fbcon
10-11-07, 01:30 PM
Problem: Twinvideo mode, fbcon console rotation, xrandr -o left (X pivot), in X => Hangs/freezes system completely on x86 kernel and x86_64 kernel.
How to reproduce: vanilla kernel (linux, 2.6.20-22), "fbcon=rotate:3" passed to kernel (console rotation) , pivot screen with xrandr "xrandr -o left", use nvidia binaries (also latest) for x86 and x86_64, start X session.
Symptoms:
Either by going idle or in active use. Composite expediates time-to-hang. System will always hang at some point, guaranteed. Some mouse control for 5 seconds (no button response), followed by artefacts. Then complete system hang.
Hardware: G6600, 256mb. Problem not to be blamed on hardware failure. I traced the problem back to that "fbcon: rotate=3". If turned off => System stable (As it should be).
Anyone know how/have the time to diagnose this further?
I have turned my screens to the side, because this maximizes my use of the screens, since I'm a programmer.
I haven't gotten time to produce a kernel dump yet. But I'm sure this can be reproduced by someone else as well.
Thanks in advance.
Update: I thought, it was the pivot parameter, but the system still managed to crash nvidia and freeze.
Update: currently testing kernel 2.6.23, checking if there's any change.
Update: testing vanilla 2.6.23 seems to be stable, even turned on composite
Update: now patched with rt1 and tuxonice, testing stability
How to reproduce: vanilla kernel (linux, 2.6.20-22), "fbcon=rotate:3" passed to kernel (console rotation) , pivot screen with xrandr "xrandr -o left", use nvidia binaries (also latest) for x86 and x86_64, start X session.
Symptoms:
Either by going idle or in active use. Composite expediates time-to-hang. System will always hang at some point, guaranteed. Some mouse control for 5 seconds (no button response), followed by artefacts. Then complete system hang.
Hardware: G6600, 256mb. Problem not to be blamed on hardware failure. I traced the problem back to that "fbcon: rotate=3". If turned off => System stable (As it should be).
Anyone know how/have the time to diagnose this further?
I have turned my screens to the side, because this maximizes my use of the screens, since I'm a programmer.
I haven't gotten time to produce a kernel dump yet. But I'm sure this can be reproduced by someone else as well.
Thanks in advance.
Update: I thought, it was the pivot parameter, but the system still managed to crash nvidia and freeze.
Update: currently testing kernel 2.6.23, checking if there's any change.
Update: testing vanilla 2.6.23 seems to be stable, even turned on composite
Update: now patched with rt1 and tuxonice, testing stability