Date: Sat, 23 Jun 2007 00:44:44 +0100 From: "Bruce M. Simpson" <bms@icir.org> To: Julian Elischer <julian@elischer.org> Cc: FreeBSD Net <freebsd-net@freebsd.org>, Marko Zec <zec@icir.org>, releng@freebsd.org, ambrisko@FreeBSD.org Subject: Re: Vimage virtual networking and 7.0 Message-ID: <467C5EEC.1000208@icir.org> In-Reply-To: <467C1E3C.1020203@elischer.org> References: <467C1E3C.1020203@elischer.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Julian Elischer wrote: > > In the future I am hoping to be able to use vimage in our products. > They are based at the moment on 6.1, but I can see in a year they will > be based on 7.x. > > Patches for 7.0 and vimage are currently available in perforce. > What I would like to see is if there are any parts of that patch that > would allow us to make adding of vimage to 7.1 an easier task. > > For example, Anything that would prevent vimage from > needing an API change that would prevent it from being added later. My concern is that this may have already happened. I've been trying to do my bit as the years edge on to clean up the networking stack and fix bugs. One of my concerns is that the vimage change, which attempts to take network stack globals and wrap them into one big structure, may intrude on this or be subject to bitrot due to other development. > > I am quite disappointed that despite Marko's best efforts, we miss the > 7.0 > release but if it can be made nonintrusive enough I'd really like to > see if it can get in 7.1. > I appreciate all the hard work Marko has done on this, though I wonder if even 7.1 is ambitious. > Personally, if I were "god" I'd put it in now because it can be > compiled out. > and it wouldn't be compiled by default.Maybe only just bits of it.. > for sure I want the ability to have many routing tables. > and I'm not thrilled about the requirement to have my own patch sets > for this and thus not allowing others to use this feature. I think there are deeper issues in the network stack overall which need to be addressed, such as our lack of support for multipathing, scoped addresses, and all the tidyups which need to happen in struct ifnet to deal with this. My concern is that vimage may be a very intrusive change indeed where these matters are concerned, unless the vimage patches are being kept up-to-date and regression tested as issues are resolved and new features added. BMS
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?467C5EEC.1000208>