Serial number in Device Tree

[notro]

Currently the serial number is passed in 2 ways by the bootloader:

• Kernel commandline as text: bcm270{8,9}.serial= (used by bcm2708/9)
• Device Tree in binary form: /system/linux,serial (used by downstream bcm2835)

This is the only in-kernel user of this information:

$ cat /proc/cpuinfo | grep Serial
Serial          : 00000000316e5228

Linux 4.2 introduced a way to pass in the serial number via Device Tree as a string:
Commit: torvalds/linux@3f59987
Doc: torvalds/linux@13dd92b

/ {
    serial-number = "00000000316e5228";
};

@pelwell can we have it a third way? :-)

[pelwell]

Sure - I've dropped a patch in the pipe.

[popcornmix]

This is in latest rpi-update firmware.

[notro]

Thanks, that worked just fine.

[kukabu]

@popcornmix what about implement local-mac-address in DT? We can use the c489565 and drop the 1bc5568

[kukabu]

WOW! The mac-address exists in DT, it seems we can drop the 1bc5568.
But can you rename the mac-address to local-mac-address? The mac-address is better use for user-defined whereas local-mac-address - vendor-defined.

[pelwell]

The of_get_mac_address() function tries "mac-address" first, then "local-mac-address", then "address", so using "mac-address" is actually a performance optimisation.

But seriously, can you think of a situation where defining "local-mac-address" instead is of some practical benefit?

[kukabu]

doc says:

• local-mac-address: array of 6 bytes, specifies the MAC address that was
  assigned to the network device;
• mac-address: array of 6 bytes, specifies the MAC address that was last used by
  the boot program; should be used in cases where the MAC address assigned to
  the device by the boot program is different from the "local-mac-address"
  property;

I think a one 'if' doesn't significantly degrade performance but the user may be able to change the mac address via DT.

[pelwell]

I don't see how it is particularly useful retaining the original MAC address (except perhaps for information), but I agree that our usage doesn't really match the spec.

Create a new issue for tracking purposes (one problem per issue, please) and I'll change it (assuming there are no problems).