It was <2020-10-01 czw 19:45>, when Florian Fainelli wrote: > On 10/1/2020 7:05 PM, David Miller wrote: >> From: Łukasz Stelmach >> Date: Wed, 30 Sep 2020 16:25:25 +0200 >> >>> Use more generic eth_platform_get_mac_address() which can get a MAC >>> address from other than DT platform specific sources too. Check if the >>> obtained address is valid. >>> >>> Signed-off-by: Łukasz Stelmach >> >> Failure to probe a MAC address should result in the selection of a >> random one. This way, the interface still comes up and is usable >> even when the MAC address fails to be probed. > > True, however this behavior is not changed after this patch is > applied, I would argue that this should be a separate patch. In both drivers the second to last line of the *_init_mac_address()[1][2] functions is eth_hw_addr_random(dev->net); and I haven't changed that. My patches enable getting a MAC address via arch/platform specific function[3] arch_get_platform_mac_address() if of_get_mac_address() fails. [1] https://elixir.bootlin.com/linux/v5.8/source/drivers/net/usb/smsc75xx.c#L758 [2] https://elixir.bootlin.com/linux/v5.8/source/drivers/net/usb/smsc95xx.c#L902 [3] https://elixir.bootlin.com/linux/v5.8/source/net/ethernet/eth.c#L502 -- Łukasz Stelmach Samsung R&D Institute Poland Samsung Electronics