[systemd-devel] [PATCH 2/2] udev: fix for devices without PCI

[Koen Kooi]

Op 3 mrt. 2013, om 19:22 heeft Kay Sievers <kay at vrfy.org> het volgende geschreven:

> On Sun, Mar 3, 2013 at 7:10 PM, Rob Clark <robdclark at gmail.com> wrote:
>> On Sun, Mar 3, 2013 at 1:06 PM, Kay Sievers <kay at vrfy.org> wrote:
>>> On Sun, Mar 3, 2013 at 6:55 PM, Rob Clark <robdclark at gmail.com> wrote:
>>>> On many arm embedded SoC's (phones, tablets, etc), there is no PCI bus.
>>>> So it is not an error if names_pci() fails to find a parent PCI device.
>>> 
>>> As the comment in the code lines above states, all these names work on
>>> for PCI based setups.
>>> 
>>> Other buses and architectures would need proper code to handle them,
>>> not just skip over the pci prefix. We cannot do that.
>> 
>> Of course, if you have a better idea in mind, I am all-ears.  I'm not
>> a systemd/udev expert, so certainly don't claim that it is the best
>> fix.  It would be nice for this to work somehow properly on the
>> various PCI-less systems out there.
> 
> It depends on the bus type used for the parent device(s), we would
> need to compose a stable and predictable name from the properties of
> the parent device(s). In most cases it's probably a simple "platform"
> bus hack, which we will see.
> 
> What does:
>  ls -l /sys/class/net
> print on these systems?

On an TI AM335x SoC (e.g. beaglebone): 

root at xfcebone:~# ls -l /sys/class/net/
lrwxrwxrwx    1 root     root             0 May  3 06:23 eth0 -> ../../devices/ocp.2/4a100000.ethernet/net/eth0
lrwxrwxrwx    1 root     root             0 May  3 06:23 lo -> ../../devices/virtual/net/lo

regards,

Koen