Description of problem: My WLAN card stopped working after updating kernel to 3.2.2-1. It seems not to be receiving any packets (monitor mode doesn't show anything incoming, not even beacons). Manually started scan using "iw wlan0 scan" also fails to find any networks (and only the transmitted probes are visible using monitor). No errors are reported via dmesg. Switching to 3.2.1-3 makes the card work fine again. I will test vanilla 3.2.2 and probably also a bisect between 3.2.1 and 3.2.2. Any advice on how to get detailed debugging output from the driver would be really welcome (as it is, everything seems fine except for frames not going in). Version-Release number of selected component (if applicable): kernel-3.2.2-1.fc16 How reproducible: Every boot to 3.2.2 exhibits this.
Vanilla 3.2.2 (with the config from kernel-3.2.2-1.fc16 but disabling unneeded modules) works fine.
(In reply to comment #1) > Vanilla 3.2.2 (with the config from kernel-3.2.2-1.fc16 but disabling unneeded > modules) works fine. Fedora 3.2.2 is using the compat-wireless stuff from 3.3-rc1.
Same here: # lspci | grep -i network 00:09.0 Network controller: Ralink corp. RT2500 802.11g (rev 01) Booting with old kernels, everything is fine. Booting new kernel (kernel-PAE-3.2.2-1.fc16.i686) => no wifi-connection:
(In reply to comment #2) > Fedora 3.2.2 is using the compat-wireless stuff from 3.3-rc1. Thanks for the hint. I've tested vanilla 3.3-rc1 and that one works fine, too. I tried to build kernel-3.2.2-1 --without backports, but ran out of diskspace building all the modules. If there is a way to bisect the fedora-modified kernel without having to run a full-blown RPM build (that takes several hours here), I'd be glad to give it a try. Or if anyone could possibly build the RPM for me (a scratch build?), that would be nice, too.
Created attachment 558174 [details] dmesg from the broken 3.2.2-1
Created attachment 558175 [details] dmesg from vanilla 3.3-rc1 This kernel has several wlan-related debug options enabled, so it reports wlan0: detected beacon loss from AP - sending probe request (these messages are printed exactly one second apart, but several are missing). Could this be somehow related to the bug?
(In reply to comment #4) > (In reply to comment #2) > > Fedora 3.2.2 is using the compat-wireless stuff from 3.3-rc1. > > Thanks for the hint. I've tested vanilla 3.3-rc1 and that one works fine, too. > I tried to build kernel-3.2.2-1 --without backports, but ran out of diskspace > building all the modules. I tried this the other way around and compiled compat-wireless-3.3-rc1-2 for 3.2.1-3. The WiFi is working fine, so compat-wireless definitely is not the culprit. Now how just to find out which of those many Fedora patches is responsible.
This has bitten me also. No error messages, but no connection and no networks found. The only messages in /var/log/messages that refer to rt2500, wlan0 or NetworkManager are: kernel: [ 18.622372] rt2500pci 0000:07:01.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17 ... [ 22.342192] ADDRCONF(NETDEV_UP): wlan0: link is not ready ... NetworkManager[1263]: (NetworkManager:1263): GLib-GIO-WARNING **: Missing callback called fullpath = /var/run/ConsoleKit/database I'll attach the full log. Gareth
Created attachment 558204 [details] /var/log/messages from a working kernel
Created attachment 558205 [details] /var/log/messages from the broken kernel
Looks like NetworkManager has problems to detect device, perhaps there are some race condition during start? Is there any difference between iwconfig and ifconfig on broken and good kernel? Does NM detect device if you restart it ("systemctl stop NetworkManager.service" and "systemctl start NetworkManager.service") ?
(In reply to comment #7) > I tried this the other way around and compiled compat-wireless-3.3-rc1-2 for > 3.2.1-3. The WiFi is working fine, so compat-wireless definitely is not the > culprit. Now how just to find out which of those many Fedora patches is > responsible. We apply these patches on compat-wireless: > ApplyPatch compat-wireless-config-fixups.patch > ApplyPatch compat-wireless-pr_fmt-warning-avoidance.patch > ApplyPatch compat-wireless-rtl8192cu-Fix-WARNING-on-suspend-resume.patch > ApplyPatch mac80211-fix-work-removal-on-deauth-request.patch > ApplyPatch b43-add-option-to-avoid-duplicating-device-support-w.patch Only one suspicious is "mac80211-fix-work-removal-on-deauth-request.patch". Tomáš, can you apply it on vanilla compat-wireless-3.3-rc1-2.tar.bz2, recompile, and check if it broke wireless?
(In reply to comment #11) > Looks like NetworkManager has problems to detect device, perhaps there are some > race condition during start? Is there any difference between iwconfig and > ifconfig on broken and good kernel? Does NM detect device if you restart it > ("systemctl stop NetworkManager.service" and "systemctl start > NetworkManager.service") ? I doubt this is NM related. Restarting NM doesn't fix anything and the problem exhibits itself by no packets coming in, so a simple "iw wlan0 scan" doesn't work on the broken kernel no matter how the interface got activated. (In reply to comment #12) > Only one suspicious is "mac80211-fix-work-removal-on-deauth-request.patch". > Tomáš, can you apply it on vanilla compat-wireless-3.3-rc1-2.tar.bz2, > recompile, and check if it broke wireless? I did and to my great disappointment, wireless kept working with that patch. However, I found out that by changing the priorities in /etc/depmod.d/dist.conf and running modprobe -rv rt2500pci && modprobe -v rt2500pci, I can switch back and forth between the broken state (backports/ modules) and working wlan (updates/ modules - my compat-wireless build). After extensive module juggling, I've pinpointed the problem to drivers/misc/eeprom/eeprom_93cx6.ko. Placing just my compat-wireless version to /lib/modules/3.2.2-1.fc16.x86_64/updates/drivers/misc/eeprom/eeprom_93cx6.ko the wireless starts working (with the rest of modules kept original). Note that the RPM doesn't ship the compat-wireless version of this module in backports/ and the original 3.2.2 one in kernel/ is used instead. Could this be just some incompatibility between the 3.2.2 EEPROM driver and the 3.3-rc1 wlan stack?
Created attachment 558293 [details] Working eeprom_93cx6.ko from compat-wireless-3.3-rc1 This is the module that fixed it for me (x86_64). Gareth, could you place this into /lib/modules/3.2.2-1.fc16.x86_64/updates/drivers/misc/eeprom/eeprom_93cx6.ko, run depmod and see whether your wifi starts working after module reload/reboot?
Thanks Tomáš! So problem is that our build-in kernel compat-wireless does not properly install eeprom_93cx6.ko into /lib/modules/`uname -r`/backports/
Thanks indeed Tomáš! Yes, your eeprom module fixes it.
*** Bug 785386 has been marked as a duplicate of this bug. ***
I made the required change -- I guess the eeprcom_93cx6 API changed along the way. Anyway, the next f16 build should address this.
Will F15 i386 get the fix for my Eee PC 901? Same issues.
F15 doesn't have compat-wireless integrated, so whatever promise you (i.e. Thomas Arnold) are seeing is something different. Please open a separate bug for that specific issue.
Great, I already did that but it was marked as duplicate of this one: https://bugzilla.redhat.com/show_bug.cgi?id=785510
(In reply to comment #21) > Great, I already did that but it was marked as duplicate of this one: > https://bugzilla.redhat.com/show_bug.cgi?id=785510 That bug is reported with an F16 kernel. I'm not sure why you think it's an F15 bug report? If you're installing F16 kernels by hand on an F15 userspace, it's still an F16 problem.
*** Bug 785510 has been marked as a duplicate of this bug. ***
OK, given that then you (i.e. Thomas Arnold) will just need to install the updated f16 kernel when it becomes available...
*** Bug 785502 has been marked as a duplicate of this bug. ***
Just a note (mainly for indexing purposes) that I am seeing identical var/log/messages traces in machines with rt2500 and rt2800 cards, so I think it is likely to be the same bug.
In my case in ASUS EEEPC 1000H no wi-fi cards are detected.
*** Bug 786713 has been marked as a duplicate of this bug. ***
*** Bug 786935 has been marked as a duplicate of this bug. ***
*** Bug 785533 has been marked as a duplicate of this bug. ***
*** Bug 786952 has been marked as a duplicate of this bug. ***
New kernel fixed this bug for me.
Same with my system - kernel 3.2.3 does work for me. Thank you!
My EEEPC 1000H is alright now with this new kernel. Please don't repeat that joke. For moments I thought my board got broken. Thank you!
I installed 3.2.3-2.fc16.i686 and I am unable to connect to my wireless card. The card detects my router, however despite having the correct WEP Key entered, my machine does not authenticate and will not establish a connection. Is this related to the problem outlined above, or is it a separate issue?
It seems a different description. In this bug the card was not detected. Your case it detects but you're not able to connect. right? I suggest you to delete your wi-fi connection. Press ALT-F2 to get to the run dialog and type nm-connection-editor, delete the non-working connection and try to register again. If it does not work, well it may be a bug, so open a new ticket.