On 07/13/2011 11:06 AM, kullmann wrote:
>
> lwfinger wrote:
>> That is not a side effect of wl. It is a matter of having a later
> kernel.
>
> But a new kernel was installed with the installation of broadcom-wl (I
> find it strange, that such a radical thing
> is hidden in the installation of another package). And during that
> installation also Radeon-stuff was installed
> (watching the messages). Hm, perhaps the Radeon-installation
> (apparently xorg-x11-driver-video-radeonhd)
> was dragged in by some other package?? But for sure, after installation
> of broadcom-wl I had to reboot, and
> a new kernel showed up on the boot-menu. The kernel I guess came with
> broadcom-wl-kmp-default.
>
> I wanted to see what /usr/sbin/install_bcm43xx_firmware says to
> “–help”, and then, so well, it installed that firmware.
> Now the Yast hardware information lists the “Wireless 1397 WLAN
> Mini-Card” again, and regarding drivers
> it says “Active: No; modprobe: Yes; modules modprobe ssb”.
>
> So apparently all what is needed now is to activate that driver?
>
> Regarding that blacklisting-issue: in /etc/modprobe.d there is no file
> blacklist.conf.
It is named 50-blacklist.conf.
> I am now confused. Looking at
> www.broadcom.com/docs/linux_sta/README.txt
> it says that b43 and ssb are other open-source drivers, which will
> conflict
> their driver and thus must be removed. But I thought that ssb and b43
> would come
> with the firmware, and thus are associated with Broadcom? Or is “wl”
> also from
> Broadcom?
Drivers b43 and ssb are part of the kernel. They do not come with the firmware,
but with the kernel package itself. To be part of the kernel, the driver must be
open source. The wl driver is not. As open-source drivers, ssb and b43 have been
reverse-engineered from Broadcom drivers, but they have no connection with
Broadcom. In fact, Broadcom placed as many restrictions as possible to prevent
the development of those drivers. Most of those restrictions are irrelevant
except for prohibiting the redistribution of firmware.
> I would be helpful for some concrete proposals what to do now. There is
> a lot
> of noise on the Internet on these things, but I don’t get any coherent
> picture
> (as usual in such things). There was a discussion August 2010 for 11.3,
> but there are no conclusions or results. Just the thing that there are
> various
> possibilities.
You may choose either ssb/b43 or wl. If you select the latter, then every time
your kernel gets updated, then you will need to be sure to also get the
appropriate -kmp- wl package, otherwise your wifi will break. Loading wl also
taints your kernel because a closed-source binary is being loaded. There may not
be any effect; however, if you discover a kernel bug that causes an oops, most
kernel developers will not even look at the problem until you repeat it without
any tainting of the kernel.
As a developer of b43, I refuse to use wl.