Re: Patch for dynaddr

JuanJo Ciarlante (irriga@impsat1.com.ar)
Fri, 13 Feb 1998 17:29:50 -0300 (ART)


On Thu, 12 Feb 1998, Erik Corry wrote:

> Hi,
>
> I have a patch for the dynaddr functionality in 2.0.33 (the 2.1
> version is not done yet). The patch fixes a bug and adds a little
> new functionality.
>
> (Dynaddr changes the IP address on a socket when the address
> of it 's network interface changes due to dial on demand and
> dynamic IP numbers).
>
> At present, the dynaddr code is only triggered if you either
> switch device, or you are masquerading and the IP address
> changes. If you are not masquerading (ie it is a local socket)
> and you don't switch device (ie you are using pppd-2.3 or ipppd,
> and not diald) then dynaddr is not triggered. The patch brings
> local sockets into line with masqeraded sockets. ISDN users
> need this bugfix asap.
>
> The code also includes an option to switch the IP source
> address even if the socket has already connected. This is
> called RST-provoking mode, and is off by default. The
> documentation says:
>
> If you enable the RST-provoking mode, then the source address will
> be changed, even if the socket is established.
> [...]
I find your changes Ok (I don't have a 2.0 PPP machine near), maybe we can
include in 2.0.34 (recall that the default is harmless). ADDR vs DEV
change detection is actually the right approach ("demand" ppp).

By sending a true src-bogus packet we bounce at the other side to shutdown
our socket; I find this good because returned RST packet will be seen in
the connection path (think about diald, eg) instead of just shutting down
the socket. But the server's socket will never know we are giving up,
maybe we can act ugly-er and nicer by can shutting down our socket and
sending a ``true'' RST.

I'll try to make a 2.1 patch.

-- Juanjo http://juanjox.home.ml.org

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu