Hugo Mills <hugo@???> wrote:
> On Sat, Nov 07, 2009 at 11:45:43AM -0000, Vic
> wrote:
>> UDP is a connectionless protocol; all that
>> matters is what's in the payload. Previous
>> packets in the datastream are irrelevant to the
>> delivery of this one.
>>
>> So the fact that it's coming from a different
>> source port to the one that the previous packet
>> went to doesn't matter; it's still a UDP
>> packet, and that doesn't depend on any
>> connection state.
That was what I expected. However, this Linux box
thinks differently.
> I could have made the same argument -- however,
> that's still not authoritative (it's merely
> highly plausible). What I was trying to say was
> that I couldn't point to the line of the RFC
> that states that it's OK(*). :)
Unfortunately, STD0006 - which was RFC0768 - is
terse, and doesn't discuss this.
However, Stevens [1] mentions that most UDP
implementations allow processes to restrict the
foreign address and port. I'll need to dig into
the source and see what it's doing.
Nick.
[1] W. Richard Stevens, "TCP/IP Illustrated, Volume
1", Addison-Wesley, 1994.
--
Nick Chalk ................. once a Radio Designer
Confidence is failing to understand the problem.
