On Wed, Dec 16, 2015 at 12:52:34AM +0200, Imre Deak wrote: > On Tue, 2015-12-15 at 21:07 +0000, Chris Wilson wrote: > > My current thinking is that the hangcheck/RPS tasks are wrong - and > > that > > we do actually have explicit wakerefs that should cover their > > lifetimes > > (but we fail to actually terminate them when we drop the associated > > wakeref). > > > > With respect to the current state (cancelling the work in > > rpm_suspend), > > the assert disabling is correct, but I think we should be indicating > > that we papering over a "bug" more strongly. > > > > i.e. something like DISABLE_RPM_WAKEREF_ASSERT(); > > But the other cases are still legitimate, so we'd keep the lower case > name for those and define the above macro as an alias simply to > emphasize the difference? Yes. If you could put it in <blink> tags that would be a bonus. -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
