On Thu, 17 Aug 2006, Ikhwan Lee wrote: > We can have the device power manager in the kernel and let the platform > designers hardcode the rules specific to their platform. It is easier to > implement and we can use that scheme right away. > > In the future however, I think it is more clear to have a "power/voltage > domain framework, and an extensible framework for system-wide operating (and > non-operating sleep) states," quoting from what David said in his response > to my message. The second is a generalized version of the first. However the issue of the scope of a "power/voltage domain framework" is independent of the nature of the userspace API for such a framework, which is what you originally were discussing. My point was that in a large number of simple cases (like this one) there is no need for a userspace API at all. To put it another way, suppose several devices have inter-related power requirements, such as a common gated voltage source. Provided each device has only two power states (ON = "in use" and OFF = "not in use") then any reasonable framework would be able to manage them adequately even with no userspace involvement. Alan Stern
