When using the wireless b43, networkmanager fails with "nm_device_wifi_new: assertion `driver != NULL' failed" and does not present the wlan0 interface to use. My analysis show that networkmanager is failing because it is trying to get the driver name from the parent of the device (at HAL) but the information is at the grandparent.
Created attachment 199167 [details, diff] This patch sarches for the grandparents for subsystems of ssb type I do not know if this patch affects other systems. It works on a b43 equipped machine and do not harm a iwlagn machine.
Could you please provide some more logs? I've got b43 wifi card, and it works like a charm with NM-0.7.1 and 2.6.30
(In reply to comment #2) > Could you please provide some more logs? I've got b43 wifi card, and it works > like a charm with NM-0.7.1 and 2.6.30 > Got the same problem, same versions. It's HAL dependent. Reverting to 0.5.12_rc1-r7 fixes problem. Patch should work too.
Created attachment 200278 [details, diff] Patch that disables the assertion OK, I can tell definitely that it is a bug in HAL, namely libhal. In fact, if you check lshal, you will see that the driver is detected by HAL and is present. However, libhal request made by NM tells that the key does not exist, which is not true. It is a bug, but it is in the HAL package itself, not in gentoo or NM or Kernel, or ssb system. I've filed a bug to freedesktop: http://bugs.freedesktop.org/show_bug.cgi?id=23114, so you can check the progress there. For now you could try to manually disable the check in NM code.The patch is attached, have a look. My HW is b43 WIFI on pci-e bus, AMD64. You can add the patch while building from source manually, it is not THAT big=)
(In reply to comment #4) > My HW is b43 WIFI on pci-e bus, AMD64. BTW, is not your hardware tx2000 series laptop by hp? Probably it is platform-related after all. This could help those guys at freedesktop.
It is similar to the tx2000, but it is a ZV6000 HP laptop. (In reply to comment #5) > (In reply to comment #4) > > > My HW is b43 WIFI on pci-e bus, AMD64. > BTW, is not your hardware tx2000 series laptop by hp? Probably it is > platform-related after all. This could help those guys at freedesktop. >
My platform: Dell D630
OK people. Have you ever seen wmaster interface at iwconfig or somthing like that? NM decides that it is the only wireless device, so it uses its HAL udi and gets out empty with driver name (maybe wmaster is not supposed to have driver). So, it is a problem in a way NM looks for wireless adapters. I do not know yet how to fix this. Any ideas? PS: I've closed libhal-related bug at freedesktop, it was invalid.
There is a wmaster interface but my debugging show that NM ignores it and uses the wlan0 interface. I do not know why but HAL now says that wlan0 has a ssb subsystem and the grandparent of wlan0 is the one that have the driver interface filled.
(In reply to comment #9) > There is a wmaster interface but my debugging show that NM ignores it and uses > the wlan0 interface. I do not know why but HAL now says that wlan0 has a ssb > subsystem and the grandparent of wlan0 is the one that have the driver > interface filled. > Then probably your patch works fine=) Hope it gets included into the ebuild soon.
Upstream submitted proper fix for this bug few minutes ago. You can expect backports over the weekend.
That's should be now fixed in 0.7.1_p20090824 Please re-open this ticket if problem is still there.
