Re: PPTP on 3.5 GENERIC / conflicts in the documentation

[Matthew Clarke <clamat_(_at_)_telus_(_dot_)_net>]

jeudi, le 28 octobre, 2004, Jon Radel nous a dit ceci:

> I installed the pptp-1.1.0 off the i386 section of the 3.5 CD on a stock 
> 3.5 machine.  Attempting to configure it has failed so far, and various 
> documents have raised almost as many issues as they've answered.
> 
> The faq, in section 6.8, states that userland ppp is used, but I suspect 
> this is just a typo, as pptp(8) states that kernel pppd is used.  True?

No.  Supposedly, you can use either.  I know for a fact that you can use
userland ppp, because that's what I do.

> If it uses pppd, I need to recompile the kernel with ppp support turned 
> on.  True?
> 
> Why does the faq state that net.inet.gre.allow=1 is necessary when 
> pptp(8) states that the kernel should be recompiled with support for the 
>  GRE pseudo-devices removed?  Or I completely wrong in thinking that 
> net.inet.gre.allow=1 exists solely to turn the GRE pseudo-device support on?

pptp(8) is somewhat out of date with respect to current OpenBSD releases.

Older kernels needed the GRE pseudo-device compiled out.  For newer
kernels, setting that sysctl should be all you need to do.  Well, that
and, if you're using pf, making sure your packet filter rules allow GRE
packets in and out through the appropriate interface(s).

> I'd love some reassurance that if I recompile the kernel with ppp turned 
> on, GRE turned off, and get proper settings in /etc/ppp/options and 
> /etc/ppp/chap-secrets that this will work.  Or am I missing something else?

Sorry, can't give you that reassurance; I use userland ppp and the GENERIC
kernel.

Matt.
-- 
For a successful technology, reality must take precedence over public
relations, for nature cannot be fooled.
				-- Richard P. Feynman