Bug 12702 (raw)
Summary: | ath5k modul disconnect with Access-Point | ||
---|---|---|---|
Product: | Networking | Reporter: | Tobias (styla.tobi) |
Component: | Wireless | Assignee: | networking_wireless (networking_wireless) |
Status: | CLOSED INSUFFICIENT_DATA | ||
Severity: | normal | CC: | kernel-bugs, linville, mcgrof, me |
Priority: | P1 | ||
Hardware: | All | ||
OS: | Linux | ||
Kernel Version: | 2.6.28-ARCH | Subsystem: | |
Regression: | No | Bisected commit-id: |
Description
Tobias
2009-02-14 04:17:05 UTC
I am seeing the same problem on CentOS 5.3 with kernel-2.6.18-128.1.6.el5. After the "assume out of range" message: - The card is dead, iwconfig reports "not associated" - Attempt to do something (e.g. turn the card off, rmmod, etc) results in the call freezing. - Only reboot (hard reboot because of the freeze) helps. The module reports the card as: ath5k phy0: Atheros AR5414 chip found (MAC: 0xa5, PHY: 0x61) P.S. Forgot to mention - this is on a dual-core 64-bit Athlon This is normal: if a probe response fails, then you will get disconnected -- this happens with all other wireless hardware. I'm curious why you need to reboot to fix it. Can't you just re-associate with the AP? NetworkManager or wpa_supplicant daemons will do this automatically. (In reply to comment #1) > I am seeing the same problem on CentOS 5.3 with kernel-2.6.18-128.1.6.el5. > After the "assume out of range" message: > - The card is dead, iwconfig reports "not associated" > - Attempt to do something (e.g. turn the card off, rmmod, etc) results in > the > call freezing. Rmmod hangs, or the whole computer hangs? Can you reproduce this on a vanilla kernel? If the whole computer hangs, try to remove the module while you're on a text VT, and take a picture of the oops if there is one. I am not sure how to get it to reassociate. In my case, there is no WEP or anything, no NetworkManager, no wpa_supplicant, just plain "ifup" script (that uses ifconfig/iwconfig/etc). If I try to bring the interface down so that I can bring it back up, the process freezes. After that even a simple "iwconfig" (without arguments) would freeze. P.S. Do not have time to mess with vanilla kernel, sorry. P.P.S. For now I removed the ath5k module and installed madwifi - seems to work OK. (In reply to comment #5) > I am not sure how to get it to reassociate. In my case, there is no WEP or > anything, no NetworkManager, no wpa_supplicant, just plain "ifup" script > (that > uses ifconfig/iwconfig/etc). If I try to bring the interface down so that I > can > bring it back up, the process freezes. After that even a simple "iwconfig" > (without arguments) would freeze. Well, that definitely sounds like some kind of locking problem. However, since I have no idea what centos has in its kernel (2.6.18 is very old -- I assume it is heavily patched), I can't really say if it's a problem in the current kernel or not, except that I haven't seen that here. It is probably worth reporting that to CentOS so others can benefit if they made a back-porting mistake. > P.P.S. For now I removed the ath5k module and installed madwifi - seems to > work OK. Ok - I hope it works better for you in the future. Disregarding the CentOS problem...FWIW, I think current RHEL kernels fix the issue reported -- http://people.redhat.com/dzickus/el5 Does the original reporter still see this problem with current (i.e. 2.6.30 or later) upstream kernels? Closed due to lack of response... |