On Wed, Oct 31, 2012 at 05:43:16PM +0000, Chris Wilson wrote: > On Wed, 31 Oct 2012 19:38:41 +0200, ville.syrjala at linux.intel.com wrote: > > From: Ville Syrj?l? <ville.syrjala at linux.intel.com> > > > > intel_pipe_set_base() never actually waited for any pending page flips > > on the CRTC. It looks like it tried to, by calling intel_finish_fb() on > > the current front buffer. But the pending flips were actually tracked > > in the BO of the previous front buffer, so the call to intel_finish_fb() > > never did anything useful. > > > > Now even the pending_flip counter is gone, so we should just > > use intel_crtc_wait_for_pending_flips(), but since we're already holding > > struct_mutex when we would call that function, we need another version > > of it, that itself doesn't lock struct_mutex. > > That function call is now superfluous as you pointed out in an earlier > review... I think we still need it when we're not doing a full modeset. Without it, intel_pipe_set_base() could overtake the flip. So the call could be moved outside of intel_pipe_set_base() so that it wouldn't be called for the full modeset path. But I suppose it's a bit more efficient to do it after we've pinned the new buffer, so simply moving it might not be the best idea. > But yes, not waking up the pending_flip_queue after a GPU hang is a > recent bug, and could explain the lockup in > https://bugs.freedesktop.org/show_bug.cgi?id=56337. OK. I suppose I'll cook up a quick patch for that as well. -- Ville Syrj?l? Intel OTC
