dhcpcd-discuss

On 26/04/2018 09:39, Mattieu Baptiste wrote:
> > The attached patch should work.
> > I'm not in a position to test it yet, maybe you could?
>
> I spent a lot of time since a few days trying to modify dhcpcd's
> source code with these kind of changes:
> - changing the values of different options to try to mimic the original request,
> - changing the order of the options.
>
> The only time I managed to get a reply from the server is when the
> authentication option is somewhere before the first ~160 bytes ; ie
> not the last option of the request.
>
> This is totally silly. Seems the server in question is this beast:
> https://infoproducts.alcatel-lucent.com/html/0_add-h-f/93-0267-HTML/7X50_Advanced_Configuration_Guide/ESMv6_IPoE_dual_stack_hosts.html#

Well, silly or not, dhcpcd should still strive to work.
The only issue I have with re-arranging the options is if another server 
down the road wants options in a different order.

Interesting note from RFC3315:
   Any DHCP message MUST NOT include more than one Authentication
   option

So your dibbler hack isn't compliant :(
So let's ignore it.

If you make authentication the first option in the list, does it always 
work? How about after the client id being sent? Do you want me to make 
some patches to do this for you or are you ok working on this?

Whilst looking over the authentication, I did find one area on 
in-compliance in dhcpcd where RDM and replay data were not set to zero 
which I've added here:
https://roy.marples.name/git/dhcpcd.git/commit/?id=96ef771991dfd6887502ad9011562106fe245628

Roy