Latest working kernel version:2.6.25-rc1 Earliest failing kernel version:2.6.25-rc2 Distribution:debian/etch Hardware Environment:x86 + bcm4306 (14e4:4320(rev 03)) Software Environment:debian/etch + new kernel mentioned above+wpa_supplicant v0.6.0 (cvs as of 02/17/2008, compiled against 2.6.25-rc1-headers). Both compiled locally by gcc-4.2.3. (Kernel configuration and full output of lswh lspcicould be send on request, since there is no possibility to attach files here.) Problem Description: When wpa_supplicant comes into play during init, the system freezes with the following messages: "wpa_supplicant[2734]: segfault at 806de20 ip 806de20 sp bf8693ac error 4 Kernel panic -- not syncing: Aiee, killing interrupt handler" After a few seconds (or else after pressing a key?) a new (unrelated?) message is added: "atkbd.c: Spurious ACK on ISA0060/Serio0. Some programs might be trying access HW directly." Steps to reproduce: Try to connect via wpa_supplicant/WPA2 to an AP (which in my case is a WRT54GS V1.0 running openwrt on linux-2.4. Other models not tried.) With 2.6.25-rc1 I already experienced the problem, that udev did not rename the wlan device to eth2, as previous kernel versions actually did (at least 2.6.23.11). I found the following workaround (within /etc/udev/rules.d/z25_persistent-net.rules). This may or may not have to do with the problem: # PCI device 0x14e4:0x4320 (bcm43xx) #SUBSYSTEM=="net", DRIVERS=="?*", ATTRS{address}=="00:90:(censored)", NAME="eth2" SUBSYSTEM=="net", DRIVERS=="?*", SYSFS{address}=="00:90:(censored)", SYSFS{type}=="801", NAME="wmaster0" SUBSYSTEM=="net", DRIVERS=="?*", SYSFS{address}=="00:90:(censored)", NAME="eth2"
Created attachment 14874 [details] Kernel configuration 2.6.25-rc2
Can you confirm that the patch from comment #2 of bug #10002 fixes it for you?
Yes, after having applied yout patch from comment #2 of bug #10002, the segfault is gone, and the device runs normally. Reading the description of #10002, I recognize it as beeing the very same behaviour. Yet I did not find it, even though I looked for "wpa_supplicant" and "segfault" before. (Was it because the author used the term "freeze" instead of segfault?)
*** This bug has been marked as a duplicate of bug 10002 ***