Li, Aubrey wrote: > Mark.Haywood at Sun.COM wrote: > > >> So this brings up a question that I think I've asked before >> and not seen >> an answer to. The discussion here seems to be around a C-state driver, >> not a CPU driver? Does this mean that the C-state support will be in a >> driver of its own? Or is it going to be functionality provided by the >> CPU driver? >> >> > C-state support will be in a driver similar as speedstep driver, they > both are included in the CPU driver. > The driver will parse ACPI tables, evaluate _CST object and cache its > data, > give a interface to the high level OSPM to change the c-state through io > port, > and have a policy to decide which c-state the processor should go into > in the next idle, according to > the latency read from _CST. >
In regards to the c-state driver's next idle c-state policy, we would like to be able to specify "hints" to the c-state driver to help it decide which c-state it should enter in the next idle. We are not sure what this interface will look like yet. For example the power-aware dispatcher may know it will not schedule threads on a core for a long time, so the core is free to go to a very low c-state. Thanks, Bill > Thanks, > -Aubrey > _______________________________________________ > tesla-dev mailing list > tesla-dev at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/tesla-dev >
