On 31/05/2022 20:19, Radhey Shyam Pandey wrote:

quoted

-----Original Message-----
From: Krzysztof Kozlowski <redacted>
Sent: Tuesday, May 31, 2022 12:40 AM
To: Radhey Shyam Pandey <redacted>; Radhey Shyam Pandey
[off-list ref]; davem@davemloft.net;
edumazet@google.com; kuba@kernel.org; pabeni@redhat.com;
robh+dt@kernel.org; krzysztof.kozlowski+dt@linaro.org;
harini.katakam@amd.com; michal.simek@amd.com
Cc: netdev@vger.kernel.org; devicetree@vger.kernel.org; linux-
kernel@vger.kernel.org; git@amd.com
Subject: Re: [PATCH net-next] dt-bindings: net: xilinx: document xilinx emaclite
driver binding

On 30/05/2022 15:21, Radhey Shyam Pandey wrote:

quoted

quoted

quoted

+        local-mac-address = [00 0a 35 00 00 00];

Each device should get it's own MAC address, right? I understand you
leave it for bootloader, then just fill it with 0.

The emaclite driver uses of_get_ethdev_address() to get mac from DT.
i.e  'local-mac-address' if present in DT it will be read and this MAC
address is programmed in the MAC core. So I think it's ok to have a
user defined mac-address (instead of 0s) here in DT example?

And you want to program the same MAC address in every device in the world?
How would that work?

I agree, for most of practical usecases mac address will be set by bootloader[1].
But just thinking for usecases where uboot can't read from non-volatile memory
user are still provided with option to set local-mac-address in DT and let linux
also configures it? Also see this in couple of other networking driver examples. 

Which is not necessarily correct approach

cdns,macb.yaml:  local-mac-address: true
cdns,macb.yaml:            local-mac-address = [3a 0e 03 04 05 06];
brcm,systemport.yaml:        local-mac-address = [ 00 11 22 33 44 55 ];

This brcm looks like an invalid MAC, so could be just placeholder.

In emaclite yaml - as it an example I will set it mac-address to 0 to align 
with common usecase.  local-mac-address = [00 00 00 00 00 00]

Thanks.


Best regards,
Krzysztof