[BUG] NVIDIA 295.09 switching to TTY, monitor turns off and crashes X

[ioni]

Hi,

i noticed instability when using nvidia 295.09 driver on nvidia 8600M GT, linux 3.1.6, on a dell 1520 laptop
01:00.0 VGA compatible controller: nVidia Corporation G84 [GeForce 8600M GT] (rev a1)

bug1:
After login into gnome and trying to switch to tty1/2/3/etc monitor is turned off and I can only go back to tty7 to turn it on again.

bug2:
Using the desktop for some time and X crashed while browsing pictures, gdm respawn and i cannot switch to tty because of bug1.
Dmesg output has and Xorg.0.log log crash
NVRM: Xid (0000:01:00): 6, PE0001

[ 3538.081] 0: /usr/bin/Xorg (xorg_backtrace+0x37) [0x81c0117]
[ 3538.081] 1: /usr/bin/Xorg (0x8048000+0x17bc8a) [0x81c3c8a]
[ 3538.081] 2: (vdso) (__kernel_rt_sigreturn+0x0) [0xb77a140c]
[ 3538.081] 3: /usr/lib/xorg/modules/drivers/nvidia_drv.so (0xb4a30000+0xd3e59) [0xb4b03e59]
[ 3538.081] Segmentation fault at address 0xb51a4000

[ultr]

Didn't you mean the 295.09 driver?

I haven't noticed this thread because of this and created a new one: http://www.nvnews.net/vbulletin/showthread.php?t=171448

[ioni]

Originally Posted by ultr

Didn't you mean the 295.09 driver?

I haven't noticed this thread because of this and created a new one: http://www.nvnews.net/vbulletin/showthread.php?t=171448

yes, is 295.09

[ioni]

ping, i'll be very sad if a new nvidia is released without this crucial issue not being visited.

[olifre]

Using 295.09 on Gentoo Linux 64 bit here, GeForce 8600M GT.
Console-switching is working fine for me, kernel 3.0.6 with tuxonice-patches, framebuffer is uvesaFB.
You appear to be using vesafb, both of you. Can you check whether the problem persists with uvesaFB (or on an UEFI-system, with efifb)?
How is your screen connected? I'm using a laptop here (LPL / DFP-0).

[ultr]

This problem still exists for me with 295.20 (i386):
http://www.nvidia.com/object/linux-d...20-driver.html
I have geforce 8600GT.

@olifre
I use Debian Wheezy's default framebuffer. Can't check it now - writing from mobile phone.
Display is connected with DVI (PC).

[olifre]

@ultr: I did not test 295.20 (yet) due to lack of time, but I guess it's a problem that occurs with vesafb. If I had more time to check, I would recompile my kernel and see whether it fails for me too with vesafb.
Uvesafb is using userspace instead of Video BIOS, so if it really works with uvesafb and fails with vesafb, we have 'pinned the error down' and nVidia should be able to reproduce. Although I guess if it fails for the 8600M GT with vesafb, they should have be able to reproduce it with any default setup.

[ioni]

is not about what framebuffer i use, i tried vesa, uvesafb and none. all have the same issue. I do have thread that has more infos

http://www.nvnews.net/vbulletin/showthread.php?t=173214

[olifre]

Originally Posted by ioni

is not about what framebuffer i use, i tried vesa, uvesafb and none. all have the same issue. I do have thread that has more infos

http://www.nvnews.net/vbulletin/showthread.php?t=173214

Thanks for the further information.
Then I am sorry I'm out of ideas, hopefully nvidia can provide help .
My working hardware is a Dell Vostro 1500 by the way, so it's a bit older (but with same graphics adapter).
If you have another machine you might want to use it to ssh into your machine and check whether you can still use "chvt" (or login blindly and try).

[artem]

I can confirm bug N1 with NVIDIA 8800GT, an LCD display connected to DVI, and Linux kernel 3.3-rc4

If nvidia-bug-report is required I can post it.

[sandipt]

Internal bug to track this issue : 922163 . Console restore broken.

[ioni]

Originally Posted by sandipt

Internal bug to track this issue : 922163 . Console restore broken.

any news about this bug? When is going to be fixed?