Re: iwn, wpa_supplicant, and wpa2 enterprise

[Sverre Froyen <sverre%viewmark.com@localhost>]

On Fri June 4 2010 16:46:09 Martin Husemann wrote:
> On Fri, Jun 04, 2010 at 08:17:26PM +0100, Roy Marples wrote:
> > Kernel notifies link is up, dhcpcd reacts to this and attempts to query
> > the interface. So any failure must be within the driver.
> 
> Why does it signal link up to the routing socket while still loading
> firmware?

Beats me :-)

But seriously:

How does the routing socket determine that the link is up? By responding to an 
ioctl? If so, this is what the patch broke, in the sense that the ioctl call 
returns EBUSY until iwn_init has completed the firmware load and 
initialization. The reason for the patch (I imagine) was that in some cases 
the ioctl that causes the firmware to load was retriggered while the first call 
was still active (resulting in a crash).

I'm hoping that the revised patch that I posted will work OK until Damien 
decides on a proper fix.

Regards,
Sverre