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

[Patrick McHardy <kaber@trash.net>]

[-- Attachment #1: Type: text/plain, Size: 337 bytes --]

Harald Welte wrote:

>>I think we should put ip_conntrack in maintenance mode, than we can
>>resync nf_conntrack and make changes like this before we submit it.
>>
>
>At last, we again agree :)
>

I've commited a TODO list for nf_conntrack (just off the top of my head)
to svn. Any further suggestions are appreciated.

Regards
Patrick


[-- Attachment #2: TODO --]
[-- Type: text/plain, Size: 716 bytes --]

TODOs for nf_conntrack (last changed Jan 14 2005)
-------------------------------------------------

Some items are controversial, so make sure to get some feedback on
netfilter-develbefore working on any of these.

- Resync with latest ip_conntrack changes
- Function/structure name unification: s/nf_conntrack/nf_ct/
- Replace get_features stuff by ct_extend ?
- Don't waste space for IPv4 tuples (currently union with IPv6 tuples) ?
- kill ->confirm, nf_conntrack_confirm is generic
- Add nf_ct_get_afinfo for things like ip_ct_attach
- Some IPv4/IPv6 registered hook functions just call generic functions, this
  means even longer call-chains. Add family argument to hook functions and
  kill them ?
- IPv4 NAT