Re: nf_conntrack [was Re: [PATCH 1/4] RFC: fast string matching infrastrure for netfilter]

[Rusty Russell <rusty@rustcorp.com.au>]

On Fri, 2005-01-14 at 09:37 +0100, Harald Welte wrote:
> On Fri, Jan 14, 2005 at 06:01:42PM +1100, Rusty Russell wrote:
> 
> > It was IRC, and for IPv6 I agree with Harald that it's probably not the
> > right mechanism.  I prefer a simple discriminated union and two slab
> > caches (IPv4 and IPv6), which is hardcoded, but still fairly easy to
> > read.
> 
> so what about the nat_info ?  If you load iptable_nat, it will be needed
> for all connections...

My current patches reduce the nat_info to one list_head.  If we were to
use a hash tree as per Gandalf's recent work, it'd be 0.  IMHO it's not
worth the pain even if it's still a list_head.

> > Does that mean I should *not* send the expectation and NAT patches to
> > DaveM now?  (Posted before the ct_extend patches):
> 
> I was wondering all the time why we keep talking about maintainance mode
> for ip_conntrack and the coincidential spike in new development for
> ip_conntrack.

That's why I asked.  Unfortunately, I only just found time to do some
netfilter work 8(

I definitely think ct_extend should wait for nf_conntrack (but I wanted
the patches out there, since I think they're complimentary).  For the
others, I think it's better to merge them now so nf_conntrack only has
to sync once (presumably it will have to sync with some of the existing
changes already), and plan nf_conntrack for 2.6.11.

Rusty.
-- 
A bad analogy is like a leaky screwdriver -- Richard Braakman