Re: [PATCHv2 4/5] clk: samsung: exynos5410: Add fixed rate clocks

From: Tomasz Figa
Date: Thu Jul 31 2014 - 17:08:40 EST


Humberto,

On 31.07.2014 23:01, Humberto Naves wrote:
> Hi,
>
> On Thu, Jul 31, 2014 at 1:45 PM, Sylwester Nawrocki
> <s.nawrocki@xxxxxxxxxxx> wrote:
>> Can you explain what is rationale behind this change ? Is it related to
>> suspend/resume ordering ?
>
> I had forgotten, but now remember the reason why I did this. If you
> see the current implementation of clk-exynos5410, you will notice it
> heavily depends on the clock "fin_pll". On the other hand, this clock
> exists because in the current dtb (exynos5410-smdk5410.dts), there is
> a node fin_pll such as
>
> fin_pll: xxti {
> compatible = "fixed-clock";
> clock-frequency = <24000000>;
> clock-output-names = "fin_pll";
> #clock-cells = <0>;
> };
>
> So far so good. But the real problem comes in when I check the rate of
> fin_pll to determine if I should install the rate table or not (and I
> really need this for my patch). More specifically
>
> if (_get_rate("fin_pll") == 24 * MHZ) {
> exynos5410_plls[apll].rate_table = apll_24mhz_tbl;
> exynos5410_plls[cpll].rate_table = cpll_24mhz_tbl;
> exynos5410_plls[kpll].rate_table = kpll_24mhz_tbl;
> exynos5410_plls[dpll].rate_table = dpll_24mhz_tbl;
> exynos5410_plls[epll].rate_table = epll_24mhz_tbl;
> exynos5410_plls[ipll].rate_table = ipll_24mhz_tbl;
> }
>
> I *have* to determine if the rate of fin_pll is 24MHz, and this is
> impossible to do if fin_pll is not available. Moreover, there is no
> way I can ensure that the fixed clock provider for fin_pll was
> initialized before mine, so there is chance that _get_rate won't work.
> The only way I fix that is to set the dependency explicitly in the
> dtb, by adding the fin_pll clock as required resource.
>
> clock: clock-controller@10010000 {
> compatible = "samsung,exynos5410-clock";
> reg = <0x10010000 0x30000>;
> #clock-cells = <1>;
> /* Add the parent clock */
> clocks = <&fin_pll>;
> clock-names = "fin_pll";
> };

This is the correct solution to your problem. The clocks and clock-names
properties should have been there from the beginning but apparently this
has been missed in review. Also see below.

>
> But in any case, the bindings with the DTB must be changed one way or
> another, because I *really* need to use fin_pll on my driver
> registration.

This is a backwards compatible change. On DTBs without clocks and
clock-names properties the PLL tables simply won't be registered which
is exactly the same behavior we have now without any tables in the
driver at all.

> If you agree with this alternative solution I previously
> described, I can change that in the next version of the patch series.

Please update the dts instead, in the way you pointed above.

Best regards,
Tomasz
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/