netdev
[Top] [All Lists]

Re: Allowing netlink_family to be any integer (was: [PATCH 2.6] iptables

To: hadi@xxxxxxxxxx
Subject: Re: Allowing netlink_family to be any integer (was: [PATCH 2.6] iptables CLUSTERIP target)
From: Evgeniy Polyakov <johnpol@xxxxxxxxxxx>
Date: Fri, 22 Oct 2004 16:32:39 +0400
Cc: Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx>, Harald Welte <laforge@xxxxxxxxxxxxx>, lmb@xxxxxxx, ahu@xxxxxxx, hch@xxxxxxxxxxxxx, netdev@xxxxxxxxxxx, "David S. Miller" <davem@xxxxxxxxxxxxx>
In-reply-to: <1098447593.1112.77.camel@jzny.localdomain>
Organization: MIPT
References: <E1CKkWZ-0005x5-00@gondolin.me.apana.org.au> <1098444579.1112.66.camel@jzny.localdomain> <20041022113939.GA812@gondor.apana.org.au> <1098447593.1112.77.camel@jzny.localdomain>
Reply-to: johnpol@xxxxxxxxxxx
Sender: netdev-bounce@xxxxxxxxxxx
On Fri, 2004-10-22 at 16:19, jamal wrote:
> On Fri, 2004-10-22 at 07:39, Herbert Xu wrote:
> > On Fri, Oct 22, 2004 at 07:29:39AM -0400, jamal wrote:
> > > 
> > > Evgeniy Polyakov(aka mr. Sean Paul) posted code a while back for
> > > something that uses netlink that he calls  "kernel Konnector". I think
> > 
> > That patch puts the ID in each message, right?
> 
> Yes, the ID is necessary for "routing" the message. Remember this is for
> a messaging subsystem so you cant avoid having something that is in the
> packet that is used to find where to go next.
> I was suggesting also messaging by name.
> 
> > That would mean paying the lookup cost for each message rather than
> > once when you create the socket.
> > 
> 
> Take a look at the patch. See how it can be made better.
> Evgeniy, Do you have something new to post?

Nothing major, only multicast group selection by id.
Although I think message sending mechanism can use unicast.
Both have it's own advantages.

> cheers,
> jamal
-- 
        Evgeniy Polyakov

Crash is better than data corruption. -- Art Grabowski

Attachment: signature.asc
Description: This is a digitally signed message part

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