connman vs. interface detection

[Marcel Holtmann]

Hi Werner,

> Stefan, yesterday, you sent me links to connman's detect.c and rtnl.c,
> saying that connman would fail to detect the WLAN interface. I already
> commented yesterday that cursory inspection didn't reveal anything
> suspicious.
> 
> I've now resolved or dummied out all the external dependencies and ran
> it on a Neo, but it seems to perform fine. The original is here:
> git clone http://git.moblin.org/repos/projects/connman.git
> My variant is here:
> http://svn.openmoko.org/developers/werner/wlan/connman/
> 
> Specifically, I tested:
> - starting it when the interface is up: detects and registers it
> - idem, but unbinding and re-binding the driver: detect.c detects
>   and registers the interface
> - starting with the interface unbound and then binding it: again,
>   detects and registers
> 
> The registration reports:
> 
> detect.c:detect_newlink() type 1 index 5
> connman_device_create(dev_00_12_CF_8F_34_CF, 2)
> connman_device_set_interface(eth0)
> 
> The 2 is CONNMAN_DEVICE_TYPE_WIFI.
> 
> So it seems that all is well in the lower layers.

if no wpa_supplicant -u is running, it will detect the WiFi devices in
your system, but not export them over D-Bus. You can start and stop
wpa_supplicant at any time you like. The WiFi devices will then
disappear or reappear.

Regards

Marcel