[Cerowrt-devel] class-e ipv4 testing and testers?

[Dave Taht]

so for the past 2 months I've been engaged in a project with john
gilmore, rob landley and a few others to prove just how feasible it
would be to make the former
"class-e" internet address space - 240.0.0.0/4 - long reserved for
"experimentation" - actually usable.

The shocking answer has been - it is *extremely* feasible to make
these genuinely usable. Apple IOS, OSX, most Linux distributions, and
Android all can accept addresses in, and route to the 240.0.0.0/4
address space. We haven't broken any switches yet, nor crashed any
servers, clients, or routers, it works just fine through vpns...

The only major linux exception thus far was openwrt, and distributions
like that that still used "ifconfig" rather than the iproute2 suite.
They can route to, but not accept addresses in, the 240/4 range.

So... one tiny kernel patch, fixes that. That patch touches such old
code that it applies cleanly back many generations of linux.

(another openwrt patch is required if you don't rebuild your toolchain
from scratch). The other major sticking points are windows, (which
doesn't work at all)...
and another is making another small patch to all the routing daemons
of the world, and waiting a few years for those fixes to propagate.

Anyway, all the (very tiny) patches to linux, openwrt's netifd, frr
(quagga), babel, & bird are now up at:

http://www.taht.net/classe/

And: it's reached the point where we are ready to submit to kernel
mainline and all the routing daemons.

but... I'm a big believer in more eyeballs on the code,
signed-off-bys, tested-bys, and so on, so if you have a few spare
minutes in the next day or three,
 to help open up the possibility of the internet one day gaining 268
million more public  ipv4 addresses, please go play with these
patches!

The linux kernel patch (
0001-linux-kernel-Allow-class-e-address-assignment-via-ifconfig-ioctl.patch
) for openwrt
can be placed in target/linux/generic/pending-4.14/ (or 4.9 or
whichever kernel you are building for)

There's a document also in the works, writing up what we've tested so
far, outlining next steps (like asking iana/ietf to let us do a public
experiment with 255/8 and 254/8 next year), but until then...


... you'll find me at milliways.taht.net.

PS I really don't want us to get into the possibly gruesome politics
that may ensue, here, and not this week! I just want to prove this
technically feasible! We'll start up a mailing list for this at some
point soon.


-- 

Dave Täht
CTO, TekLibre, LLC
http://www.teklibre.com
Tel: 1-831-205-9740