We still need a guest device. So therefore we aren't re-inventing the
wheel. We may yet end up using the new devices for the host end of things,
but before we get that far we still have to figure out how to build the
thing into plex86 to begin with.
Drew Northup, N1XIM
> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf
> Of Todd T. Fries
> Sent: Tuesday, December 19, 2000 10:24 AM
> To: [EMAIL PROTECTED]
> Subject: Re: [plex86] brain fart regarding networking
>
>
> Instead of inventing something new, why not program to an existing api?
>
> BSD has had for a long time the 'tun' and 'tap' interfaces that allow
> raw access to the ip layer and ethernet layer respectively. Aka a
> device /dev/tun0 corresponds to an network device 'tun0' .. and one
> program, /usr/sbin/ppp (no I didn't forget the 'd') can open the
> tun0 device, and open the serial port, and wallah, we have userland ppp,
> which is easier to debug than kernel land pppd+kernel combo.
>
> The point of all of this, is that linux has recently added
> bpf/tun/tap style
> devices. I'd highly recommend using them, as passing network/ip fragments
> to/from userland to/from the kernel is exactly what these types of things
> were designed for .. don't re-invent the wheel all the time ;-)
> --
> Todd Fries .. [EMAIL PROTECTED]
>
