Hi, I have a few i386 boxes running OpenBSD hostapd, each with 2 rum devices (mix of manufacturers, revisions, with/without USB cables). Most of them panic, crash hard, or one/both rum interfaces stop transmitting every few hours/days.
So far I seem to think there are numerous, separate issues in OpenBSD's if_rum.c making it unstable right now, but I'm trying to take time to work on this: Issue 1: beacons didn't work (fixed in CVS HEAD, rev 1.99 -- thanks) Issue 2: physically pulling a rum device that is 'up' and busy (sending lots of broadcast pings, hostapd not running) can trigger a panic (about 1 time in 10); I think I'm close to having a patch for this; Issue 3: occasional 'rum0: device timeout' during operation; after this, flag IFF_OACTIVE stays set, no more packets go out, and if PF/cbq is enabled, the queue fills right up; Issue 4: after a timeout, 'ifconfig rum0 stop' or a reboot will then crash the box hard, so far I've seen a panic message on the serial port, and with ddb.panic=0 it fails to automatically restart; Issue 5: a 'silent' tx lockup of the device; with no mention of any 'device timeout', and IFF_OACTIVE not set, I noticed a device stop responding to Association Requests; with hostapd still running, I did "ifconfig rum1 down ; ifconfig rum1 up" and got 'rum1: failed to queue raw tx frame' 4 times before the box crashed hard. Issue 6: deauthentication frames are constantly sent out to devices that haven't been in range for hours, to the same BSSIDs over and over; as if it is about to purge them from the node cache, but actually doesn't; I think these frames are what triggered the 'failed to queue raw tx frame' mentioned for Issue 5 above while the interface was going down. Thanks for any help or suggestions. I'll continue to share any more successes/failures I have with this. Regards, -- Steven Chamberlain [email protected]
