Description of problem: Transferrate drops or complete stalls due to firmware resets. Occasional 'unknown notification' errors Version-Release number of selected component (if applicable): kernel 2.6.15-1.1861 - 1863 How reproducible: Always after generating IP traffic, say 'yum upgrade', over the wireless link. Dmesg shows ipw2200: Firmware error detected. Restarting. Numerous times. Transferrate drops or stalls as a result. Steps to Reproduce: 1. Generate some IP traffic over the wireless link 2. Transferrate drops or completely stalls 3. Check for errors with dmesg Actual results: Transferrate drops or completely stalls Expected results: No firmware resets, no tranferrate degradation of stalls. Additional info: firmware 2.4 from Sourceforge ipw2200.sf.net. Configured with 128-bit WEP. The firmware restarts seem to be happening since ipw2200 driver > 1.01 (or version included in FC4 kernel). Occasional 'Unknown notification' errors look like: ipw2200: Unknown notification: subtype=40,flags=0xa0,size=40
Created attachment 123685 [details] Complete dmesg output with all ipw2200 messages over a 20 min period
I have the same problem in FC4 with testing kernel 2.6.15-1.1823.
If a download just stalls, no more connections work after that, but network device restart helps. Sometimes a stalled download in accompanied by the fact that ksoftirqd starts eating 100% cpu, and this problem can only be "fixed" only by restarting the system.
This bug shows up in FC4 with all the latest updates, so probably the version above should be changed to 'fc4' and priority to 'high'?
Jurgen, does the kernel from http://people.redhat.com/davej/kernels/Fedora/devel/ fix this for you ? John merged a fix recently that may solve this problem, but due to the test3 freeze, I don't think it made it to rawhide yet. Anton, the same fix has been merged in FC4 CVS, and will go out in the next update soon.
OK, I've tried the devel kernel from feb. 19th and used my wireless link to do a yum update. After downloading about 650Mb over the wireless connection still no firmware resets! So for now it seems the bug is gone. Thanks.