On Sat, 4 Sep 2010, Rafael J. Wysocki wrote: > OK, so in the context of my other message, I think we need these types of > helpers: > > (1) "suspend X ms from now" > (2) "suspend X ms from now if the device has been inactive for Y ms" > > where Y is set by user space (and X may be 0). > > To me, the both of them make sense. (2) is really: "starting X ms from now, wait until the device has been inactive for Y ms and then suspend". Well and good, but what about pm_runtime_suspend? Should it respect the Y ms delay? Even if the last helper called was (1)? Suppose the driver uses either helper (1) or helper (2) and then some other piece of code calls pm_runtime_get_sync followed by pm_runtime_put_sync. The get_sync will cancel the timer, but what should the put_sync do? Your helper (1) bypasses the user's requirement for Y ms of inactivity before a runtime suspend. When should the bypass end and normal behavior (i.e., using the Y-ms delay) be restored? The next time either helper is called? Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
