[PATCH] usb: onboard-hub: Update the bind function based on "peer_hub" property

[Michal Simek]

On 12/3/24 17:24, Marek Vasut wrote:
> On 12/3/24 2:44 PM, Michal Simek wrote:
>>
>>
>> On 12/3/24 05:49, Venkatesh Yadav Abbarapu wrote:
>>> Add the bool variable "peer_hub" and set this only for the
>>> hubs which have the "peer-hub" property in their DT nodes.
>>> Skip the bind function for usb hubs which don't have "peer-hub"
>>> property.
>>>
>>> Fixes: 57e30b09fc ("usb: onboard-hub: Bail out if peer hub is already probed")
>>> Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu at amd.com>
>>> ---
>>>   common/usb_onboard_hub.c | 8 ++++++++
>>>   1 file changed, 8 insertions(+)
>>>
>>> diff --git a/common/usb_onboard_hub.c b/common/usb_onboard_hub.c
>>> index 6f28036e09..0a8fac38ee 100644
>>> --- a/common/usb_onboard_hub.c
>>> +++ b/common/usb_onboard_hub.c
>>> @@ -26,6 +26,7 @@ struct onboard_hub {
>>>   struct onboard_hub_data {
>>>       unsigned long reset_us;
>>> +    bool peer_hub;
>>>       unsigned long power_on_delay_us;
>>>       int (*init)(struct udevice *dev);
>>>   };
>>> @@ -178,10 +179,15 @@ err:
>>>   static int usb_onboard_hub_bind(struct udevice *dev)
>>>   {
>>> +    struct onboard_hub_data *data =
>>> +        (struct onboard_hub_data *)dev_get_driver_data(dev);
>>>       struct ofnode_phandle_args phandle;
>>>       const void *fdt = gd->fdt_blob;
>>>       int ret, off;
>>> +    if (!data->peer_hub)
>>> +        return 0;
>>> +
>>>       ret = dev_read_phandle_with_args(dev, "peer-hub", NULL, 0, 0, &phandle);
>>>       if (ret)  {
>>>           dev_err(dev, "peer-hub not specified\n");
>>> @@ -212,12 +218,14 @@ static int usb_onboard_hub_remove(struct udevice *dev)
>>>   }
>>>   static const struct onboard_hub_data usb2514_data = {
>>> +    .peer_hub = false,
>>>       .power_on_delay_us = 500,
>>>       .reset_us = 1,
>>>   };
>>>   static const struct onboard_hub_data usb5744_data = {
>>>       .init = usb5744_i2c_init,
>>> +    .peer_hub = true,
>>>       .power_on_delay_us = 1000,
>>>       .reset_us = 5,
>>>   };
>>
>> I don't think this is correct solution. I think if optional property peer-hub 
>> is not present in DT bind should return 0 and that should fix issue on stm 
>> board where only one hub is present.
> Shouldn't the 'peer-hub' property be mandatory for this hub ?

You can use this hub with usb2.0 only or with usb3.0 only. It means peer-hub is 
correctly optional property. It is just saying when it is there hub should be 
configured only once not twice.

It means right fix is when peer-hub property is missing bind should just pass 
because it is just hub which don't use another ID.

Thanks,
Michal