[Top] [All Lists]

Re: Move fib_alias out of fib_hash.c

To: "David S. Miller" <davem@xxxxxxxxxxxxx>
Subject: Re: Move fib_alias out of fib_hash.c
From: Robert Olsson <Robert.Olsson@xxxxxxxxxxx>
Date: Wed, 29 Sep 2004 23:30:48 +0200
Cc: Robert Olsson <Robert.Olsson@xxxxxxxxxxx>, netdev@xxxxxxxxxxx
In-reply-to: <20040929125359.12a00ba7.davem@xxxxxxxxxxxxx>
References: <20040928214722.11aef8e0.davem@xxxxxxxxxxxxx> <16730.53965.503605.943263@xxxxxxxxxxxx> <20040929125359.12a00ba7.davem@xxxxxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
David S. Miller writes:

 > You cannot find the longest matching prefix and just use that.
 > Rather, you must iterate through all matching prefixes in the
 > table from longest to shortest, trying fib_semantic_match() on
 > each one until it says OK.

 You're right we have to match semantics then first after that return 
 longest prefix to comply w. current matching.

 > This makes using a new algorithm very non-trivial.

 Yes and will take resources compared to simple longest prefix.

 > Probably what you should do is keep an array on the function
 > stack, recording shorter prefix entries you see on your walk
 > down to the longest matching prefix.  Then you process the array
 > one entry at a time back to the root, trying fib_semantic_match()
 > on each one.

 We have backtracking as we search to leaves and as leaves does not 
 always has a matching prefix.  So we have to look into how semantics 
 match can be done here instead and backtrack until we got a semantics 
 match. Was about to test w. ipv6 to start with due to ipv4 complexity.
 Anyway ipv4 fib-code is cleaner now.


<Prev in Thread] Current Thread [Next in Thread>