Why not aging the BSS in use?

[Jouni Malinen]

On Fri, Apr 12, 2013 at 09:07:43AM +0000, Mauger, MatthieuX wrote:
> We are currently facing a tricky problem, we are using a WiFi chip that is sometime not so reliable and we not always receive a deauth or beacon loss from the chip when leaving the range of a network.
> In such situation, we can trust that we are still connected even at kilometers from the BSS!
> 
> As a workaround we are considering to use the aging mechanism in order to not keep such BSS in the scan list.
> However we saw in function "wpa_bss_update_end" in bss.c that the aging is voluntary not applied to the current BSS. 
> We are wondering if is there a good reason to skip it? What could be the risk to apply aging on the BSS in use?

Yes, there are areas where the BSS entry for the current BSS is assumed
to be there. In many cases, the BSS entry within wpa_supplicant is not
updated during the association and as such, it could expire without the
code that is used to skip expiration of the current BSS.

It sounds like the workaround should really be in the driver code rather
than wpa_supplicant..

-- 
Jouni Malinen                                            PGP id EFC895FA