On Friday, 6 April 2007 10:48, Johannes Berg wrote: > On Fri, 2007-04-06 at 02:17 +0200, Rafael J. Wysocki wrote: > > > > Not sure, it might be different for different suspend methods. We > > > actually need to do some platform-function stuff inbetween, and if we > > > ever want some S4-like state then we might need to do it differently. > > > > Ah, OK > > Keep in mind that I don't know that yet, and am not totally sure I ever > will implement something S4-like (it would probably require kexec or > similar tricks). Also, these handlers are not even called fro the > suspend to disk case right now (and documented that way.) > > I will repost with some BUG_ON() assertions, but should I change it to > have 4 handlers before_irq_off/after_irq_off/before_irq_on/after_irq_on > instead of the two I have now? Frankly, I'm not sure. For practical purposes the BUG_ON() assertions will suffice, so I think you can keep the two handlers. I'd change the names, though, to something like quiesce() and activate(), for example. [Hm, it feels more appropriate to define them for all platforms and make them call local_irq_save() on the platforms that don't need to do anything more.] BTW, please remember to update the SNAPSHOT_S2RAM ioctl accordingly (well, I think we should move the common code to a separate function). Greetings, Rafael _______________________________________________ linux-pm mailing list linux-pm@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/linux-pm
