IPv6 on Ben Nanonote

EdorFaus edorfaus at xepher.net
Sun Jun 17 21:49:50 EDT 2012


On 06/17/2012 02:17 PM, Wolfgang Spraul wrote:
>>  From what I remember, Qi doesn't have a MAC address range to assign
>> to the devices, and are using the dynamic ones on purpose.
>
> http://en.qi-hardware.com/wiki/IEEE_OUI_assignments

Ah. I stand corrected, they apparently *do* have a range they 
potentially could assign (some of) to the NanoNotes.

However, they apparently haven't done so, at least according to that 
page, which only lists MM1 assignments at the moment...

[[ TL;DR for the below: I believe we can get the desired effect, with 
less fuss and no more work for us, if we instead just configure our own 
devices to use the MAC address the wiki page [2] already shows. ]]

And the bit about the dynamic ones being used on purpose seems to also 
be supported by Xiangfu in comment 1 on issue 7 (see [1])... Though he 
also explains how to set the address to a static one in comment 2 
(including a specific address pair in the example, that I'm currently 
using on mine, though I'm not sure if that was his intention...).

There's also the problem of assigning the addresses uniquely now that 
the devices are already shipped out. I think the current owners would 
have to be involved directly, to set the address on their own device 
(since it's in the bootargs flash, not in the kernel or userland), and 
many probably wouldn't care enough for that.

Hmm. Assuming I'm correct that we technically really only need a single 
MAC address for all NanoNotes (since there can only be one NN on a 
single link-layer network), maybe Qi could assign 10-E2-D5-00-00-7F 
(which is the next free one) to Ben NanoNote, so RC4 (or whatever is 
next for MM1) would start at the round number, ..-80? ;)

That way we could just use the instructions in the wiki, or build it 
into the distros, or whatever, to use that particular address when 
setting a static one, which would keep the instructions relatively 
simple (compared to unique addresses), and be fairly easy to find (or 
point to) for those who care.

...

OK, I just did a bit more reading, and it turns out that the benefits I 
mentioned in the above paragraph do not require any assignment from Qi's 
OUI space, as we can all just use locally administered addresses with 
exactly the same effect. (Under the same assumption.)

Apparently, MAC addresses do not require the OUI; as long as the second 
least significant bit of the most significant byte of the MAC address is 
set to 1, the address is a locally administered one instead of a 
globally unique one. That means it does not include a OUI, and as such 
does not really belong to anyone, which I believe means that we are free 
to use any such address that we want to.

(Note that the least significant bit of the same byte also has special 
meaning, so we probably want to use addresses where it is 0.)

I notice that the addresses Xiangfu used in that comment (which are 
replicated on the wiki) have this bit set, meaning they are locally 
administered ones - which means he apparently knew exactly what he was 
doing when he used those particular ones. (I'm not really surprised by 
this.)

So, basically, I believe we could end up with the same desired effect if 
we all simply set the address in our own NanoNotes to the address in 
that comment - or any such, really, which might help for those with more 
than one NN - in exactly the same manner as we would have to do if they 
did assign a range and gave out unique addresses to each of us.

-- 
Regards,
Frode Austvik

[1] http://projects.qi-hardware.com/index.php/p/ben-nanonote/issues/7/
[2] http://en.qi-hardware.com/wiki/Ethernet_over_USB#Static_Mac_Address




More information about the discussion mailing list


interactive