Hi Everyone, This is a follow-up for the discussion on the patchset at https://lore.kernel.org/linux-pm/2314745.iZASKD2KPV@xxxxxxxxxxxxx/ that suggested an adjustment of the approach. Accordingly, this series modifies the PM core and the users of DPM_FLAG_SMART_SUSPEND to take it into account only if it is consistently used in dependency graphs, as described in the changelog of patch [3/3]. Patches [1-2/3] are preparatory and they arrange for the handling of devices with no runtime PM support in a meaningful way. Thanks!