Fatal trap 12: page fault while in kernel mode

[Gardner Bell]

On Tue, Apr 26, 2011 at 04:25:26PM +0200, Bernhard Schmidt wrote:
> On Tuesday, April 26, 2011 15:15:45 Gardner Bell wrote:
> > On Tue, Apr 26, 2011 at 4:12 AM, Bernhard Schmidt <bschmidt at freebsd.org> wrote:
> > > On Tuesday, April 26, 2011 01:09:42 Gardner Bell wrote:
> > >> Downloading a torrent with many peers on a toshiba satellite notebook
> > >> using an Atheros AR5006 wireless nic caused the following panic.  This
> > >> is an i386 system running 8.2-STABLE from around April 06.
> > >
> > > Can you reproduce that?
> >
> > So far I've not been able to reproduce this.
>
> Ok. I assume this only happens when loosing the connection and trying
> to re-associate. At least that is the only possible scenario I can
> think of where a timeout for mgmt frames is involved. Probably we
> aren't bumping a refcount correctly or something. Actually that sounds
> rather plausible as it panics exactly when trying to access ni which
> should, for a station, always point to iv_bss, which can in turn be
> free'd almost unconditionally if someone's telling net80211 to
> associate to another (or even the same) network. Hmm.. tracing refcount
> it is.
>
> Were you running wpa_supplicant at that point? Any messages before
> the panic happened?
>

Yes, I'm running wpa_supplicant with the following settings:

network={
    ssid="xxxxx"
    psk="xxxxx"
}

Other settings for the wireless card I have in rc.conf:

wlans_ath0="wlan0"
ifconfig_wlan0="WPA DHCP"
ifconfig_wlan0_alias0="inet 192.168.0.12 netmask 0xffffffff"


The last messages seen on the console before the panic are wlan0:
ieee80211_new_state_locked: pending SCAN -> AUTH transition lost
and several UP/DOWN events.


> --
> Bernhard

--
Gardner Bell