On Tue, Jan 31, 2006 at 03:06:09AM +0100, Quentin Garnier wrote: > On Tue, Jan 31, 2006 at 11:24:01AM +1100, Daniel Carosone wrote: > > On Tue, Jan 31, 2006 at 01:16:13AM +0100, Juan RP wrote: > > > On Tue, 31 Jan 2006 09:26:38 +1100 > > > Daniel Carosone <dan%geek.com.au@localhost> wrote: > > > > > > > How hard would it be to make these cloning-type devices, created on > > > > demand? (the devices, not the /dev nodes, that's another bikeshed :) > > > > > > Do you mean a device clonning as was implemented for bpf(4) or tap(4) > > > or are you talking about something else? > > > > Yes, just like that, though perhaps a closer implementation analogue > > would be the way that sd(4) instances or wedges can come and go > > dynamically, just because its disks and not network devices. > > However the kernel creates the instances, you still need, for vnd, a > userland entrypoint to send the 'create' command. That's what cloning > devices are for. > > It shouldn't be too hard to do; I might have time to look at it this > week, but I wouldn't mind if someone else had a try. tap(4) should have > all the necessary logic. The only thing to take care of is providing an > API so that vnconfig can tell the user which device it has created, and > that's exactly how the cloning device of tap works. It is different from > bpf(4) in that respect, because bpf is "open once for each use", i.e. the > created device doesn't last longer that the time it is opened. So, I've thought about that a bit more. I don't think we need a cloning device here. We just need vnd(4) to create vnd<N> when the user does 'vnconfig -c vnd<N>'. -- Quentin Garnier - cube%cubidou.net@localhost - cube%NetBSD.org@localhost "When I find the controls, I'll go where I like, I'll know where I want to be, but maybe for now I'll stay right here on a silent sea." KT Tunstall, Silent Sea, Eye to the Telescope, 2004.
Description: PGP signature