[OLPC-devel] Build image 76.

[Dan Williams]

On Thu, 2006-08-24 at 14:31 -0400, Jim Gettys wrote:
> Interesting.  I'm glad we now know what was going on.
> 
> In theory, I would think the device driver should have responsibility
> for enabling the power on devices; in Linux has knowledge of the bus
> topology.
> 
> But I'll defer to Marcelo on what he thinks the right fix is.

Wouldn't we have to special-case this for OLPC?  It doesn't seem quite
right to have the 'enable' magic in the libertas driver, since that
driver works just fine for the USB dongles that we can plug into our
laptops, and is pretty clear trying to enable some random pin on an EC
my laptop doesn't have would fail :)

Is there a generic "OLPC" device driver that this sort of thing can live
in?  I just don't think we should be doing this in the libertas driver
because it's not actually anything to do with the 8388.  It's the EC
that needs driving here.

Dan

>                                   Regards,
>                                           - Jim
> 
> 
> On Thu, 2006-08-24 at 13:27 -0500, Richard Smith wrote:
> > > > No, but the libertas driver doesn't seem to send anything to dmesg under
> > > > Insyde either, so that doesn't help.  The device is missing from lsusb
> > > > and /proc/bus/usb/devices; it's "Bus 001 Device 004: ID 1286:2001" when
> > > > booted in Insyde, and that line doesn't appear inside LB.  I've checked
> > > > that ehci/uhci are both loaded.
> > > >
> > > > - Chris.
> > > I wonder if this is some magic being done by our friend the VSA ...
> > 
> > Nope looks to me to be a simple enable pin.  The EC has a singnal
> > marked WLAN_EN that appears to control the 3.3V to the marvel chip.
> > 
> > Its hooked up to GPIO18 on the EC.  So enable that output and wireless
> > should come back.
> > 
> > Should linuxbios do this or leave it up to a userspace program?
> >