When i use my machine and suddenly things get frozen for few seconds and refreshes back in working condition with a pop-up named "KWIN Window Manager" with description "Desktop effects were restarted due to a graphics reset". This doesn't not happens frequently but i do get it regularly (annoying)and i don't know whats causing the problem. So how do i find whats causing the problem and troubleshoot this issue ?


Desktop: KDE Plasma 5.4.3
Distro: openSUSE Leap 42.1
I am using the proprietary graphic driver.
Graphics: Card: NVIDIA GT218 [GeForce 210] bus-ID: 01:00.0
Display Server: X.Org 1.17.2 drivers: nvidia (unloaded: fbdev,nv,vesa,nouveau)
Resolution: 1366x768@59.8hz
GLX Renderer: GeForce 210/PCIe/SSE2 GLX Version: 3.3.0 NVIDIA 340.93 Direct Rendering: Yes

My compositor settings:
Scale Method: Accurate
Rendering backend: OpenGL 2.0
OpenGL interface:GLX
Tearing prevention("vsync"): Automatic
Keep window thumbails: Only for shown Windows.

Your GPU took a nap.
Check dmesg (notably grep it for NVRM, but also see the tail right after the incident) on whether it contains valuable information.
Maybe there's also notion in /var/log/Xorg.0.log

luebking wrote:Your GPU took a nap.
Check dmesg (notably grep it for NVRM, but also see the tail right after the incident) on whether it contains valuable information.
Maybe there's also notion in /var/log/Xorg.0.log

Found this last 4 lines of dmesg after i got "desktop effects were restarted due to a graphics reset" :
[ 57.831047] kactivitymanage[1427]: segfault at 7f47c3960d10 ip 00007f47c3ba769a sp 00007ffd4123f638 error 4 in libQt5Sql.so.5.5.1[7f47c3b92000+3f000]
[ 387.468677] perf interrupt took too long (2514 > 2500), lowering kernel.perf_event_max_sample_rate to 50000

dmesg | grep -iC 3 "NVRM"

Code: Select all

[    1.756672] hub 2-0:1.0: 2 ports detected
[    1.761717] vgaarb: device changed decodes: PCI:0000:01:00.0,olddecodes=io+mem,decodes=none:owns=io+mem
[    1.762252] [drm] Initialized nvidia-drm 0.0.0 20150116 for 0000:01:00.0 on minor 0
[    1.762259] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  340.93  Wed Aug 19 16:49:15 PDT 2015
[    1.764327] nvidia_uvm: Loaded the UVM driver, major device number 250
[    1.780223] hub 3-0:1.0: USB hub found
[    1.780566] hub 3-0:1.0: 2 ports detected


journalctl | tail - http://dpaste.com/3TCRVZ5

[ 387.468677] perf interrupt took too long (2514 > 2500), lowering kernel.perf_event_max_sample_rate to 50000

This may go along with the problem but doesn't indicate the cause. The kactivitymanagerd crash suggests its connection to the X11 server broke.
Older dmesg entries (they're actually only two lines) may still hold the answer, the other log data is meaningless.

Could find only these two related to "KWin" in .xsession-errors-:0 in home folder.

Code: Select all

(uint32 1,)
kdeinit5: PID 1938 terminated.
kf5.kded: found kded module "touchpad" by prepending 'kded_' to the library path, please fix your metadata.
QMetaMethod::invoke: Unable to handle unregistered datatype 'KWin::EffectWindow*'
Trying to use rootObject before initialization is completed, whilst using setInitializationDelayed. Forcing completion
kf5.kded: found kded module "freespacenotifier" by prepending 'kded_' to the library path, please fix your metadata.

Code: Select all

OpenGL vendor string:                   NVIDIA Corporation
OpenGL renderer string:                 GeForce 210/PCIe/SSE2
OpenGL version string:                  3.3.0 NVIDIA 340.93
OpenGL shading language version string: 3.30 NVIDIA via Cg compiler
Driver:                                 NVIDIA
Driver version:                         340.93
GPU class:                              G80/G90
OpenGL version:                         3.3
GLSL version:                           3.30
X server version:                       1.17.2
Linux kernel version:                   4.1.13
Requires strict binding:                no
GLSL shaders:                           yes
Texture NPOT support:                   yes
Virtual Machine:                        no
Currrent active notifications: QHash()
Guessing partOf as: 0
 New Notification:  "KWin Window Manager" "Desktop effects were restarted due to a graphics reset" -1 & Part of: 0

The indication would (if at all) be in dmesg, not xsession errors.


Debug statements seem disabled*, otherwise kwin would at least tell you whether the reason is the KWin GL context, some other context or just unknown.

*Afaik that's not default so your distro will (thakfully, it's really spammy have changed that but I don't know in which way.

~/.config/kdelogging.ini

Code: Select all

[Rules]
*.debug=true
kwin_core=true


QT_LOGGING_CONF=~/.config/kdelogging.ini kwin_x11 --replace &

Hi,

I have the same problem on openSUSE 13.1 (3.11.10-29-desktop x86_64) and can't make any sense of it. The freeze occurs frequently, mainly when using eclipse-cpp IDE.
I'm happy to learn how to fix this, because it is quite annoying.

Cheers

Code: Select all

kded4 --version
Qt: 4.8.5
KDE Development Platform: 4.11.5
KDE Daemon: 4.11.5

My compositor settings:
Scale Method: Accurate
Rendering backend: OpenGL 2.0
OpenGL interface:GLX
Tearing prevention("vsync"): Automatic
Keep window thumbails: Only for shown Windows.

Code: Select all

glxinfo | grep render
direct rendering: Yes
OpenGL renderer string: GeForce 9500 GT/PCIe/SSE2
GL_NVX_conditional_render, GL_NVX_gpu_memory_info,
GL_NV_ES1_1_compatibility, GL_NV_blend_square, GL_NV_conditional_render,
GL_NV_path_rendering, GL_NV_pixel_data_range, GL_NV_point_sprite,
GL_NVX_conditional_render, GL_NVX_gpu_memory_info,
GL_NV_ES1_1_compatibility, GL_NV_blend_square, GL_NV_conditional_render,
GL_NV_path_rendering, GL_NV_pixel_data_range, GL_NV_point_sprite,


Code: Select all

dmesg | grep NVRM
[    2.500305] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  340.96  Sun Nov  8 22:33:28 PST 2015
[    4.952363] NVRM: Your system is not currently configured to drive a VGA console
[    4.952367] NVRM: on the primary VGA device. The NVIDIA Linux graphics driver
[    4.952370] NVRM: requires the use of a text-mode VGA console. Use of other console
[    4.952371] NVRM: drivers including, but not limited to, vesafb, may result in
[    4.952372] NVRM: corruption and stability problems, and is not supported.
[ 8756.480049] NVRM: GPU at PCI:0000:02:00: GPU-883fc05e-43b2-9731-b447-e04ff70d1b6f
[ 8756.480059] NVRM: Xid (PCI:0000:02:00): 3, C 00000001 SC 00000000 M 00000188 Data e0ff0200
[ 8758.484007] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
[ 8775.525806] NVRM: Xid (PCI:0000:02:00): 8, Channel 00000001
[ 8777.527631] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
[ 8781.531206] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
[ 8783.533026] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
[ 8861.759750] NVRM: Xid (PCI:0000:02:00): 8, Channel 00000001
[ 8863.761576] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
[ 8867.765143] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
[ 8869.766962] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
[ 8869.782110] NVRM: Xid (PCI:0000:02:00): 69, Class Error: ChId 0001, Class 00008297, Offset 00001b0c, Data 9000f010, ErrorCode 00000000
[ 8869.784222] NVRM: Xid (PCI:0000:02:00): 9, Channel 00000001 Instance 00000484 status 0000001d
[ 8869.786427] NVRM: Xid (PCI:0000:02:00): 13, Graphics Exception: ChID 0003, Class 00000000, Offset 00000000, Data 00000000

> [ 4.952363] NVRM: Your system is not currently configured to drive a VGA console

https://wiki.archlinux.org/index.php/GR ... ramebuffer

> [ 8777.527631] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
iieee.
- could simply be due to the above
- blacklist nouveau kernel module (same problem)
- did you already update to X11 ABI v20 (ie. what's the version of the Xorg server?)
- output of "dmesg | grep vmalloc"
- overheat? (check nvidia-settings for sensor data)
- your GPU may be dying

Thanks for the info, I appreciate your help. If I can narrow down the problem at least I know..

https://wiki.archlinux.org/index.php/GRUB/Tips_and_tricks#Disable_framebuffer

- Added "GRUB_TERMINAL_OUTPUT=console" to /etc/default/grub
- blacklisted nouveau
- "dmesg | grep NVRM" doesn't always show the CPU stuff, only the VGA console message
- "dmesg | grep vmalloc" doesn't show anything
- nvidia-settings thermal sensor output seems fine, at least right now

Could very well be a hardware problem. I have a BIOSTAR GF8200E motherboard and even after exhausting the internet I still never could get the onboard NVIDIA GeForce 8200 graphics card to work with the proprietary NVIDIA linux drivers (neither manual installation nor repository). Nouveau was giving me annoying freezes, so I bought a used GeForce 9500 GT PCIE card on ebay and installed it. Works most of the time except for the reported behavior. Of course, I would like to figure out what the actual problem is, so I can either repair it or let it rest. Maybe the GF9500GT GPU is bad or maybe it's even the old board itself? Occasionally it doesn't recognize the USB port block where my USB keyboard is plugged in, so I have to shutdown, fiddle with the USB plugs on the board, and restart. Not very reassuring..

Is there a way to tell for sure what's going wrong on my system?

The 9500 gt should be faster anyway, but maybe your system is underpowered?

Does the 9500 have a dedicatetd power plug (and is it used) and is the AC sufficiently sized (what's not necessarily the case if it was supposed to run the embedded system - the 9000 series wasn't as much a power **** as the 8000 series, but the in-generation level isn't comparable and your USB issues make me wonder)
You can try to downclock the GPU, force the desktop mode or similar, add

Code: Select all

Option  "CoolBits"    "1"


to the [Device] section (you'll likely either have a /etc/X11/xorg.conf or some /etc/X11/xorg.conf.d/*nvidia* snippet) to possibly gain more options in nvidia-settings.

I created xorg.conf via nvidia-xconfig and added the option to the Device section.
Now that I have changed some settings I'll wait till I have issues again. It might work..
The 9500 doesn't have a dedicated power plug as far as I can tell. So you reckon the main power supply might be too weak and a stronger one could be the cure?

That /can/ be the cause, what are its specs? (General Watt and max Ampere on the 12V lane?)

500W / 27A (+12V)
tacens radix eco 500 if you need to google more specs

No, that's sufficient for sure.

Had another couple of KDE restarts and/or freezes.
The last one just now after opening the window and letting cold air into the room - so some loose contact or could it be something else?
I'm thinking of trying a used geforce 8xxx PCIe card from ebay for a few bucks in case the board just can't handle the 9500gt. Can you think of anything else I could check or any other advice before I invest the money?