MMKernelDevice ID_USB_INTERFACE_NUM property isn't set anymore

[Dan Williams]

On Wed, 2017-05-03 at 20:21 +0200, Aleksander Morgado wrote:
> On Wed, May 3, 2017 at 5:08 PM, Dan Williams <dcbw at redhat.com> wrote:
> > > > > Do you have a /lib/udev/rules.d/60-serial.rules in your
> > > > > system
> > > > > that
> > > > > sets ID_USB_INTERFACE_NUM? Maybe we shouldn't rely on having
> > > > > that?
> > > > > 
> > > > 
> > > > No, I do not have that 60-serial.rules file, but I did obtained
> > > > the
> > > > result I
> > > > wanted adding the following rule to 77-mm-telit-port-
> > > > types.rules
> > > > 
> > > > SUBSYSTEMS=="usb", ATTRS{bInterfaceNumber}=="?*",
> > > > ENV{ID_USB_INTERFACE_NUM}="$attr{bInterfaceNumber}"
> > > 
> > > We need to fix this; we shouldn't be relying on a systemd
> > > property. I
> > > originally thought this was provided by udev itself... although
> > > with
> > > the systemd/udev development being done together, it may actually
> > > be
> > > a
> > > packaging decision; no idea.
> > 
> > I think it originally was a udev property.  And I see it's still
> > provided by systemd-udev, so I'd expect it to be present if udev is
> > installed...
> 
> In my Arch system, it looks like it's the "systemd" package is the
> one
> installing that file.

Does Arch have a separate udev package then?  If so, that would seem
like a packaging bug.  In Fedora we have "systemd" and "systemd-udev"
as separate packages, and systemd-udev owns the udev rules.

That said, our systemd-udev package depends on the systemd package, so
in the end it doesn't make a difference on an installed system.

Dan