Interface auto-cloning bug or feature?

[Alexey Shuvaev]

On Fri, Sep 19, 2008 at 03:51:31PM -0700, Maksim Yevmenkin wrote:
> On Fri, Sep 19, 2008 at 3:43 PM, Maksim Yevmenkin
> <maksim.yevmenkin at gmail.com> wrote:
> > [....]
> >
> >>> That what has caused me to look into this issue. You can find patch for
> >>> security/vpnc to prevent unbounded interface cloning here:
> >>>
> >>> http://sobomax.sippysoft.com/~sobomax/vpnc.diff
> >>>
> >> Ok, the patch prevents interface cloning, but I think it doesn't solve
> >> the actual problem.
> >> Let's wait for Maksim :)
> >
As for vpnc-script, yes I also think using kldstat to check if the module is
present is more correct than stat-ing /dev/tun.
Actually I have commented these lines out in my previous mail instead of
applying the patch %-)

> > ok, how about attached patch. i put it together *very* quickly and
> > only gave it a light testing. its for tap(4), because i could compile
> > it as a module and tun(4) is compiled into kernel by default, but the
> > idea should identical for tun(4). should be even simpler for tun(4)
> > because it does not have to deal with 2 kind of devices (i.e. tap and
> > vmnet). give it a try, and see if it works. please try both cloning
> > paths, i.e.
> >
> > 1) cat /dev/tap (/dev/vmnet) with and/or without unit number
> >
> > and
> >
> > 2) ifconfig tapX (vmnetX) create/destroy
> >
> > in the mean time i will prepare something similar for tun(4).
> 
> attached is similar patch for tun(4). i only made sure it compiles :)
> rebuilding kernel now...
> 
Ok, I have tried both patches and in both cases 'cloning bug' is went away.
I have tried
cat /dev/tap (creates only /dev/tap0)
cat /dev/tapX (creates exactly /dev/tapX)
stat /dev/[tun|tap|vmnet] (creates only '0' device)
and also running unpatched security/vpnc.

Everything works fine!

Thanks!
Alexey.