On Thu, Oct 25, 2012 at 10:23 AM, Lee Jones <lee.jo...@linaro.org> wrote: > On Thu, 25 Oct 2012, Linus Walleij wrote: >> >> Yeah drop it for now. > > Actually, a quick question before I do: > > If it's better/faster to prepare the clock and keep it prepared > while you do clk_enable/clk_disable,
It is generally faster that is why we call it fastpath. E.g. if the clock hardware can do this in IRQ context by just chaninging one bit in a quickly written register from 1->0 and then the clock goes off from some silicon. Whether it's "better" or not is a transcendental question, as it requires a ruler to measure betterness. > why don't we do that in all > drivers? Why do we bother to prepare/unprepare each time if all > it does is take up cycles? Usually to save power. Albeit saving power may be at odds with gaining the maximum performance and/or latency. So depending on the demands and use case the answer to whether or not you want to do this will be different. That's for the clock API. In the ux500 case specifically, you can drill down to the clock implementation and ask the question whether or not we want to do this for this instance of the pin controller in this case, I'll leave that for Ulf to answer... but remember that this driver is also used for the Nomadik NHK8815. Yours, Linus Walleij -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/