Hi, The original purpose of this set was as stated below: On Wednesday, September 6, 2017 1:41:37 AM CEST Rafael J. Wysocki wrote: > On Monday, September 4, 2017 1:05:57 AM CEST Rafael J. Wysocki wrote: > > Hi, > > An update related to the dependency on intel-lpss. > > > The point here is to avoid runtime resuming i2c designware devices during > > system suspend in the driver's ->suspend callback in the case when the > > ACPI PM domain is not present. That will allow us to deal with the > > ACPI PM domain case going forward, among other things. > > > > The first patch cleans up the runtime PM handling in the i2c-designware-platdrv > > _probe() routine so as to make it enable runtime PM in all cases, but prevent > > the device from being runtime suspended via pm_runtime_forbid() if > > pm_disabled is set. > > The second one pushed the intel-lpss system suspend/resume callbacks to the > late/early stages of suspend/resume, respectively, so that the i2c-designware-platdrv > ones can be pushed too. > > > The second one get rids of some ugly code and makes the PM callbacks of the > > driver handle runtime-suspended devices during system suspend/resume. > > This is the third one and it does less than it did before now. > In the meantime, though, Rajat reported that there actually were problems with i2c-designware being suspended too early (as in some cases dependent device drivers attempted to use i2c during the late suspend phase of system suspend), so here it goes again as a fix. Patches [1-2/3] have not changed since v3 (other than by acquiring tags) and patch [3/3] has only been rebased on top of 4.14-rc1 (and has acquired some tags too). Thanks, Rafael
