Re: [dhcwg] "client identifier" in server replies

On Monday 07 July 2003 23:39, narasimha.nelakuditi@nokia.com wrote:
> Coming back to 'xid', I think checking 'xid' alone is not a good idea,
> considering erroneous servers, multiple clients using same xids etc.

That's right.   We hacked the ISC DHCP client to check chaddr in the reply 
because someone was in fact getting broadcast replies in high-traffic 
environments that were caused by xid collisions.   The problem is that a 
really good random number generator for xid generation is difficult to write 
in a portable way, and so we wind up using a fairly crappy one, which does 
the trick but certainly wouldn't be appropriate for doing crypto.   More to 
the point, in some environments I would expect collisions to be pretty much 
inevitable.   Collisions are a pretty serious problem, so I think the point 
that is being raised here is a serious one.

_______________________________________________
dhcwg mailing list
dhcwg@ietf.org
https://www1.ietf.org/mailman/listinfo/dhcwg