In July / August this year I installed Ubuntu 10.10 on my Thinkpad, I noticed when I logged in that the wireless wasn't working. After some research I discovered that my Thinkpad is running a Broadcom BCM43224 wireless network card and that there were a number of known issues with the Broadcom drivers. Luckily for me there was a patch available that would solve all my problems automagically, took a few hours and a ethernet connection but I managed it.
I recently upgraded Ubuntu to 11.10 and haven't had much of a chance to play around on it. I did notice on Sunday night that the wifi wasn't working, ethernet was working fine which is why I hadn't noticed the issue earlier.
Deja Vu much? After the hoops I had to jump through to get wireless working on Ubuntu 10.10 I couldn't believe I was having the same issues in Ubuntu 11.10...
This time there is no patch sitting waiting to be downloaded that will automagically fix the wifi for me. After some research I found a few "work arounds" (could someone explain to me please why we have to find work arounds for known issues?)
The work arounds suggested deleting modules and installing other ones, although the work arounds had been well received I don't entirely feel comfortable jumping in and deleting modules and bits left, right and centre unless I know what it is that they do. I initiated a game of 20 questions with Mr. O.G.
Thinkpad running Broadcom BCM43224 wireless addapter, Ubuntu detects the wireless network card but proceeds to install the wrong drivers.
Solution:
Add bcma to the blacklist file created by bcmwl-kernel-source's postinst script.
- Edit /etc/modprobe.d/blacklist.conf to include bcma
For full information feel free to have a look at the Ubuntu bug report raised ~8weeks ago & suggested bcmwl-kernel-source patch information.
What bothers me most is that this bug effects a multitude of people and has been reported several times. I could understand that it could be considered a low priority issue and that valuable resources are being allocated to more pressing matters, but seriously accompanying this bug reported 8weeks ago was a patch to fix it... It would take very little to review, approve and make said patch available... my mind boggles!
I thought we got rid of this bcm issue back in 09 with DKMS, no?
ReplyDeleteThis blacklist workaround crap is no better than it was when the only option was ndiswrapper...and how long ago was that??
I have no idea what most of what you said means.
ReplyDeletehehe **noob** remember! :-)
NDISwrapper was a software that "wrapped" around the windows driver files. It was an ugly hack. Not too long after was b43-fwcutter, which extracted the firmware form the windows drivers. Kind of the same thing, but more natively supported after the firmware was actually extracted. A common problem with both of these methods was that you still had to tell the kernel to ignore the native "support" and use these hacks instead. Hence the blacklist.conf edits.
ReplyDeleteThen came along DKMS (Dynamic Kernel Module Support), which promised to dynamically load hardware modules which were not a part of the kernel, and it just happened to include the b43 module for the broadcom chips included in your wireless card. Being a 'native' way to include obscure drivers in kernel support, one was no longer required to edit blacklist.conf anymore, since the kernel already looked to DKMS for these drivers to begin with.
Here in your blog post, it seems we have a major regression. As you experienced, it can be kind of a PITA. How is a normal desktop user to feel confident resolving this?