Hi, could anyone tell me if a new wifi driver needs to add a DT binding if it optionally retrieves the MAC address of the device using of_get_mac_address, as an alternative to chip EFUSE (which is known to contain invalid data on some devices)? The specific case, if relevant: I have a patch set to add support for the Realtek RTL8723CS SDIO wifi chip to the existing rtw88 driver, currently in review on linux-wireless [1]. The chip is used in the Pinephone, and in all samples I've seen the MAC EFUSE is six bytes of 0xff. I don't know if this affects other devices using the same chip. rtw88 falls back to a random MAC if the EFUSE value is invalid. of_get_mac_address lets me get a stable address if the bootloader provides one, which is obviously preferable (users can easily randomize the address if they prefer, the other way around it's more difficult). Is adding a DT binding required here? I'm asking because the wireless wiki tells me [2]:
For submitting a new wireless driver the requirements are: [...] * document Device Tree usage in devicetree bindings and review them with DT maintainers
Thank you and kind regards, Fiona Klute [1] https://lore.kernel.org/linux-wireless/20240202121050.977223-1-fiona.klute@xxxxxx/T/ [2] https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches#new_driver