Page 1 of 2 12 LastLast
Results 1 to 10 of 14

Thread: Logout freeze using Nvidia Proprietary driver (latest and even older versions)

  1. #1
    Join Date
    Apr 2012
    Posts
    26

    Default Logout freeze using Nvidia Proprietary driver (latest and even older versions)

    I've built a HTPC running Linux (Ubuntu 12.04) w/ AMD Liano processor and a Fusion motherboard with HDMI. Initially I was using AMD processor and Catalyst driver output to HDMI.

    Then I realized that video decoding acceleration is poor with this combination. I've installed vaapi acceleration packages but XBMC refused to work with it. Not to mention even flash acceleration is not working as it requires vdpau.

    So I brought a Nvidia GT 520 card for this. Lucky the card is quiet cheap

    Once I installed the card with Nvidia proprietary driver, everything seems to now well with output going to the HDMI in Nvidia card. Sound works great, video acceleration works as well with vdpau as I can see Chrome Flash youtube is now consuming very few processing power. Of course I've the blue tint problem with the flash but I managed to fix it with the /etc/adobe/mms.cfg thing.

    However a problem happens.

    Whenever I try to logout (either my Unity session or a delicated XBMC session), the system freeze with some console message (I can't recall exactly "SYSV service stop......" or "using libgcrypt-1.5 ...." and I cannot get back to the login screen (I'm using ubuntu so it should be lightdm).

    I tried to ping the machine and SSH to it. Ping responses with either problem with connection or reply with a very long latency. SSH doesn't work (I think probably due to long network latency). Seems like the machine is still on but with a very high CPU utilization.

    Keyboard / mouse is not responding (alt-ctrl-f1 doesn't work) and I've to hard reset the machine.

    Anyone have similar issue and any fix? Now I've reverted back to AMD (and open source radeon driver which seems to work pretty well). Really hope that I can get some video decoding acceleration.

  2. #2
    Join Date
    Oct 2011
    Location
    Toruń, Poland
    Posts
    160

    Default

    Do you insist on using Ubuntu? Try using a different distribution. Ubuntu family tends to have tons of small and weird bugs, thanks to their custom patch frenzy. I found Fedora to be user friendly equally to Ubuntu. If you go down Fedora route, don't forget about RPMFusion repository for non-free software. Also, take a look at "Fedora spins" for a choice of environments different than GNOME 3.

  3. #3
    Join Date
    Apr 2012
    Posts
    26

    Default

    The problem doesn't happen when I am using the ATI on-board card, either the open source or proprietary driver. So I don't know if it is a problem with nvidia, ubuntu or anything else. Will probably give fedora a try but would still like to know how it can be fixed here.

  4. #4
    Join Date
    Oct 2011
    Location
    Toruń, Poland
    Posts
    160

    Default

    Quote Originally Posted by mikejuni View Post
    The problem doesn't happen when I am using the ATI on-board card, either the open source or proprietary driver. So I don't know if it is a problem with nvidia, ubuntu or anything else. Will probably give fedora a try but would still like to know how it can be fixed here.
    The warning that you got could be useful to track the issue. Not sure however, where Ubuntu stores its logs.

  5. #5
    Join Date
    Apr 2012
    Posts
    26

    Default

    I've reinstalled everything hoping to find out little bit more.

    Now I found out that the issue generally occurs after logout, when the x server is shutdown and reloaded for lightdm. The I suspected lightdm issue and use gdm instead.

    Still the same issue occurs.

    The xorg log immediately before freeze indicates xorg and nvidia driver unloaded successfully. Then the system seems to hang. I can actually ssh to the server and the response is extremely slow, which I think indicates some program locks the system with some kind of infinite loop.


    Other than that doesn't seem to have any unusual log. The problem occurs only with the nvidia restricted driver or the driver I downloaded from nvidia directly. Nouveau doesn't have the same problem, neither do the ati card (radeon or catalyst).

    Now I am sticking to radeon driver. The catalyst driver have other problems (xbmc not able to play video with vaapi enabled)

  6. #6
    Join Date
    Aug 2007
    Posts
    6,623

    Default

    You can not play h264 l5.1 with xvba-video (nor would it work with xbmc-xvba). I requested a hotkey to disable accelleration without going thru the menus but nobody cared

  7. #7
    Join Date
    Apr 2012
    Posts
    26

    Default

    Yes, and that's why I reverted to AMD open source driver. My system is APU based so it is on board.

    Actually performance is quite ok for HTPC (I can decode blu-ray 1080p without problem), just no hardware video acceleration so no fancy things such as motion compensation or other post-processing.

    Really hope I can get the nvidia driver working so that I can use GT520 for all these with VDPAU. The application support is much wider (flash, XBMC, VLC and almost everything else).

  8. #8
    Join Date
    Apr 2012
    Posts
    26

    Default

    I just want to update my experience and share with everyone here about the issue.

    My card is a ASUS GT520 (passive cooling). The kernel freeze happens with 295.40 driver.

    Today I installed the 295.49 driver, the "freeze" had disappeared. I got response from the machine and SSH works fine now.

    However, still having Xid messages in the kernel log and still X wouldn't start up again after I logoff, kill / restart lightdm (am using Ubuntu) and the error messages are consistent.

    I tried 302 beta driver however there is no HDMI sound coming out. Hence it is not an option. I tried to ask for help in nvnews.net and simply cannot register in the forum (the register link is broken).

    Just want to know if there are any other help channels?

  9. #9
    Join Date
    Aug 2007
    Posts
    6,623

    Default

    If you use kdm try this:
    Code:
    sed -i 's/^[#]*\(TerminateServer=\).*/\1true/' /etc/kde4/kdm/kdmrc
    Otherwise look for a similar option. I have got no HDMI issues with kernel 3.2 and 302.xx or 295.xx using pa (and nv 405 (rebranded 210). Maybe you use the wrong system
    Last edited by Kano; 05-11-2012 at 01:23 AM.

  10. #10
    Join Date
    Apr 2012
    Posts
    26

    Default

    Don't think it is a config issue since I have no hdmi issue with 295.49 driver. Just that 295.49 driver I annoy have X restarted.

    I have similar finding reports in nvnews.net with exactly same configuration gt520

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •