Re: [PATCH 0/3] [RFC] Support for ADSL modems

On Tue, 2011-05-10 at 19:25 +0300, Pantelis Koukousoulas wrote:
> On Mon, May 9, 2011 at 9:38 PM, Dan Williams <dcbw redhat com> wrote:
> > On Fri, 2011-05-06 at 14:23 +0300, Pantelis Koukousoulas wrote:
> >> Hi, before I leave for, this patch series implements the initial
> >> version of the support for ADSL modems in network manager.
> >>
> >> Notes:
> >>    * PPPoA-only for now. I hope that PPPoE can be added with not too much trouble
> >>      as well, if it is possible to just create a new ethernet device for the "nas0"
> >>      interface that br2684ctl creates and reuse the already existing "wired" device
> >>      PPPoE support.
> >>
> >>      This way the adsl part can manage the ATM part of the equation just like it
> >>      does now for PPPoA.
> >
> > Well, the PPPoE bits aren't that complicated, so for now I'd rather
> > duplicate them for the ADSL code, and we can re-merge it later.  Most of
> > the code is simply detecting whether PPPoE should be used, and if so,
> > creating a new NMPPPManager object for the interface, then connecting to
> > the signals that it provides.
> The thing is that br2684ctl creates a device that is indistinguishable from
> an ethernet interface from what I can tell. So, probably udev and netlink
> will report it as an ethernet device, NetworkManager will configure it
> for PPPoE and everything will be ok? (in theory at least).

We'd want to figure out some way of distinguishing that interface from a
normal ethernet interface (since it's really not one), maybe by devtype
from sysfs, or maybe by the name ("nasX" where of course we can give
br2684ctl the -c parameter to specify the 'X' and look for that
interface when it shows up).  Then when we get a new interface
notification for the nasX we can start the PPPoE.  The bluetooth DUN
code does something like this, except there we actually get told what
the exact interface name will be instead of just picking a number.

> > But lets get PPPoA working first and PPPoE can come later.
> Agreed, let's wait and see how it will work in practice.

Yeah, sounds good.  THanks!


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]