Results 1 to 4 of 4

Thread: Nvidia GPU Lockup - Window Manager Freezes Randomly, Xorg, 100% CPU, Proprietary Drv

  1. #1
    Join Date
    May 2005
    Beans
    153

    Nvidia GPU Lockup - Window Manager Freezes Randomly, Xorg, 100% CPU, Proprietary Drv

    Hi,

    I'm not a "great" troubleshooter, but I thought I would just mention this, as it was a simple fix for me, but the solution was*very* obscure and challenging for me to locate by just searching UbuntuForums and the web in general.

    There seems to be a certain number of Nvidia GPU users who are having, what appear to be random, "GPU Lockups". After searching around,there are lots of suggestions regarding, nouveau vs. ubuntu-repo-proprietary vs. newest-nvidia-website-proprietary driver search being the cause of the lockups.

    Similarly,there are a lot of opinions regarding the compositing function (thatis, turning off composting) of your window manager (compiz, marco, orother, depending on your window management environment) being the culprit and losing composting may fix your issues.

    And,these may be the correct suggestions for many people. It was different for me though. First, here are some of my system details (Iam running the most current proprietary driver for my GTX 750 from the Nvidia website (NVIDIA-Linux-x86_64-384.59) ):

    Code:
    Summary
    -------
    
    
    -Computer-
    Processor: 8x AMD FX(tm)-8320 Eight-Core Processor
    Memory: 16332MB (8495MB used)
    OperatingSystem : Ubuntu 16.04.3 LTS
    UserName : promet (promet)
    Date/Time: Fri 18 Aug 2017 08:40:34 PM CDT
    -Display-
    Resolution: 1920x1080 pixels
    OpenGLRenderer : GeForce GTX 750/PCIe/SSE2
    X11Vendor : The X.Org Foundation
    -Multimedia-
    AudioAdapter : HDA-Intel - HDA ATI SB
    AudioAdapter : HDA-Intel - HDA NVidia
    -InputDevices-
     PowerButton
     PowerButton
     LogitechUSB Keyboard
     LogitechUSB Receiver
     LogitechUSB Keyboard
     LogitechUSB Receiver
     AresonUSB Device
     AresonUSB Device
     SIwilress mouse & keyboard
     SIwilress mouse & keyboard
     EeePC WMI hotkeys
     HDAATI SB Front Mic
     HDAATI SB Rear Mic
     HDAATI SB Line
     HDAATI SB Line Out Front
     HDAATI SB Line Out Surround
     HDAATI SB Line Out CLFE
     HDAATI SB Line Out Side
     HDAATI SB Front Headphone
     HDANVidia HDMI/DP,pcm : 3=
     HDANVidia HDMI/DP,pcm : 7=
     HDANVidia HDMI/DP,pcm : 8=
    -Printers-
    Noprinters found
    -SCSIDisks-
    ATAADATA SP900
    ATAADATA SP900
    WDMy Passport 25E1
    WDSES Device
    WDCWD50 00BEKT-75KA9T0
    
    
    OperatingSystem
    ----------------
    
    
    -Version-
    Kernel: Linux 4.4.0-92-generic (x86_64)
    Compiled: #115-Ubuntu SMP Thu Aug 10 09:04:33 UTC 2017
    CLibrary : Unknown
    DefaultC Compiler : GNU C Compiler version 5.4.0 20160609 (Ubuntu5.4.0-6ubuntu1~16.04.4) 
    Distribution: Ubuntu 16.04.3 LTS
    -CurrentSession-
    ComputerName : shivux
    UserName : promet (promet)
    HomeDirectory : /home/promet
    DesktopEnvironment : MATE (mate)
    -Misc-
    Uptime: 22 hours, 16 minutes
    LoadAverage : 2.25, 2.13, 2.06
    
    
    KernelModules
    --------------
    
    
    -LoadedModules-
    input_leds: Input -> LEDs Bridge
    joydev: Joystick device interfaces
    msr: x86 generic MSR driver
    xt_multiport: Xtables: multiple port matching for TCP, UDP, UDP-Lite, SCTP andDCCP
    zram: Compressed RAM Block Device
    lz4_compress: LZ4 compressor
    binfmt_misc
    snd_hda_codec_hdmi: HDMI HD-audio codec
    nvidia_uvm
    kvm_amd
    kvm
    irqbypass: IRQ bypass manager utility module
    crct10dif_pclmul: T10 DIF CRC calculation accelerated with PCLMULQDQ.
    crc32_pclmul
    snd_hda_codec_realtek: Realtek HD-audio codec
    ghash_clmulni_intel: GHASH Message Digest Algorithm, acclerated by PCLMULQDQ-NI
    snd_hda_codec_generic: Generic HD-audio codec parser
    aesni_intel: Rijndael (AES) Cipher Algorithm, Intel AES-NI instructionsoptimized
    snd_hda_intel: Intel HDA driver
    aes_x86_64: Rijndael (AES) Cipher Algorithm, asm optimized
    lrw: LRW block cipher mode
    gf128mul: Functions for multiplying elements of GF(2^128)
    glue_helper
    snd_hda_codec: HDA codec core
    ablk_helper
    eeepc_wmi: Eee PC WMI Hotkey Driver
    cryptd: Software async crypto daemon
    asus_wmi: Asus Generic WMI Driver
    sparse_keymap: Generic support for sparse keymaps
    snd_hda_core: HD-audio bus
    video: ACPI Video Driver
    mxm_wmi: MXM WMI Driver
    snd_hwdep: Hardware dependent layer
    snd_pcm: Midlevel PCM code for ALSA.
    serio_raw: Raw serio driver
    fam15h_power: AMD Family 15h CPU processor power monitor
    snd_seq_midi: Advanced Linux Sound Architecture sequencer MIDI synth.
    snd_seq_midi_event: MIDI byte <-> sequencer event coder
    snd_rawmidi: Midlevel RawMidi code for ALSA.
    edac_mce_amd: AMD MCE decoder
    edac_core: Core library routines for EDAC reporting
    k10temp: AMD Family 10h+ CPU core temperature monitor
    snd_seq: Advanced Linux Sound Architecture sequencer.
    snd_seq_device: ALSA sequencer device management
    snd_timer: ALSA timer interface
    snd: Advanced Linux Sound Architecture driver for soundcards.
    soundcore: Core sound module
    shpchp: Standard Hot Plug PCI Controller Driver
    i2c_piix4: PIIX4 SMBus driver
    tpm_infineon: Driver for Infineon TPM SLD 9630 TT 1.1 / SLB 9635 TT 1.2
    8250_fintek: Fintek F812164 module
    mac_hid
    wmi: ACPI-WMI Mapping Driver
    ipt_REJECT: Xtables: packet "rejection" target for IPv4
    nf_reject_ipv4
    nf_log_ipv4: Netfilter IPv4 packet logging
    nf_log_common
    xt_LOG: Xtables: IPv4/IPv6 packet logging
    xt_limit: Xtables: rate-limit match
    xt_tcpudp: Xtables: TCP, UDP and UDP-Lite match
    xt_addrtype: Xtables: address type match
    nf_conntrack_ipv4
    nf_defrag_ipv4
    xt_conntrack: Xtables: connection tracking state match
    ip6_tables: IPv6 packet filter
    nf_conntrack_netbios_ns: NetBIOS name service broadcast connection tracking helper
    nf_conntrack_broadcast
    nf_nat_ftp: ftp NAT helper
    nf_nat
    nf_conntrack_ftp: ftp connection tracking helper
    nf_conntrack
    iptable_filter: iptables filter table
    ip_tables: IPv4 packet filter
    x_tables: {ip,ip6,arp,eb}_tables backend module
    parport_pc: PC-style parallel port driver
    ppdev
    lp
    parport
    autofs4
    btrfs
    xor
    raid6_pq: RAID6 Q-syndrome calculations
    dm_mirror: device-mapper mirror target
    dm_region_hash: device-mapper region hash
    dm_log: device-mapper dirty region log
    ses: SCSI Enclosure Services (ses) driver
    enclosure: Enclosure Services
    uas
    usb_storage: USB Mass Storage driver for Linux
    hid_generic: HID generic driver
    usbhid: USB HID core driver
    hid
    nvidia_drm
    nvidia_modeset
    nvidia
    drm_kms_helper: DRM KMS helper
    psmouse: PS/2 mouse driver
    syscopyarea: Generic copyarea (sys-to-sys)
    r8169: RealTek RTL-8169 Gigabit Ethernet driver
    sysfillrect: Generic fill rectangle (sys-to-sys)
    mii: MII hardware support library
    sysimgblt: 1-bit/8-bit to 1-32 bit color expansion (sys-to-sys)
    fb_sys_fops: Generic file read (fb in system RAM)
    drm: DRM shared core routines
    ahci: AHCI SATA low-level driver
    libahci: Common AHCI SATA low-level routines
    fjes: FUJITSU Extended Socket Network Device Drive
    I ran through various ubuntu/proprietary drivers and various composite/non-composite modes, but could not stop these freezes/crashes. Incidentally, these crashes would freeze the window-manager, sometimes the mouse-cursor as well, sometimes the mouse would still move, sometimes not.

    The system would still be running though, and could be ssh-ed into. On connecting via ssh, I learned that these lockups also caused the Xorg process to go to a constant 100% and, also, become "un-killable",which is super annoying. This problem also make the Lightdm service unstoppable, so there was no way to get back to the window manager without rebooting, which required a full-on "REISUB"reboot, also, suuuuper-annoying

    So the fix for me was adding the "nogpumanager" boot parameter to my /etc/default/grub file, as per the below (always remembering,of course, to run...

    Code:
    sudo update-grub


    ...to get the system to incorporate the change into the system's next grub boot), like so:


    Link to a sample of my /etc/default/grub file

    Now this would lead me to believe that something is broken in the way the gpu-manager is interacting with Nvidia cards, at least mine and, at least on 16.04.2. As mentioned before, I don't really have the "chops" to say I know that for sure. Perhaps something for a more skilled user to check out though? At any rate, I hope this will be of use to someone in future.

    Cheers,

    P

    ************************************************** ****
    HUZZAH UBUNTU, HUZZAH!

  2. #2
    Join Date
    May 2005
    Beans
    153

    Re: Nvidia GPU Lockup - Window Manager Freezes Randomly, Xorg, 100% CPU, Proprietary

    Actually, it may be not. After doing all the above and having a "statistically-longer-period-of-non-failure-that-made-me-feel-comfortable-about-claiming-the-issue-as-solved"; I just had two "hard" failures, back-to-back. So...back to the drawing board. Though I do feel that I am on to something. Standby...
    ************************************************** ****
    HUZZAH UBUNTU, HUZZAH!

  3. #3
    Join Date
    May 2005
    Beans
    153

    Re: Nvidia GPU Lockup - Window Manager Freezes Randomly, Xorg, 100% CPU, Proprietary

    Just a quick update. After quite a bit of effort. I had reached a point where any combination of kernels and drivers would only get me to the Lightdm login, the keyboard would become unresponsive and the login screen would freeze.

    I then, just for kicks, replaced lightdm with gdm3, and now the system boots, logs in and, for the moment, seems to be working fine...so, if you've gone down this particular rabbit hole, hopefully that is helpful.

    Best,
    ************************************************** ****
    HUZZAH UBUNTU, HUZZAH!

  4. #4
    Join Date
    Nov 2017
    Beans
    1

    Re: Nvidia GPU Lockup - Window Manager Freezes Randomly, Xorg, 100% CPU, Proprietary

    I feel you bro.. N i did managed to recreate the problems many times. After quite *lots* of tries n retries i foung out its due the graphic drivers... lightdm is very shy on using drivers thats not rock solid in its dri connections... n after months of retries finally i got it just right for the hardware im using... and i'm now a happy man

Bookmarks

Posting Permissions

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