From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Denis V. Lunev" Subject: Re: namespace acceptance process. bad news Date: Wed, 05 Dec 2007 14:22:49 +0300 Message-ID: <47568A09.7050305@sw.ru> References: <475680A2.2000907@sw.ru> <47568471.1060503@fr.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=KOI8-R Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <47568471.1060503-NmTC/0ZBporQT0dZR+AlfA@public.gmane.org> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: containers-bounces-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org Errors-To: containers-bounces-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org To: Daniel Lezcano Cc: Pavel Emelianov , "Eric W. Biederman" , Linux Containers , Alexey Kuznetsov , Benjamin Thery List-Id: containers.vger.kernel.org Daniel Lezcano wrote: > Denis V. Lunev wrote: >> Hello, All! >> >> We are completely bite to ground with the current Eric's patchset today >> by Dave Miller. flowi tagging considered wrong. The same opinion has >> been received from Alexey Kuznetsov :( >> >> So, it seems that we can't push this approach. > > Argh ! > >> >> Daniel, Benjamin, should I merge your code to our git after this news or >> we should stop a bit and think? We have talked on OLS that if Dave stop >> us with current approach we could try global context as in OpenVz. > > IMHO, doing netns switching has no sense now we are so far in the netns > implementation. > >> I think I'll code this a bit and see a reaction, but we need to have >> some agreement here :) > > I am more inclined to think about how to handle this problem before > doing anything. > > Let's try to understand why flowi tagging is considered wrong first. > > Alexey seems to disagree with this approach, is it possible to elaborate > a little bit ? > > Here is a quote from Miller: | I'm not applying this, it's going to have a negative impact on routing | performance. | | It also changes the semantics of the flowi object in a way I very | much dislike, in that there is now non-clobberable state in there. | | Previously only addressing identifying objects were present in the | flow, you could use it any context, and there were no pointer | dereferencing or object references from this thing. It was very | simple. | | That is no longer the case after your patch and I don't want us | to go down this path. | | Please find another way to implement this. flowi marking is a way to deliver the namespace into the routing code, as far as I can understand the implementation. Regards, Den