On Fri, Oct 11, 2013 at 06:22:55PM +0200, Daniel Vetter wrote: > On Fri, Oct 11, 2013 at 11:14 AM, Chris Wilson <ch...@chris-wilson.co.uk> > wrote: > > > > void > > @@ -4711,11 +4707,9 @@ i915_gem_lastclose(struct drm_device *dev) > > if (drm_core_check_feature(dev, DRIVER_MODESET)) > > return; > > > > - mutex_lock(&dev->struct_mutex); > > - ret = i915_gem_idle(dev); > > + ret = i915_gem_idle(dev, false); > > I think this here should also set disable=true, in case X died > prematurely and forgot to do the leavevt dance. At that point I think > we could rip out that disable parameter again and just set it when in > ums mode. Also this change should be in a separate patch, at least I > don't understand why we need to change it here together with the > untangling of the gpu_idle/gem_idle confusion.
Seems reasonable, I just wanted to mimic the existing behaviour closely. I would also rename i915_gem_idle() then, perhaps i915_gem_freeze(), _suspend(), _disable(), _plug() or _stop() -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/intel-gfx
