Comment 370 for bug 541511

@chis

I've installed the module from your PPA on a cuple of new-installed boxes, and unfortunatly this seems to makes things even worse than the drivers released within Lucid until now.

I'm able to get the boxes running installation and then running Lucid from harddisk, but ONLY by supplying the extra kerne parameter "i915.modeset=1" in Grub in both cases. Then after opgrading the installation from your PPA, I'm still not able to boot without this parameter - the boxes completly freeses just before X starts without it. Then supplying this kernel parameter while using the modules from your PPA, I got the boxes up and running, but with som odd experinces in the desktop enviroment (KDE) - like menu entries suddenly changing colours and making them nearly dissapear into the menu-background. Videoes now "playes" without imidiate X crash though, but without picture ever seen (black screen in the gecko-mediaplayer plugin). And after I've stopped the player X and closed the browser, X sometimes crashes completely.

But I have a question though, why can I make an unmodified Lucid run perfect on those boxes by using the kernel image 2.6.32-17, 2.6.33-5, 2.6.34 or now 2.6.35 from the Ubuntu Mainline archive (http://kernel.ubuntu.com/~kernel-ppa/mainline/) ?. No kernel parameter needed, compiz and video playing runs fine and even functions like Brightness control, OSD (found ind KDE) for sound and brightness among other things (not seen working with the released Lucid kernel ever) just works. And then why was KMS disabled for i855 in_last_minute_before_release, just to make Lucid uninstallable to most ordenary i855 user, and making a lot of fuss when they try to upgrade (is this good for a LTS rel.) ?.

I can't help wonder why workarounds for a bug, witch to me seems introdused/exposed by the Ubuntu specifik kernels (according to above), are hunted so bad in modules eksternal to the kernel. The real problem must reside in backporting things from the 2.6.33 kernel to the 2.6.32 kernel as both mainline kernel 2.6.32 and 2.6.33 seems working okay - and more people have in different i855 related bugreports replied that they use those kernels as a workaround, some from the backported kernel PPA (https://launchpad.net/~kernel-ppa/+archive/ppa) though.

It would be great if Maverick isn't going suffer this fate, and it have until now (Alpha 1, 2 & 2+) been running okay on one of those (i855) boxes with its default 2.6.35 kernel...