[Dnsmasq-discuss] domain-search option "." (Re: Announce: dnsmasq-2.55)

[Simon Kelley]

clemens fischer wrote:
> Simon Kelley wrote:
> 
>> That's what I would expect: a single zero byte is the correct
>> representation of the the domain name "." in the encoding used, which
>> is <count> characters <count> characters <zero>, so
>>
>> thekelleys.org.uk
>>
>> is <10> thekelleys <3> org <2> uk <0>
>>
>> The important thing is, does your DHCP client do the right thing with
>> this?
> 
> I think so.  Up until yesterday, a windows client would query for names
> that could not possibly reside in the assigned home domain, looking like
> ordinary third-party resources.  It happened not often, but during every
> session.  Today this didn't happen at all.  Also, tcpdump marks BOOTP
> options looking obviously wrong, which didn't happen for a while.
> 
> I'm not sure about this one:
> 
>     FQDN Option 81, length 18: [SO] 255/255 "xxx.aaaa.bb.ccc"
> 
> (from tcpdump).  The length seems odd.  The actual name is obfuscated,
> but not the length of the labels.

That's fine: there are three bytes of binary data before the name 
(decoded by tcpdump as [SO] 255/255), the IETF seems to think it's a 
good idea to invent a new, tricky, encoding for each new option if they 
can: see RFC 4702 for the details of this one.


Cheers,

Simon.