Re: [net-next: PATCH v2 4/7] net: mvmdio: simplify clock handling

On Thu, Jun 17, 2021 at 2:25 AM Marcin Wojtas [off-list ref] wrote:

śr., 16 cze 2021 o 21:48 Andrew Lunn [off-list ref] napisał(a):

quoted

quoted

+     dev->clks[0].id = "core";
+     dev->clks[1].id = "mg";
+     dev->clks[2].id = "mg_core";
+     dev->clks[3].id = "axi";
+     ret = devm_clk_bulk_get_optional(&pdev->dev, MVMDIO_CLOCK_COUNT,
+                                      dev->clks);

Kirkwood:

                mdio: mdio-bus@72004 {
                        compatible = "marvell,orion-mdio";
                        #address-cells = <1>;
                        #size-cells = <0>;
                        reg = <0x72004 0x84>;
                        interrupts = <46>;
                        clocks = <&gate_clk 0>;
                        status = "disabled";

Does this work? There is no clock-names in DT.

Neither are the clocks in Armada 7k8k / CN913x:

                CP11X_LABEL(mdio): mdio@12a200 {
                        #address-cells = <1>;
                        #size-cells = <0>;
                        compatible = "marvell,orion-mdio";
                        reg = <0x12a200 0x10>;
                        clocks = <&CP11X_LABEL(clk) 1 9>,
<&CP11X_LABEL(clk) 1 5>,
                                 <&CP11X_LABEL(clk) 1 6>,
<&CP11X_LABEL(clk) 1 18>;
                        status = "disabled";
                };

Apparently I misread the code and got convinced that contrary to
devm_clk_get_optional(), the devm_clk_get_bulk_optional() obtains the
clocks directly by index, not name (on the tested boards, the same
clocks are enabled by the other interfaces, so the problems

Me too. Sorry for the wrong suggestion. I think we need something that
actually gets clocks by indices in a bulk, but this is another story.

I will drop this patch. Thank you for spotting the issue.

I'm fine with this.

-- 
With Best Regards,
Andy Shevchenko