Hi, On 01/08/2014 02:19 AM, Ma Haijun wrote:
Hi, I think there is no much reason to use dummy clock names. It is better to access them by index, and use clock-names for documentation only. Then it can support arbitrary number of clocks easily.num_clks = of_count_phandle_with_args(np, "clocks", "#clock-cells"); for ( id = 0; id < num_clks; id++) { clk[id] = of_clk_get(np, id); }
Except that we will want to use a devm version of of_clk_get, which means adding a new devm_get_clk_by_index function. Certainly doable. But the same argument could be made for the phy, so that means adding a devm_get_phy_by_index oh wait maybe we also want to add support for regulators for some boards, so that is a devm_get_regulator_by_index, oh but it is an optional regulator so we want devm_get_regulator_optional_by_index. So this will lead to an explosion of devm_get_foo_bar_by_index functions additions which I want to avoid. TL;DR: 1) We want to use devm_get_foo functions 2) All devm_get_foo functions take an (often mandatory) name argument 3) We don't know the exact clk names since we're trying to do a generic driver 4) Therefor we use generic clock-names: "clk0", "clk1", etc. Regards, Hans -- You received this message because you are subscribed to the Google Groups "linux-sunxi" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/groups/opt_out.
