Re: [PATCH] OLPC Mesh interface support

[Daniel Drake <dsd laptop org>]

On Wed, 2009-07-15 at 14:15 -0400, Dan Williams wrote:
> A few things...  can you post the udev rules file so we can add that
> too?  I'd like to change that from NM_DEVICE_TYPE to ID_NM_OLPC_MESH to
> be more in keeping with udev's property naming.

Ok, then the rule will be

KERNEL=="msh*", SUBSYSTEM=="net", DRIVERS=="usb", ATTRS{idVendor}=="1286", ATTRS{idProduct}=="2001", ENV{ID_NM_OLPC_MESH}="1"

For some reason, the fact that it's driven by libertas doesn't get
exposed in the sysfs tree.

> Second, the autoconnect inhibit stuff is pretty elegant the way you've
> implemented it, but I'd like to see if we could implement it with the
> inhibit stuff I've got going on in another branch (which is a property
> on the device instead of a signal).  If you can't do that then maybe I
> can.

OK, where can I find this branch?

> Third, we try not to let the devices themselves have direct access to
> the manager for a few reasons (Tambet was the one who originally
> architected it that way) so I'd like to either revisit that restriction
> or have the manager pass the device list into the device creators and
> then call device-added/removed functions.  That basically duplicates
> signals (bad) but doesn't require the manager object (good).
> 
> I'll try to poke Tambet on #3.  OLPC is obviously the first device we've
> had that is so tightly tied to another.
> 
> So this means I've applied 1, 2, 3 and 5, but not the
> autoconnect-inhibit and not the main OLPC mesh patch yet.

Great, thanks!

I'll be able to work on the udev rule tweaking and the autoconnect stuff
tomorrow (if you can point me to it).

After that, I'm off to do volunteer work on the project in Nepal but
hopefully will be able to itch away at this in the evenings.

Daniel