Re: vnd.c 1.254

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Sun, Jan 17, 2016 at 08:13:46PM +0700, Robert Elz wrote:
>     Date:        Sun, 17 Jan 2016 11:52:42 +0100
>     From:        Manuel Bouyer <bouyer%antioche.eu.org@localhost>
>     Message-ID:  <20160117105242.GC1498%asim.lip6.fr@localhost>
> 
>   | This is how vnconfig -l has been working for a long time.
> 
> Not that long, only since vnd was made a cloning device, which means

I mean, vnconfig -l (without other arguments) has been showing available
devices for a long time:
osiris:/#uname -a
NetBSD osiris.lip6.fr 5.1_STABLE NetBSD 5.1_STABLE (ISIS) #0: Thu Mar  3 11:20:35 CET 2011  bouyer@swing:/home/src/src-5/src/sys/arch/i386/compile/obj/ISIS i386
osiris:/#vnconfig -l
vnd0: not in use
vnd1: not in use
vnd2: not in use
vnd3: not in use


> [...]
> 
>   | HEAD has been changed, which means that HEAD won't list free devices
>   | which exists in /dev. I think this behavior is brocken.
> 
> I disagree.  I think it should be changed further to not list any
> free devices at all.

this is a major behavior change, which may well break existing setups.
You remove existing and working functionality to fix a marginal backward
compatibility issue ? But removing this functionality is breaking
backward compat, in a much more important way.

> 
>   | We are running a vnconfig from netbsd-7.
> 
> That would do it.   Update it to what is on the netbsd-7 branch or
> the netbsd-7-0 branch.

we *are* already running an up to date vnconfig, dammit !

> 
>   | netbsd-7 vnconfig was broken to fix a problem with netbsd-7
>   | kernel+netbsd-6 userland. This is inacceptable.
> 
> No, it was the right choice, there are (and particularly were) far more
> people running NetBSD 6 than NetBSD 7.   What's more, NetBSD 7 has had a
> lot of bug fixes pulled up, there really should be a 7.0.1 release
> very soon, and the current 7.0 should be forgotten.

not until this problem is fixed. Breaking XEN3_DOM0 support is a real
problem.

> 
>   | In addition it seems that in some case, vnconfig fails to configure
>   | a free device.
> 
> If that is happening we need to find out why.   Can you add some debug to
> the script to figure out what error is occurring (perhaps removing the
> redirect to /dev/null on the vnconfig).

Unfortunably it's transient. After a view vnconfig manipulations the
problem is gone for me (and vnconfig -l again show all devices,
used or free). cd_ndevs is now at 8 (checked with gdb against /dev/mem)
unfortunably I didn't check it when it was in a brocken state

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--