lock(&work->dev->struct_mutex);
-	i915_gem_object_unpin(work->obj);
-	drm_gem_object_unreference(work->obj);
+	i915_gem_object_unpin(work->old_fb_obj);
+	drm_gem_object_unreference(work->old_fb_obj);
 	mutex_unlock(&work->dev->struct_mutex);
 	kfree(work);
 }
@@ -3986,7 +3987,7 @@ void intel_finish_page_flip(struct drm_device *dev, int pipe)
 	work = intel_crtc->unpin_work;
 	if (work == NULL || !work->pending) {
 		if (work && !work->pending) {
-			obj_priv = work->obj->driver_private;
+			obj_priv = work->pending_flip_obj->driver_private;
 			DRM_DEBUG_DRIVER("flip finish: %p (%d) not pending?\n",
 					 obj_priv,
 					 atomic_read(&obj_priv->pending_flip));
@@ -4011,7 +4012,7 @@ void intel_finish_page_flip(struct drm_device *dev, int pipe)
 
 	spin_unlock_irqrestore(&dev->event_lock, flags);
 
-	obj_priv = work->obj->driver_private;
+	obj_priv = work->pending_flip_obj->driver_private;
 
 	/* Initial scanout buffer will have a 0 pending flip count */
 	if ((atomic_read(&obj_priv->pending_flip) == 0) ||
@@ -4060,7 +4061,7 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc,
 	work->event = event;
 	work->dev = crtc->dev;
 	intel_fb = to_intel_framebuffer(crtc->fb);
-	work->obj = intel_fb->obj;
+	work->old_fb_obj = intel_fb->obj;
 	INIT_WORK(&work->work, intel_unpin_work_fn);
 
 	/* We borrow the event spin lock for protecting unpin_work */
@@ -4089,13 +4090,14 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc,
 	}
 
 	/* Reference the old fb object for the scheduled work. */
-	drm_gem_object_reference(work->obj);
+	drm_gem_object_reference(work->old_fb_obj);
 
 	crtc->fb = fb;
 	i915_gem_object_flush_write_domain(obj);
 	drm_vblank_get(dev, intel_crtc->pipe);
 	obj_priv = obj->driver_private;
 	atomic_inc(&obj_priv->pending_flip);
+	work->pending_flip_obj = obj;
 
 	BEGIN_LP_RING(4);
 	OUT_RING(MI_DISPLAY_FLIP |