Framebuffer gets messeed up as soon as xorg loads. [solved]

[woodwizzle]

I got genkernel installed and running fine durring boot. However once x.org loads on tty7, if I switch to any of the other terminals I find that the framebuffer has offset the screen so that parts of what should be visible are, for lack of a better explaination, rendered outside of my monitor's viible area. I've tried different framebuffer resolutions, none fix the problem however each resolution sems to offset the framebuffer completely new ammounts and in different directions. I'm running x86_64 with an nvidia 8600gt. The offset happenes right about the them the Nvidia logo shows up.

here is my grub.conf entry:
title=Gentoo Linux 2.6.22-r2
root (hd0,0)
kernel /boot/kernel-2.6.22-gentoo-r2 root=/dev/hda3 splash=silent,fadein,theme:livecd-2007.0 vga=0x318 video=vesafb:mtrr:3,ywrap quiet CONSOLE=/dev/tty1
initrd (hd0,0)/boot/fbsplash-livecd-2007.0-1024x768

I'm at a total loss! =) Please help.
Corey

[dusik]

Which version of nvidia-drivers are you using? I've had that problem with version 100.14.09 on my 8800 GTS.

Just unmask and install x11-drivers/nvidia-drivers-100.14.11. Let us know if that doesn't fix the problem.

[hansolocg]

I just might add here, that i had exactly the same problem until 3 minutes ago . It also put some ugly colorful lines on the screen, just before X starts, probably connected to the breakage of the console.
But the new driver fixes it perfectly.

[woodwizzle]

Thanks! Worked like a charm!

[pjv]

Ì might have something similar. However the new drivers don't fix it for me. I have a GeForce 8600 GS.

I first was looking at this problem thread because I was using vesafb-tng and my screen got messed up when leaving X (I don't see any of you using vesafb-tng so you might have missed this symptom). The main symptom is this: Once I've started X (in any way, but only with the nvidia binaries), I cannot do Ctrl+Alt+Fx or even logout. As soon as X closes (after a short black pause) the screen shows horizontal grey lines at random different intervals. The are not very contrasted (almost like normal terminal text). Almost like spectroscopic linepatterns. They are on a background of similar but slightly coloured and vertical lines. In ten seconds time all the lines slightly fade out to just a blank illuminated screen (a little bit more on the edges). Regular garbage I can still type and reboot by login in and typing reboot. X even doesn't succeed in starting itself a second time (when logging out to gdm).

But now that I think of it I am also seeing strange things BEFORE X gets started: Right before X starts I first see a quick flash of a black screen with a horizontal bar of garbage (pixels in different colours). Were you seeing that too? Next, about when the nvidia logo (but i disabled it) should come up, I see a black screen with in the bottom right corner something that looks like a part of a previous screen. It could be the top left of a moved screen. It's 70% wide but only 2% high in the bottom right corner. It's a clear picture and it actually sometimes looks like a piece of my gnome panel. Were you seeing that too?

In the framebuffer afterwards I don't see parts of a moved screen, but it could be that it has been moved all the way of the screen. What were you seeing at the parts the moved screen wasn't covering? Was anybody using anything else than the plain vesafb? Could you switch back to X after the moved screens?

Just trying to rule things out, thx.

[dusik]

pjv, I'm pretty sure you're looking at a different problem than the one we had, but to me it also smacks of driver misbehaviour. According to the documentation I'm looking at ( $bzless /usr/share/doc/nvidia-drivers-*/NVIDIA_Changelog.bz2 ), it looks like support for the 8600 series was added in 100.14.09, so you don't have much choice in nvidia-drivers versions.

You're right, I'm using the original vesafb, since I'm on x86_64 hardware (so I can't test with -tng). But maybe you can switch to vesafb to test? Also, maybe you could try using nvidiafb (I remember it being problematic last time I tried it, but it was a long long time ago).

BTW, at the top of the changelog, it's indicated that since 100.14.09: