On a machine with two 4850 connected in CrossFireX, loading the kernel module radeon with modeset=1 makes the screen go blank. The machine is still alive as I can access it via ssh. Didn't tried with a single card in the machine. Reproducible: Always Actual Results: The screen connected on the machine becomes blank (a signal seems to be detected as the led on the front panel of the screen does not blink) and the following appears in /var/log/messages :
Created attachment 213075 [details] /var/log/messages when inserting the radeon kernel module with modeset=1
http://bugs.freedesktop.org/show_bug.cgi?id=25682
Thanks for reporting upstream. We'll follow that bug.
If I can do something to make things advance, please, let me know.
You could try the latest drm-next kernel from "airlied" on kernel.org [1], he's the DRM maintainer. Maybe the bug is already fixed there. And if not, upstream will ask you to test that kernel anyway... Thanks [1] http://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git with branches "drm-next", "drm-linus", "drm-radeon-testing", etc.
Tested again with Linux 2.6.33-rc4 : All quiet on the western front but a slight difference (my monitor's LED now blinks when the screen becomes blank). Joined a new log excerpt in the upstream report.
Tested with 2.6.33-rc7, I still get a blank screen :-(
Ping the upstream bug, that's where the fix is going to come from. Don't hesitate to reopen this bug once upstream fixes this issue. Closing upstream. Thanks
A follow-up on this one : the problem is solved ! Simply upgrade the kernel to Linux 2.6.35 (-rc3 at this date), install the latest revision of x11-drivers/radeon-ucode and you should see a nice framebuffer with a lot of text inside :-) Also, note that if you don't use initramfs images, you *must* set 'radeon' as being a module (it requires to load several firmware files). Note : Blank screen on every releases prior to 2.6.35