Yes I see that; and it doesn't look like the upstream changes were simple otherwise I'd try it locally. I guess I'll wait for us to sync to a later upstream.
Neil. ---- On Tue, 16 Dec 2014 00:46:05 -0800 Francois Tigeot<[email protected]> wrote ---- > On Mon, Dec 15, 2014 at 11:05:29PM -0800, kyuupichan wrote: > > 0 1143 0 1142 111 220 0 163000 51584 lwe D0L ?? > > 0:28.20 /usr/local/bin/X :7 -auth /home/neil/.serverauth.1129 (Xorg) > > 0 1143 1 1142 0 152 0 163000 51584 umtxsl S0L ?? > > 0:00.00 /usr/local/bin/X :7 -auth /home/neil/.serverauth.1129 (Xorg) > > There are two Xorg threads and both appear to be sleeping. > > The first one is waiting in a wait_event() call > > The drm/radeon driver only uses the wait_event_timeout() and > wait_event_interruptible_timeout() variants so it's unlikely to be the > culprit. > > Howewer, the drm/ttm memory manager may use the straight wait_event() > variant > without a timeout in the ttm_bo_wait_unreserved() function. > > Our ttm_bo_wait_unreserved() function is from Linux 3.9 and I see its > implementation has been changed in more recent Linux versions to not use > wait_event() anymore. > > We may have inherited an upstream bug here. > > -- > Francois Tigeot >
