Viresh,
On Fri, May 16, 2014 at 8:26 AM, Viresh Kumar viresh.kumar@linaro.org wrote:
On 16 May 2014 20:50, Doug Anderson dianders@chromium.org wrote:
Right, so I think on exynos no functionality will be broken once Thomas's cpufreq-cpu0 change lands (udelay will only run long, never short). ...but from the purist standpoint we will be transitioning from 1.6 GHz => 800 MHz => 1.7 GHz without any notification about the 800 MHz. You could imagine someone registering for cpufreq notifications that would care about the 800MHz transition.
...so it seems like we could wait for Thomas's patches to land as-is (since they make things better) and then atop that see about adding support for intermediate frequencies to cpufreq-cpu0.
Hmm, don't know. I think these patches aren't aimed at solving exynos's problem but rather a general solution which must have already been there.
If some platform can work without it then its fine, but otherwise they should use it, even if udelay does work for them..
So, I would propose to go ahead with these patches in linux-next and lets see who all would use it.
Ah. I wasn't suggesting to wait on your patches. I think it's fine to get your patches landed and to get Thomas's patches landed (without actually intermediate frequencies). ...and then both sets have landed then we can modify cpufreq-cpu0 / exynos to actually use the intermediate freq.
-Doug