[PATCH] net: dwc_eth_qos: add support of device tree configuration for reset delay

[Patrick DELAUNAY]

Hi,

On 4/9/21 2:22 PM, Marek Vasut wrote:
> On 4/9/21 10:00 AM, Patrick Delaunay wrote:
>> The gpio reset assert/deassert delay are today harcoded in U-Boot driver
>> but the value should be read from DT.
>>
>> STM32 use the generic binding defined in linux:
>> Documentation/devicetree/bindings/net/ethernet-phy.yaml
>>
>>    reset-gpios:
>>      maxItems: 1
>>      description:
>>        The GPIO phandle and specifier for the PHY reset signal.
>>
>>    reset-assert-us:
>>      description:
>>        Delay after the reset was asserted in microseconds. If this
>>        property is missing the delay will be skipped.
>>
>>    reset-deassert-us:
>>      description:
>>        Delay after the reset was deasserted in microseconds. If
>>        this property is missing the delay will be skipped.
>>
>> See also U-Boot: doc/device-tree-bindings/net/phy.txt
>
> Since this is a PHY property, shouldn't that be handled in 
> drivers/net/phy/ instead of MAC driver ?


I was my first idea but I don't found found the correct location in phy 
(driver or uclass)

to manage these generic property and the generic property "reset-gpios" 
was already

managed in eth driver, so I continue to patch the driver.


But I come back to this idea after your remark....

=> in linux these property are managed in

     drivers/net/mdio/of_mdio.c::of_mdiobus_phy_device_register

         parse DT node and add info in mdio

     drivers/net/phy/mdio_device.c::mdio_device_reset

         called in  mdio_probe / mdio_remove


In my first search, I don't found the same level in the U-Boot drivers 
in drivers/net/phy/

but I miss the uclass defined in drivers/net/eth-phy-uclass.c


Finally I think I need to manage the generic binding property

(reset-gpios, reset-assert-us, reset-deassert-us) directly  in

=> drivers/net/mdio-uclass


The GPIO RESET will be managed in mdio  ops: pre_probe/ post_remove

as it is done in linux

warning: today post_remove ops don't exit in u-class.


Do you think it is the correct location ?


Do you think it should be a new serie (migrate the eqos property in mdio)

after this eqos is accepted

or I shoudl sent a new serie to replace this serie.


Regards


Patrick