drm/doc: document fallback behaviour for atomic events

Worst case if the hw can't support completion signalling in a
race-free way we want the event to be too late, not too early.

Text adapted from a proposal from Laurent - the other side of how to
make hw work correctly where it's possible is imo already sufficiently
documented.

v2: Review from Laurent.

Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170308141257.12119-7-daniel.vetter@ffwll.ch

include/drm/drm_crtc.h

@@ -204,6 +204,12 @@ struct drm_crtc_state {
 	 * drm_crtc_arm_vblank_event(). See the documentation of that function
 	 * for a detailed discussion of the constraints it needs to be used
 	 * safely.
+	 *
+	 * If the device can't notify of flip completion in a race-free way
+	 * at all, then the event should be armed just after the page flip is
+	 * committed. In the worst case the driver will send the event to
+	 * userspace one frame too late. This doesn't allow for a real atomic
+	 * update, but it should avoid tearing.
 	 */
 	struct drm_pending_vblank_event *event;