Date: Thu, 21 Feb 2008 13:56:40 -0800 From: Sam Leffler <sam@errno.com> To: =?ISO-8859-1?Q?Dag-Erling_Sm=F8rgrav?= <des@des.no> Cc: arch@freebsd.org Subject: Re: dev.* analogue for interfaces Message-ID: <47BDF398.3060108@errno.com> In-Reply-To: <868x1eja16.fsf@ds4.des.no> References: <86odacc04t.fsf@ds4.des.no> <20080221072410.GC96595@funkthat.com> <86ablua4pk.fsf@ds4.des.no> <20080221193217.GF96595@funkthat.com> <868x1eja16.fsf@ds4.des.no>
next in thread | previous in thread | raw e-mail | index | archive | help
Dag-Erling Smørgrav wrote: > John-Mark Gurney <jmg@funkthat.com> writes: >> My case is perfectly clear. We already have dev.* for this, and you >> want to add a second, confusing, place to put similar/same information... >> Yes, this is specific for network interfaces, but what makes a network >> interface special that it's configuration can't live in dev.*? You >> stated that you were fine w/ some items being in dev.* and others in >> net.if.* for the same device, which is why I objected. > > If you can't tell the difference between a struct ifnet and a device_t, > I'm afraid we're going to have to agree to disagree. > > DES I think you need to experiment with this before you push a proposal. In net80211 I've had parallel net.wlan.X tree's that are companion to dev.* tree's and it's worked out ok but mostly because there is a clear layering/distinction between the two. I believe the original motivation for this was for s/w only devices that don't otherwise have a dev.* entry. I recently handled something like this for the cryptosoft driver by arbitrarily attaching it to nexus and it worked out very well. I personally would just attach these other devices under net. as that's existing practice but I'm open to your suggestion. Sam
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?47BDF398.3060108>