On Mon, Dec 05, 2011 at 10:41:21PM +0100, Linus Walleij wrote: > Just wanted to mention that I had this discussion with Magnus Damm > about how they do this in shmobile: in there their central runtime PM > policy in arch/arm/mach-shmobile/pm_runtime.c adds in the > pm_clk_notifier from drivers/base/power/clock_ops.c to gate/ungate > the clocks centrally acting on bus notifiers without any per-driver > hooks like these.
Yeah, I'm familiar with that - some other platforms were looking into a similar scheme I think. > I clearly see that in this platform it's *not* going to work since > for this one driver atleast, there are two clocks, not just one, > that need to be managed for the device. Well, that's not something that's a blocker to central management - the core can always work with two clocks. What's more of an issue is when devices need to work with their clocks independantly of the core, but that's not insurmountable. I do also think that if we do decide to move more platforms to central management it's going to be easier to first transition all their drivers to a repetitive style of clock management and then do a big factor out once the pattern is clearly visible. It's a big undertaking for things like the Samsung family of processors where you've got a huge set of chips to work with going all the way back to s3c24xx so anything we can do to make it easier is going to be a win. ------------------------------------------------------------------------------ All the data continuously generated in your IT infrastructure contains a definitive record of customers, application performance, security threats, fraudulent activity, and more. Splunk takes this data and makes sense of it. IT sense. And common sense. http://p.sf.net/sfu/splunk-novd2d _______________________________________________ spi-devel-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/spi-devel-general
