[PATCH v2 2/2] phy: rockchip: inno-usb2: Add usb2 phys support for rk3576

frawang frawang.cn at gmail.com
Tue Sep 24 18:42:35 PDT 2024


Hi Heiko,

On 2024/9/24 18:01, Heiko Stuebner wrote:
> Am Dienstag, 24. September 2024, 10:55:10 CEST schrieb Frank Wang:
>> From: William Wu <william.wu at rock-chips.com>
>>
>> The RK3576 SoC has two independent USB2.0 PHYs, and
>> each PHY has one port.
> Can you please split the content into "converting to clk_bulk" (see
> additional comment below) and "add rk3576" please?
>
> That would make the patch a lot cleaner.

OK, I shall amend in the next patch.

>
>> @@ -376,6 +378,7 @@ rockchip_usb2phy_clk480m_register(struct rockchip_usb2phy *rphy)
>>   {
>>   	struct device_node *node = rphy->dev->of_node;
>>   	struct clk_init_data init;
>> +	struct clk *refclk = of_clk_get_by_name(node, "phyclk");
> Doesn't this create an imbalance - with the missing put?
> I think ideally just define clk_bulk_data structs for the
> 1-clock and 3-clock variant, attach that to the device-data
> and then use the regular devm_clk_bulk_get ?
>
> That way you can then retrieve the clock from that struct?

How about keep the clk_bulk_data and num_clks member in rockchip_usb2phy 
structs, and retrieve the clock by "clks.id" here?
Just like the following codes.

@@ -378,8 +378,9 @@ rockchip_usb2phy_clk480m_register(struct 
rockchip_usb2phy *rphy)
  {
         struct device_node *node = rphy->dev->of_node;
         struct clk_init_data init;
-       struct clk *refclk = of_clk_get_by_name(node, "phyclk");
+       struct clk *refclk = NULL;
         const char *clk_name;
+       int i;
         int ret = 0;

         init.flags = 0;
@@ -389,6 +390,13 @@ rockchip_usb2phy_clk480m_register(struct 
rockchip_usb2phy *rphy)
         /* optional override of the clockname */
         of_property_read_string(node, "clock-output-names", &init.name);

+       for (i = 0; i < rphy->num_clks; i++) {
+               if (!strncmp(rphy->clks[i].id, "phyclk", 6)) {
+                       refclk = rphy->clks[i].clk;
+                       break;
+               }
+       }
+

BR.
Frank

>
> Thanks
> Heiko
>



More information about the linux-phy mailing list