PATCH RFC: wpa_supplicant influencing device operstate

[Stefan Rompf]

Hi,

some weeks ago, there has been a quite heated discussion on linux-netdev about 
extending functionality of netif_carrier_on()/off(). In the end, a patch 
written by me has been applied for 2.6.17. For an overview, see the 
documentation at http://www.flamewarmaster.de/software/operstates.txt . The 
patch itself (applies to 2.6.14+) can be found on this site too. Main point 
concerning this list is that it allows an userspace supplicant to signal the 
kernel when the interface is ready for user traffic after authentication, and 
afterwards, kernel notifies other programs via the existing IFF_RUNNING 
interface.

The quagga routing daemon, vrrpd and my new dhcp client 
(http://www.flamewarmaster.de/software/dhcpclient/) can act on these 
messages.

I have attached a patch against wpa_supplicant 0.5.1 to this mail that 
implements signalling functionality in the WEXT driver. Note that I haven't 
tested how this affects drivers that rely on WEXT functionality - this is a 
very first version.

With this extension, I can start wpa_supplicant and my DHCP client in 
parallel. Whenever the supplicant has completed association, and WPA 
authentication if needed, the DHCP client will automatically update the 
interface IP configuration. Together with a graphical wpa_cli (will try 
wpa_gui), this increases usability of wireless LANs a lot. In the long run, 
I'd like this feature to be added to wpa_supplicant.

Thoughts?

Stefan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: wpa_supplicant.diff
Type: text/x-diff
Size: 7523 bytes
Desc: not available
Url : http://lists.shmoo.com/pipermail/hostap/attachments/20060304/331eea09/attachment.diff