[lfs-dev] Scripts modified to add IPv[46] dual-stack support

Bruce Dubbs bruce.dubbs at gmail.com
Mon Feb 6 09:18:24 PST 2017


jpbion at westvi.com wrote:
> I made the following modifications to my LFS system to provide for having
> the inclusive or of (IPv4-static or IPv6-static) to exist on an interface,
> while allowing a single ifdown <dev> and ifup <dev> command.
>
> The changes are:
>
> 1) In ifconfig.dev file, I allow for the following new options: IP6,
> PREFIX6, GATEWAY6. (although most IPv6 situations have you nicely learning
> the default gateway from the router, I allow it to still be set). I
> created a new service called "ip46-static"
>
> 2) In the new ip46-static service, I added ability to read the IP6,
> PREFIX6, and GATEWAY6 variables from the ifconfig.dev file, and set the
> appropriate addresses on the interface.
>
> 3) In the same service, I disabled the removing of the IPv4 gateway from
> the routing table altogether. The reason being that removing the address
> from the interface removes all routes pointing out from the interface - so
> it won't appear in the table anyway. The existing code looked like it had
> a logic error, so wouldn't be executed anyway. I just commented it out -
> but if someone believes it should be there, easy enough to leave in.
>
> 4) In /sbin/ifup, I added the ability to set the IPv6 default gateway
> (again, this really should be learned from the router in IPv6, but the
> support is there, if desired).
>
> 5) I installed the BLFS traceroute program. I believe if static IPv6 is
> added to LFS vs. BLFS, that this package, installed after inetutils,
> should be part of LFS as well, as the lack of a v6-capable traceroute on
> an IPv6-enabled system is not good.
>
> 6) Here's an area that leaves me unhappy: In /sbin/ifdown, I removed the
> check for all addresses being removed from an interface before bringing it
> down. I just bring it down. If calling 'ifdown' - you kind of want to
> bring the interface down.
>
> 7) Another reason I am unhappy: It doesn't support the use of dynamic IPv6
> addressing - but then again, neither does LFS (BLFS does).
>
> Rationale for changes the way they are:
>
> a) Most of the time, if someone has both IPv4 and IPv6 addresses on an
> interface, they want them both up simultaneously. They want "ifdown <dev>"
> and "ifup <dev>" to do that for them, without thinking of issuing multiple
> commands. They also want the same logic to work if they just have one of
> v4 or v6.
>
> b) Supporting this is simple for the user - just add the IPV6, PREFIXV6,
> and perhaps a GATEWAY6 parameter to the same files they already edit.
>
> c) IPv6 is becoming more common. LFS 'standard' should support it.
>
> In summary:
>
> The changes work for me. I am not happy with (6) all that much at all,
> even though it works for me, it may not work for all (I didn't try the
> sub-interface (secondary address) case - that's where it may break.) I am
> not sure how many LFS users are creating secondary interfaces.
>
> If they are viewed as useful, I will ship the changed files off to
> someone. They really weren't much work - perhaps a total of an hour was
> spent on this; largely it is copy/paste of other code.

Please send the files directly to me at bdubbs at linuxfromscratch.org.

The problem that I have is that I do not use iv6 from my isp (not evne 
sure they assign one), but I can check on that.

That said, most of the ip addresses I see are ipv4.  Somewhere a nat 
process ipv4 <--> ipv6 is happening.  I suspect that is in a router 
somewhere in the isp's system.

   -- Bruce




More information about the lfs-dev mailing list