I upgraded to nvidia driver 4496, and then switched from xfree to console (using control-shift f1), and found my console to be totally corrupted. It was filled with flashing blinking multi-colored garbled text. I then masked nvidia-kernel and nvidia-glx version 4496, and downgraded to the previous version. This fixed the problem. Reproducible: Always Steps to Reproduce: 1. Run gentoo linux kernel in framebuffer mode with bootsplash, and nvidia drivers version 4363 2. Upgrade to nvidia driver version 4496 Actual Results: The console was garbled, and unusable. Expected Results: Correctly displayed the console Running kernel 2.4.22 nvidia geforce 2 gts pro (made by asus)
what frame buffer driver were you using ?
I am using the standard vesa frame buffer driver, here is an excerpt from dmesg: vesafb: framebuffer at 0xf0000000, mapped to 0xe080d000, size 3072k Console: switching to colour frame buffer device 128x48 fb0: VESA VGA frame buffer device
Could we please have an update please. Also are you using any form of TV out? Can you please try disabling this if so and see if the problem persists.
I'm not using TV out functions. I am still using the older nvidia drivers (4363) without any problems. I assume this is a problem with the 4496 nvidia driver itself. Perhaps my card is too old, and they decided not to care about it anymore? I can only hope future versions of the driver some how magically fix this problem. One of the reasons I posted this to the gentoo bugzilla (aside from the fact that I use gentoo), is that I was hoping the gentoo team might put some pressure on nvidia to fix the problem. Certaily this would go over much better after have being tested by a support team for a project as large as gentoo, rather than a single individual emailing them to complain about their linux console in frame buffer mode.
Same problem for me. Except I'm running Kernel 2.6.3-gentoo-r1 same frame buffer driver Nvidia drivers 5328 and 5336 failed. Will try 4363... Hope it works... And it doesn't come from your geforce 2 gts pro being too old because i've got a geforce ti 4400 and it doesn't work better !
Another confirmation. Running the default framebuffer (not sure what it is), AMD64, kernel 2.6.3 r2, 5700 FX. Pressure on nvidia to fix this would be wonderful.
Forgot to say, I'm using using the 5332r1 drivers.
Created attachment 27000 [details] working 4363 ebuild for 2.6 kernels. Don't forget the patch!
Created attachment 27001 [details, diff] NVIDIA kernel 4363 patch (used by the ebuild)
I have the same problem, using a Geforce 4MX. The problem applies to all NVIDIA-drivers >= 4496. The 4363 driver works perfect, but will not compile on 2.6 kernels. I created two attachments, an ebuild and patch for the 4363 kernel-module. Using these you can succesfully compile this one. - Point the PORTDIR_OVERLAY variable in /etc/make.conf to /usr/local/portage (example). - Copy the media-video/nvidia-kernel directory to ${PORTDIR_OVERLAY} - Copy the .ebuild file to the ${PORTDIR_OVERLAY}/nvidia-kernel/ - Copy the .patch file to the ${PORTDIR_OVERLAY}/nvidia-kernel/files/ - emerge =nvidia-kernel-1.0.4363-r3
Ofcourse you should use ${PORTDIR_OVERLAY}/media-video where I wrote ${PORTDIR_OVERLAY}. But you got the point, didn't you :-)
I cant reproduce this under amd64 or x86 using either of the Xfree branches. That is not to say I havent seen this happen before but that was usually with the drivers prior to version 4363. Is this still a problem with the 533* drivers (5336 for x86 and 5332 amd64)?
These are the steps needed to reproduce it: - Enable the following kernel options: Code maturity level options ---> [*] Prompt for development and/or incomplete code/drivers Processor type and features ---> [*] MTRR (Memory Type Range Register) support Graphics support ---> [*] Support for frame buffer devices [*] VESA VGA graphics support Console display driver support ---> [*] Video mode selection support <*> Framebuffer Console support - Install the latest nvidia-drivers (1.0.5336 for me) ACCEPT_KEYWORDS="~x86" emerge -v nvidia-kernel ACCEPT_KEYWORDS="~x86" emerge -v nvidia-glx - Append the following options to your kernel: video=vesafb:ywrap,mttr vga=0x317 I also tried to disable agpart in my kernel, and added the NvAGP option to my XF86Config. After that, i also tried to use _only_ vga=0x317 as boot parameter. Both didn't work. You'll see that this problem is related to the combination framebuffer & nvidia-kernel (& graphics card?). Often the bootsplash scripts/patches are blamed for it... I really don't understand that you cannot reproduce it. I read a lot of postings from people with the same problem. By the way, I created a photo of my corrupted screen: http://download.delete-it.nl/files/nvidia-corrupt.jpg Sander de Leeuw
I just installed my first gentoo on a Precision m60 (that uses a Nvidia QuadroFXGo 700 card), and XFree was either locking (on a white/black/red/blue slowli alternating fullscreen) with the nvidia driver (the default 4496), or showing a very garbled screen that corrupted the frame buffer (after ctrl-alt-backspace) with the nv driver. I'm using a genkernel kernel. I followed the procedure indicated in comment #10, and now the nvidia driver works perfectly.
As crazy as this might sound... I think this happens when the actual text output to the console from Xserver causes the framebuffer to scroll. Can you all please test this... ie 1) Start up an Xserver and kill it / switch vts right away, does the problem persist 2) Start up an Xserver and somehow ensure that your console will be flooded by messages (just add heaps of bad font entries to XF86Config or xorg.conf), then switch or kill the server what happens? I was able to make this happen randomly on amd64 with 5332 (at least I thought it was random until I noticed this trend, anyone else see this happening)?
I just tried what was suggested in comment 15, and I couldn't reproduce that. I redirected both stdout and stderr so the framebuffer console would not scroll, and I ended up with the same behavior. The best I can say is that you are killing the xserver before the nvidia drivers actually take over, and thus avoiding the problem.
I had a very same situation, but on debian.. I wrote this here, because same problem, and I have one solution. I use nvidia 1.0.5336 driver, on a debian sarge (2.4.25) dist, with GForce 4 MX440 64Mb. When I was updated to this driver, I lost the console screen, just as mentioned above. I was trying to find out what casued it.. and finaly I found this: My lilo.conf file was using vga=8 and if I change it back to normal, everything is fine. so my "solution" - more like workaround - for this problem was: edit the /etc/lilo.conf file to vga=normal and run /sbin/lilo After reboot, my consol works fine! (I did not try other values.. )
Can someone setup ssh on their box and ssh to it after exiting X and see what kernel messages (dmesg) are dumped when leaving X / entering X For this too work best, please do the following $ su // need root for a second # rmmmod nvidia # modprobe nvidia # exit // back to used $ startx --- do whatever you do and then exit X SSH in from another machine, run dmesg and post everything after the nvidia insert messages.
I might have a solution for this bug: When I did "emerge system" today I encountered this bug. I downgraded to nvidia kernel 4363 with the ebuild given here an the bug was gone. I emerged nvidia kernel 4496 again to see the dmesg output as suggested. What i saw after rmmod/modrobe nvidia was: "0: nvidia: loading NVIDIA Linux x86 nvidia.o Kernel Module 1.0-4363 ..." Then i noticed that there were two modules in /lib/modules/2.6.5-gentoo-r1/video/: nvidia.o and nvidia.ko. I tried to unmerge nvidia-kernel but the modules still remained. I deleted them, emerged nvidia-kernel 4496, modprobed nvidia, started x und now it works (dmesg shows that 1.0-4496 is loaded). There is only nvidia.ko in the modules directory now.
Try 6106 on bug #55714
The bug still exists with the 6106 drivers.
I should clarify. When I first subscribed to this bug, the framebuffer was indeed corrupted. After re-installing my gentoo system about 3 weeks ago, the 'corrupted' framebuffer became a completely blank framebuffer. This could indicate that something was fixed in the meantime, and I'm now suffering from bug http://bugs.gentoo.org/show_bug.cgi?id=54523. Or it could be a different symptom of the same bug. I'll test this tonight and post my findings.
Please make your way over to the folowing site: http://whitebunny.demon.nl/hardware/chipset_nvidia.html that I mentioned in the folowing post: http://forums.gentoo.org/viewtopic.php?t=162955&highlight= altho i misquoted my new BIOS version in the post the process still worked. I was using 4.17.45 on a GeForce4 MX440 AGP 4x. I then upgraded to 4.17.69. I know your thinking that a .24 VideoBIOS bump may not be worth it, but I can now report that the green splotchies and the garbled text are GONE!!! I can now perfectly switch from X to a terminal with no problem. As a side note, I'm also happy to report that the 6106 kernel works flawlessly on gentoo-dev-sources with 4k stacks. At the time of this comment my kernel is 2.6.7-gentoo-r11
Another confirmation with the 2.6.7-gentoo-r11 kernel and the latest nvidia 6106. My framebuffer also got corrupted :(
Please test this with 6111 which just went into cvs. Report back on success or failures. Thanks.
I fixed my framebuffer corrupotion by adding Option "IgnoreDisplayDevices" "TV" in my xorg.conf. Can you try this and report if it works or not?
Bump. Does this happen for 6111 or 6629?
I had no problems before trying 1.0.6629 on amd64. Now with it, the framebuffer gets broken. When the X-Server starts the DFP gets a noisy pattern with some sort of feedback. Then the NVidia Logo comes up and everything is fine. But if I want to switch back to the framebuffer this feedbacky pattern comes up back and the console is invisible. If I enter login and password blindly I can enter commands that work, but I do not see anything more than this pattern. If I then switch back to VT7 I get my wonderful working X back. If I then switch again back to framebuffer console, I get pink blinking rectangles with sometimes green characters in them on my screen. If I then enter some characters by Keyboard these Noise-Characters change in a more random manner.
what's the latest on this, people?
I haven't had any problems in quite some time. I don't recall which update fixed it.
I still had corruption on 6629 when using TwinView, but haven't seen it at all (yet) with 7167. That said, it's always been very intermittent for me. But after another week without issues I would be happy to agree that it seems fixed.
Pretty much a given issue if using vesafb. vesafb-tng works fine here though.
Ok, then I'm going to mark this fixed since the latest builds from nVidia seem to fix this problem. If that is not the case, then please reopen
