Date: Wed, 15 Apr 2015 10:44:58 -0700 From: Nick Rogers <ncrogers@gmail.com> To: Gleb Smirnoff <glebius@freebsd.org> Cc: Luigi Rizzo <rizzo@iet.unipi.it>, "freebsd-net@freebsd.org" <net@freebsd.org> Subject: Re: moving ALTQ out of contrib Message-ID: <CAKOb=Ybox7XJT%2BVPfKqWdTPfMWStedZ-7tZNbD_u0fr2mQDa1Q@mail.gmail.com> In-Reply-To: <20150415122627.GZ883@glebius.int.ru> References: <20150414135346.GU883@FreeBSD.org> <20150415073823.GA94402@onelab2.iet.unipi.it> <20150415122627.GZ883@glebius.int.ru>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Apr 15, 2015 at 5:26 AM, Gleb Smirnoff <glebius@freebsd.org> wrote: > On Wed, Apr 15, 2015 at 09:38:23AM +0200, Luigi Rizzo wrote: > L> > With the new ifnet KPI, that is now being developed in > projects/ifnet, > L> > the ALTQ will need some tweaking. It is discontinued by initial author > L> > for a decade now, and it has already experienced direct commits in > L> > our tree. Thus, I see no good reasons to continue keeping it in > contrib. > L> > In NetBSD they have it in sys/altq. > L> > > L> > I'd prefer to move it to sys/net/altq. > L> > > L> > Any objections or better ideas? > L> > L> my first question is what is the expected residual lifetime of altq ? > > If I get it working properly in projects/ifnet, I see no reasons to > remove it. It is going to be a plugin into network stack and will no > longer require editing drivers. It will run on drivers that aren't > supported by ALTQ now. However in the latter case the ALTQ will sit > on top of interface own queue, and will start to work only when > interface's own queue overflows. But if we later add a new interface > method to modify length of own queue at runtime, this issue will > go away. > For what its worth, I maintain 300+ systems acting as PF+ALTQ routers, among other things. I've been doing this for the last 8 years or so (since 7.2 and now running 10.1), and have dealt with all the issues and workarounds surrounding lack of multiqueue support and driver compatibility, but at the end of the day ALTQ still does the job despite its performance issues. I would be in bad shape if it were simply removed at some point in the future. The changes you propose to not only keep it around officially, but make it more driver-agnostic and possibly behave better with multiqueue, sound absolutely fantastic. Thank you. > L> If it is destined to be removed soon (and probably that is not > L> unlikely given its unmaintained state, the absence of multiqueue > L> support etc.) maybe we could live for the next > L> couple of years just leaving it where it is now and avoid the > L> repo churn. > L> > L> If we really plan to relocate the code, I guess the options are > L> > L> sys/altq as in netbsd > L> > L> sys/netaltq this would be an alternative location to > L> the above one, justified by the fact that > L> we have already a bunch of net* subdirs > L> > L> sys/net/altq as you propose, i guess to stay close to > L> the rest of the ifnet code (and perhaps > L> as a first step in cleaning up sys/net > L> by putting stuff in various subdirs) > L> > L> In any case my preference would really be to leave it where it is. > > I don't like to keep in contrib a code maintained and edited by the > project. Especially I don't like tautological path of contrib/altq/altq. > I don't like extra glue in Makefiles, especially modyfing CFLAGS for the > whole kernel build. > I'm not a major committer other than the occasional patch submission for various things, but the contrib/altq location has always perplexed me. I don't see it as a contribution. > If it is a regular piece of kernel code, let it be like the rest of > kernel code. > > -- > Totus tuus, Glebius. > _______________________________________________ > freebsd-net@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-net > To unsubscribe, send any mail to "freebsd-net-unsubscribe@freebsd.org" >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAKOb=Ybox7XJT%2BVPfKqWdTPfMWStedZ-7tZNbD_u0fr2mQDa1Q>