On Fri, Apr 07, 2017 at 09:18:17AM -0000, Patchwork wrote: > == Series Details == > > Series: drm/i915: Don't call synchronize_rcu_expedited under struct_mutex > URL : https://patchwork.freedesktop.org/series/22646/ > State : success > > == Summary == > > Series 22646v1 drm/i915: Don't call synchronize_rcu_expedited under struct_mutex > https://patchwork.freedesktop.org/api/1.0/series/22646/revisions/1/mbox/ Slow mbox. So reply here instead. Hmm, so that fixes the immediate problem with struct_mutex, but there is still a danger that we are underneath some other mutex that is being waited upon. I think it is too dangerous to live and that kswapd should have the role of expediting the RCU cleanup. Since the patch improves the situation, Reviewed-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> But please do follow up with a - synchronize_rcu_expedited(); + /* We rely on the system flushing the RCU slabs to free our structs */ -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
