Description of problem: This is rawhide, using ATrpm's ipw3945 packages. This is a Dell Precision M65 laptop with a Intel Pro Wireless 3945 ABG card. The laptop has a "kill switch" on the side which is supposed to cut power to the Wifi and Bluetooth radios. Version-Release number of selected component (if applicable): Linux lithium 2.6.18-1.2798.fc6 #1 SMP Mon Oct 16 14:37:32 EDT 2006 i686 i686 i386 GNU/Linux ipw3945-kmdl-2.6.18-1.2798.fc6-1.0.12-11.1.fc6.at ipw3945-ucode-1.13-2.at ipw3945d-1.7.22-3 How reproducible: Either (a) by removing the module while the radio is killed, or (b) by turning off the kill switch when the module is loaded. Steps to Reproduce: Scenario a) 0. With laptop powered off, turn kill switch off (to normal position.) 1. Boot system, get wifi link up and running as normal. 2. Turn on the RF kill switch (to kill position.) 3. Modprobe -r ipw3945 4. Wait for oops... 5. Only one cpu on my dual-core system locks up. The rest of the system is responsive for a few seconds (mouse works, apps work), then soon the whole thing goes down. SysRq-RSEIUB works. Scenario b) 0. With system powered off, turn kill switch on (to kill pos.) 1. Boot system. 2. Modprobe ipw3945 3. Turn off the kill switch. 4. Wait for oops... 5. System completely locks up immediately, no mouse, unlike above. Can't emergency sync the HD, or unmount it, but Alt-sysRq-B reboots it. Actual results: Kernel panics. Expected results: Wifi subsystem should not take the "kill switch" literally!
Created attachment 140673 [details] output of lspci and lsmod
Created attachment 140674 [details] Syslog from scenario (A)
Created attachment 140676 [details] Syslog from scenario (B)
Created attachment 140679 [details] better lspci & /proc info
as the ipw3945 driver isn't part of the Fedora kernel, report this to the upstream developers. There's nothing to fix in the Fedora kernel until it gets merged.