From: Andrej Podzimek <andrej@po...> - 2008-09-30 02:41:48
Here's a piece of good news: The problem still occurs quite often, but it does *not* cause fatal network failures any more. There are no IPv6 blackouts. Router advertisements do not seem to get blocked after stuck beacon events. Moreover, router solicitation works now! (It has never worked with madwifi before.)
Unreported beacon loss seems to have been resolved. I used to receive about one beacon in five seconds (!!!) with some of the earlier revisions. Everything seems to work fine right now. There are some rare delays in beaconing approximately once a minute. (This is neither unusual, nor harmful.) Beacons are emitted regularly for the rest of the time.
Using kernel 126.96.36.199 with revision 3866, branch madwifi-hal-testing.
The only problem is my VoIP phone (using WPA2-PSK) that stopped working. (Calls are interrupted ater a few seconds with lots of error messages on the hostapd side.) All the other clients (using WPA2-EAP-TLS) work just fine. No inexplicable disconnections and the like.
Many thanks to all the brilliant hackers who fixed the driver.