[Mageia-dev] kernel 3.0 is a big mistake in cauldron

Sun Jul 17 16:37:26 CEST 2011

+>and both i586 and x86_64 kernels were tested on a i7-860 workstation and 
+>on a Acer TravelMate 5720  laptop before submitting to the buildsystem...
+
+
+tmb, for the Wi-Fi issue, I identified the culprit.
+
+With the 2.6.38 kernel, dkms-broadcom-wl was useless (Broadcom STA
+
+is a hybrid driver, including a binary blob from upstream, but in Mageia 
+
+dkms-broadcom-wl is not like Ubuntu's broadcom-wl, it still asks for
+firmware, which is stupid, as STA is redistributable; eventually, I had to use 
+
+fw-cutter and to populate /lib/firmware/b43 with *.fw, so that BCM4311 rev.1
+works w/o dkms-broadcom-wl!).
+
+With the 3.0.0 kernel, dkms-broadcom-wl tries to build a module after booting,
+but it doesn't work at all (I haven't checked what it does, whether it has the binary 
+
+blob or not, if it logs messages, etc.), and it blocks b43. 
+
+Uninstalling dkms-broadcom-wl still didn't fix the issue, but again, I have not investigated
+-- maybe there are some remnants or smth that impedes b43.
+
+Still, with the 3.0.0 kernel, my TravelMate 5310 fails to shutdown or reboot, when told
+so kdeinit4 takes 100% CPU and the system becomes unresponsive. Back to 2.6.38,
+everything is smooth again.
+
+Experimenting with various packages and their breakages is one thing, but the Linux
+kernel is my personal enemy, I hate the too many changes between one release and 
+
+the next one, it's practically impossible to design & run any relevant regression tests. 
+
+If I were in charge with any distro's policy, I would be particularly conservative
+when comes to the kernel (e.g. slackware-current is still on 2.6.38.7)
+
+>There is no forcing...
+>the old kernel is still installed so you can use it if you need.
+
+>
+>And if you dont like automatic kernel upgrades, uninstall kernel-*-latest
+
+
+I did so, but w/o kernel-*-latest I suppose I won't get any kernel update at all?!
+This is not what I intended...
+
+
+>its always good to know my work is appreciated...
+
+
+Your work in patching and packing and so on is one thing, and the policy of
+when the kernel goes and where it goes and what kind of kernel-*-latest 
+
+metapackages exist is a different thing. I still believe that, even in cauldron,
+there should be 2 distinct kernel "branches", not a unique kernel-*-latest path.
+(Maybe I'm influenced by the conservative way Slackware is approaching the kernels.)
+(Or maybe I'm just getting old.)
+
+R-C aka beranger
+
+