Re: the right spot for ifa_add...

2010-10-23 Thread Kenneth R Westerback
On Sat, Oct 23, 2010 at 01:33:51PM +0200, Henning Brauer wrote: > I received a fair number of positive test reports and would like to > commit this, aka looking for oks. > ok krw@ Ken > * Henning Brauer [2010-09-28 11:48]: > > hello? > > > > * Henning Brauer [2010-09-24 10:35]: > > > I'

Re: the right spot for ifa_add...

2010-10-23 Thread Henning Brauer
I received a fair number of positive test reports and would like to commit this, aka looking for oks. * Henning Brauer [2010-09-28 11:48]: > hello? > > * Henning Brauer [2010-09-24 10:35]: > > I'd really like more and heacy testing on this. has the potential for > > a great speedup on boxes wit

Re: the right spot for ifa_add...

2010-09-28 Thread Henning Brauer
hello? * Henning Brauer [2010-09-24 10:35]: > I'd really like more and heacy testing on this. has the potential for > a great speedup on boxes with many IPs. > > * Henning Brauer [2010-09-23 13:51]: > > oups, one superfluous check forgotten to remove. > > > > note to self: in main tree on anak

Re: the right spot for ifa_add...

2010-09-24 Thread Henning Brauer
I'd really like more and heacy testing on this. has the potential for a great speedup on boxes with many IPs. * Henning Brauer [2010-09-23 13:51]: > oups, one superfluous check forgotten to remove. > > note to self: in main tree on anakin (ryan sez it's the next level of > krautcomputing) > >

Re: the right spot for ifa_add...

2010-09-23 Thread Stuart Henderson
On 2010/09/23 13:44, Henning Brauer wrote: > oups, one superfluous check forgotten to remove. > > note to self: in main tree on anakin (ryan sez it's the next level of > krautcomputing) works so far, including multicast.

Re: the right spot for ifa_add...

2010-09-23 Thread Henning Brauer
oups, one superfluous check forgotten to remove. note to self: in main tree on anakin (ryan sez it's the next level of krautcomputing) Index: net/if.c === RCS file: /cvs/src/sys/net/if.c,v retrieving revision 1.225 diff -u -p -r1.22

Re: the right spot for ifa_add...

2010-09-23 Thread Henning Brauer
and if you feel truly adventurous you can try this, which maintains and uses the rb tree. known broken with ipvshit router advertisement shit until that shit is fixe^Wmade a little less sucky. did i mention shit yet? has a little debug goo in it: if you ifconfig destroy a non-existant but destroya

Re: the right spot for ifa_add...

2010-09-22 Thread Henning Brauer
sigh. whoever wrote that co^Wmess^Wdesaster better never gets close to me, physically. the below doesn't break on 0.0.0.0 any more. problem found by sthen with pppoe. note to self: diff is in the main tree on luke Index: net/if.c =

the right spot for ifa_add...

2010-09-22 Thread Henning Brauer
so, eventually i want to move us to use an rb tree for interface addresses instead of a stupid list which is linerily walked. the basics have been discussed, there is no downside for the "one real interface with one real IP" usage case, but the "lots of addresses in the system" case would eventuall