Re: [PATCH v2 2/9] Ignore rild modem devices.

[Dan Williams <dcbw redhat com>]

On Tue, 2016-06-14 at 15:28 -0400, Tony Espy wrote:
> On 06/14/2016 12:42 PM, Dan Williams wrote:
> > On Fri, 2016-06-10 at 17:56 -0400, Tony Espy wrote:
> > > From: Mathieu Trudel-Lapierre <mathieu trudel-lapierre canonical 
> > > com>
> > >
> > > Gbp-Pq: Name Ignore-rild-modem-devices.patch
> > > ---
> > >   src/nm-manager.c | 8 ++++++++
> > >   1 file changed, 8 insertions(+)
> > >
> > > diff --git a/src/nm-manager.c b/src/nm-manager.c
> > > index e64c68a..be21baa 100644
> > > --- a/src/nm-manager.c
> > > +++ b/src/nm-manager.c
> > > @@ -2057,6 +2057,14 @@ platform_link_added (NMManager *self,
> > >                   gboolean ignore = FALSE;
> > >                   gs_free_error GError *error = NULL;
> > >
> > > +         /* Ignore rild modem devices, which will be
> > > handled
> > > by their modem parent */
> > > +         if (g_str_has_prefix (plink->name, "rmnet") ||
> > > +                 g_str_has_prefix (plink->name,
> > > "rev_rmnet")
> > > > >
> > > +                 g_str_has_prefix (plink->name, "ccmni"))
> > > {
> > > +                 _LOGW (LOGD_HW, "Ignoring rild modem
> > > device:
> > > %s", plink->name);
> > > +                 return;
> > > +         }
> > > +
> > >                   device = nm_device_factory_create_device
> > > (factory,
> > > plink->name, plink, NULL, &ignore, &error);
> > >                   if (!device) {
> > >                           if (!ignore) {
> > I pushed an alternate version of this patch to th/review/ofono;
> > Thomas
> > is right that to put this into the WWAN plugin would require them
> > being
> > tagged with devtype=WWAN, and none of the drivers do that, at least
> > from sources I can find.  Since they aren't upstream, we can't fix
> > them.
> Right. We'll see if this is something we can push through on our
> end, 
> but I'm not too optimistic.
>
> I'm also surprised that from what I've read, this isn't something
> that 
> could be handled by a udev rule, but apparently DEVTYPE isn't
> something 
> that can be fixed up.

Yeah, I don't have hope of getting the devtype correct until (if ever?)
the drivers hit the kernel.  At least for rmnet there's a possibility,
somebody from Sony was recently working on the basic SMD bits on which
rmnet would get based.

> > Instead, I think we should tag them with TYPE_WWAN_NET, and they'll
> > get
> > automatically ignored by the WWAN plugin.
> I guess this is the best we can do, however it still leaves us
> facing 
> the same problem with the next ISV/ODM unwilling to 
> "do-the-right-thing".  ;)-

Oh, by "tagging" here I meant internally in src/platform, not via udev
rules or anything.  Sorry for the confusion.  Nothing for the ISV/ODMs
to do here, and it's basically just moving your interface name check to
a different place.

Dan