Date: Sat, 12 Mar 2005 17:55:09 +0100 From: Andre Oppermann <andre@freebsd.org> To: Sam Leffler <sam@errno.com> Cc: net@freebsd.org Subject: Re: Giant-free polling [PATCH] Message-ID: <42331EED.D7714E05@freebsd.org> References: <20050311110234.GA87255@cell.sick.ru> <E1D9kbt-000FAj-00._pppp-mail-ru@f22.mail.ru> <20050311141450.GF9291@darkness.comp.waw.pl> <42320A3E.1020708@elischer.org><42322875.4030404@errno.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Sam Leffler wrote: > > Pawel Jakub Dawidek wrote: > > On Fri, Mar 11, 2005 at 01:14:38PM -0800, Julian Elischer wrote: > > +> >P> There is still an unresolved problem (in your and our patch as well) of > > +> >P> using ifnet structure fields without synchronization, as we don't have > > +> >P> access tointerface's internal mutex, which protects those fields. > > +> > > > +> > > > +> > > +> you need to add an interface method that has access to it.. > > > > I was thinking more about moving interface mutex into ifnet structure, > > but Robert has some objections IIRC. > > > > I don't know what Robert's objections are but I've considered doing it > for a while to deal with some locking issues in net80211-based drivers. > The only issue I can see is if this mutex boxes drivers into a locking > model that interlocks the rx+tx paths. We don't want this. This would paint us into a corner with modern high speed hardware that can hanle the rx+tx paths simulaneously. Depending on the hardware DMA model and driver architecture you want to have a different locking model. I agree with Robert in objecting to this. -- Andre
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?42331EED.D7714E05>