I've been trying to install a Plasma 5 desktop for the last week.
I have tried Debian Testing, Fedora 22, OpenSuse, Manjaro, Kubuntu 15.04 and all with the same result.
I have been installing these onto a spare SSD that has no other OS/software on it. I have Linux Mint 17.2 Cinnamon installed as my primary OS on my main SSD and Windows 7 relegated to a separate HDD. I can install the same distros with Mate/Gnome/XFCE/Cinnamon without any problems. I can also install distros using KDE4 without any problems.

Back to the problem - after install the KDE Plasma 5 loading screen appears - then one of two things happen:
1. - The loading bar stops at just after halfway and everything freezes - no mouse cursor, no responses to keyboard input and I have to use the reset button on my PC to get out of it.
2. - The desktop actually loads and is usable for anywhere between approximately one minute and five minutes. Then the whole thing freezes - in this case my mouse cursor continues to respond to movement but will not interact with the desktop. Again there are no responses to keyboard input. Yet again my only option is the reset button.
I have made no changes to anything - I haven't had time! I have tried using a different keyboard and got the same results.

Is there any answer to this? Plasma 5 looks fantastic and I would really like to try it as I have always preferred KDE.
Here are my basic hardware details:

Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz
RAM 32115 MiB
GPU nVidia GeForce GTX 780
MB Asus P9X79 Deluxe
SSD Samsung 850 EVO 500GB - primary with Linux Mint installed
SSD Samsung 830 Pro 500GB - spare
HDD Hitachi 7210 1TB - Win 7 installed
USB Connected Hardware:
Sound Creative SB X-Fi 5.1 Pro
Mouse Razer Taipan
Keyboard Corsair K65 RGB
Wacom Intuos Pro Graphics Tablet
Logitech Extreme 3D Joystick

Hard freeze (even mouse movement) is most likely a video driver issue (a driver that claims to support opengl, but crashes/freezes when software tries to actually use it).

rdieter wrote:Hard freeze (even mouse movement) is most likely a video driver issue (a driver that claims to support opengl, but crashes/freezes when software tries to actually use it).

Yes, I was assuming it was probably something to do with the graphics driver. The installs I've tried have used the default graphics driver that comes with the distro. Which I assume is the nouveau driver. The freezing certainly doesn't give me time to install the nVidia driver or tinker with any settings.
I suppose I could add the 'nomodeset' entry in the boot menu and see if that makes a difference.

Hi,

on my system I have the same problem. I'm using a Gentoo system with live ebuilds of KDE 5. In difference to the thread creator I have a AMD R9 270X graphic card and use the open source ati driver with mesa 11.0.3. The behaviour is exactly the same as described... only the mouse freezes all time.

In order to solve the problem I rebuild all kde frameworks ebauilds, but this makes no difference. Any suggestions?

The OP didn't say the mouse freezes, but that there's no visible cursor (what could simply mean it's hidden by eg. the splash screen and never re-shown)

If you've a frozen cursor (you can see it, but it doesn't react to input (and you ensured the wire didn't just drop out of the USB slot that means the X11 server processes no input (and apparently nothing else) or the scanout buffer is frozen (problem in the graphics driver)

=> do you get past the splash screen before things freeze, or soes it freeze in the splash screen?

As general advise on the free radeon driver, check /var/log/Xorg.0.log on whether it uses "GLAMOR" acceleration, if yes, try EXA:
/etc/X11/xorg.conf.d/20-radeon.conf

Code: Select all

Section "Device"
   Identifier  "AMD R9 270X"
   Driver   "radeon"
   Option   "AccelMethod"           "EXA"
EndSection


It seems that the "radeon" driver do not support EXA... I got an error while starting the X server.

But... I rebuild all KDE5 relevant packages and now it worked for a while. I still get those freezes of mouse and key inputs but it's not happening in the first 2 or 3 minutes. Sometimes I have to wait for an hour or longer and then suddenly all freezes.... I have no idea

An error? What error? If radeon mandates GLAMOR, you're in quite some trouble.

I assume there'll be a texture leak. Disable compositing or use the xrandr backend ("kcmshell5 kwincompositing") for a start and see whether that improves anything.

I have the exact same problem, and it also reproduces against at least 2 Linux distributions:
openSUSE Leap 42.1 shows the behavior you described in point 2 (freezes 1-5 min into the working plasma session)
Fedora 23 shows the behavior you described in point 1 (freezes at end of splash screen and need reset button)

It is also uncanny to what degree our hardware specs match (leading me to believe it's related to the GTX 780 or the SSD):
Intel(R) Core(TM) i7-4770S CPU @ 3.10GHz
RAM 32 GiB
GPU nVidia GeForce GTX 780
MB Asus
SSD Samsung 850 EVO 500GB (on which OS is installed, and an 840 pro which is idling)

Hoes does one go about debugging this issue? As you mentioned you need to perform a hard reset to get out of it. I noticed it is possible (test under openSUSE) to restart the system using SysRq magic.

Just briefly getting back to this. I did not manage to debug the issue but after a clean install of Fedora + Plasma5 left me hanging with a non functional KDE suite I decided to give openSUSE Leap 42.1 another go and did a clean install based on the Net installer, this worked. This leads me to suspect that the problem is probably related to a specific Qt library. This hypothesis is supported that in the end (after installing all updates) even under Fedora Gnome my system would utterly freeze as soon as I started any complex KDE application (KDE System Settings or Dolphin). So not a solution, but a semi-dissatisfying workaround.

Here's something I tried.
I kept an eye on the clock then wrote down the time it froze. I'd then logout/login then wait for the freeze after doing this a couple of times I'd then reboot. Since I usually don't need to wait long after the next freeze this'd be the 3RD freeze I rebooted again. This time though I logged into an XFCE desktop. With no threat of freezes there I brought up a terminal & typed in

Code: Select all

journactl

I then noticed that the freezes & there times corresponded to an hp-systray error.
Perhaps someone else can try this out to either confirm or deny this as cause, or get a clue as to what this is.
Also if this helps I'm using Opensuse Tumbleweed with an Nvidia chipset & Nvidia GT610 GPU.

Some additional journalctl info:
In KDE

Nov 11 00:20:52 geekoshome systemd[1950]: Reached target Paths.
Nov 11 00:20:52 geekoshome systemd[1950]: Reached target Timers.
Nov 11 00:20:52 geekoshome systemd[1950]: Reached target Sockets.
Nov 11 00:20:52 geekoshome systemd[1950]: Reached target Basic System.
Nov 11 00:20:52 geekoshome systemd[1950]: Reached target Default.
Nov 11 00:20:52 geekoshome systemd[1950]: Startup finished in 76ms.
Nov 11 00:20:52 geekoshome sddm-helper[1954]: Adding cookie to "/home/robbie/.Xauthority"
Nov 11 00:20:53 geekoshome org.a11y.Bus[2006]: Activating service name='org.a11y.atspi.Registry'
Nov 11 00:20:53 geekoshome org.a11y.Bus[2006]: Successfully activated service 'org.a11y.atspi.Registry'
Nov 11 00:20:53 geekoshome org.a11y.atspi.Registry[2031]: SpiRegistry daemon is running with well-known name - org.a11y.atspi.Registry
Nov 11 00:21:12 geekoshome hp-systray[2245]: hp-systray[2245]: error: option -s not recognized
Nov 11 00:21:16 geekoshome hp-upgrade[2272]: hp-upgrade[2272]: error: HPLIP upgrade is disabled by openSUSE for security reasons, see https://bu
Nov 11 00:44:09 geekoshome systemd-coredump[2468]: Process 2467 (kinfocenter) of user 1000 dumped core.
Nov 11 00:44:59 geekoshome org.gtk.vfs.Daemon[2006]: A connection to the bus can't be made
Nov 11 00:45:00 geekoshome systemd[1950]: Stopped target Default.
Nov 11 00:45:00 geekoshome systemd[1950]: Stopped target Basic System.
Nov 11 00:45:00 geekoshome systemd[1950]: Reached target Shutdown.
Nov 11 00:45:00 geekoshome systemd[1950]: Starting Exit the Session...
Nov 11 00:45:00 geekoshome systemd[1950]: Stopped target Sockets.
Nov 11 00:45:00 geekoshome systemd[1950]: Stopped target Paths.
Nov 11 00:45:00 geekoshome systemd[1950]: Stopped target Timers.
Nov 11 00:45:00 geekoshome systemd[1951]: pam_unix(systemd-user:session): session closed for user robbie
Nov11 00:45:00 geekoshome systemd[1950]: Received SIGRTMIN+24 from PID 2520 (kill).
Nov 11 00:45:00 geekoshome org.kde.kglobalaccel[2006]: The X11 connection broke: I/O error (code 1)
Nov 11 00:45:00 geekoshome org.kde.kglobalaccel[2006]: XIO: fatal IO error 6 (No such device or address) on X server ":0"
Nov 11 00:45:00 geekoshome org.kde.kglobalaccel[2006]: after 381 requests (379 known processed) with 0 events remaining.

While writing the above in KDE the desktop did that odd freeze allowing me to type this message in Fx. At 00:44:09 I tried to get the kinfocenter(to see if it froze). It did, so I rebooted into XFCE.
Then one of the errors happened there to but AFAIK the XFCE desktop doesn't have a systray so it didn't freeze. This is what journalctl showed:

Nov 11 00:47:42 geekoshome pulseaudio[2132]: [pulseaudio] pid.c: Daemon already running.
Nov 11 00:47:45 geekoshome org.gtk.vfs.AfcVolumeMonitor[2001]: Volume monitor alive
Nov 11 00:47:46 geekoshome org.gnome.OnlineAccounts[2001]: goa-daemon-Message: goa-daemon version 3.18.1 starting
Nov 11 00:47:46 geekoshomehp-upgrade[2203]: hp-upgrade[2203]: error: HPLIP upgrade is disabled by openSUSE for security reasons, see https://bu
Nov 11 00:47:46 geekoshome org.gnome.OnlineAccounts[2001]: (goa-daemon:2202): goa-daemon-CRITICAL **: Error preparing AM: The name org.freedeskt
Nov 11 01:02:50 geekoshome su[2437]: The gnome keyring socket is not owned with the same credentials as the user login: /run/user/1000/keyring/c

TO THE ADMINS:
I'm very sorry that I didn't use the code bars for the KDE part however I found that colors didn't carry into the code parts.

First of all, I'm not sure that this freeze actually manifests itself in systemd's log / that you can find it by going through journalctl.

Second, I find it very unlikely that hp-systray or anything hp-* would be causing these severe crashes/freezes, especially since the OP tried so many distributions and they all give the same result. I thought a clean install of openSUSE Leap 42.1 had saved me but after I left the computer idle for sometime I got again a freeze (only a hard reset or SysRq+REISUB gets me out).

The last thing I'm trying before putting openSUSE Leap in the fridge again, is it to install the Nvidia drivers (which I am actually trying to avoid because of an oscillating fan bug and leave it idle for half a day with some applications running [1]. I honestly wonder why such a major bug in Plasma 5 was not caught by all these distributions (especially openSUSE and Fedora). Perhaps all the experts have ditched NVidia cards?

I would love to help with the debug process but I would need some documentation on how to tackle a bug that completely freezes my system (perhaps doing a controlled crash using SysRq at the time of the freeze would help? Not exactly sure how to set that up though.

[1] Update: Indeed after installing the proprietary NVidia drivers the system has been running Plasma 5 smoothly for the last 16 hours. Now how to formulate these experiences into a bug report ...

@hardyheron
Thanks for your comment! It caused me to take another look. My eye got distracted by the red color.
Also a question if I may which nvida drivers are you using? 352.41? 352.55? or the beta 358.09?

In the meantime yes I admit the hp could be a red herring, however there is this which is common to all of the freeze times what is this?

Code: Select all

Nov 11 18:43:11 geekoshome org.a11y.atspi.Registry[1985]: XIO:  fatal IO error 11 (Resource temporarily unavailable) on X server ":0"
Nov 11 18:43:11 geekoshome org.a11y.atspi.Registry[1985]: after 653 requests (653 known processed) with 0 events remaining.
Nov 11 18:43:11 geekoshome org.kde.kglobalaccel[1964]: The X11 connection broke (error 1). Did the X11 server die?
Nov 11 18:43:11 geekoshome org.gtk.vfs.Daemon[1964]: A connection to the bus can't be made


As I said the red hp stuff grabbed my attention and i totally missed these lines. The error takes place at the times of the freeze even now.

If I look at my journalctl logs from the last time I remembered it freezing (before installing the NVIDIA drivers [1]) I see this:

Code: Select all

 journalctl -b -2
Nov 11 06:02:44  kernel: nouveau E[ PIBUS][0000:01:00.0] HUB0: 0x000000 0x00000000 (0x0e000200)
Nov 11 06:02:44  kernel: nouveau E[ PFIFO][0000:01:00.0] read fault at 0x0000450000 [PTE] from GR/SCC on channel 0x007ed8c000 [y2base[12233]]
Nov 11 06:02:44  kernel: nouveau E[ PFIFO][0000:01:00.0] PGRAPH engine fault on channel 14, recovering...

Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: SIGNATURE
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: ch 15 [kscreenlocker_g[14299]] subc 0 mthd 0x0000 data 0x00000000
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: SIGNATURE
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: ch 15 [kscreenlocker_g[14299]] subc 0 mthd 0x0000 data 0x00000000
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: GPPTR PBENTRY SIGNATURE
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: ch 15 [kscreenlocker_g[14299]] subc 0 mthd 0x0000 data 0x00000000
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: GPPTR SIGNATURE
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: ch 15 [kscreenlocker_g[14299]] subc 0 mthd 0x0000 data 0x00000000
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: SIGNATURE
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: ch 15 [kscreenlocker_g[14299]] subc 0 mthd 0x0000 data 0x00000000
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: SIGNATURE
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] PBDMA0: ch 15 [kscreenlocker_g[14299]] subc 0 mthd 0x0000 data 0x00000000
Nov 11 06:09:35  kernel: nouveau E[ PFIFO][0000:01:00.0] read fault at 0x0000000000 [PTE] from PBDMA0/HOST on channel 0x007ec7f000 [unknown]


Seemingly related bug reports:
https://bugzilla.redhat.com/show_bug.cgi?id=1278071
https://bugzilla.redhat.com/show_bug.cgi?id=1214663 <- this one describes fits very nicely
https://bugs.freedesktop.org/show_bug.cgi?id=76414

[1] x11-video-nvidiaG04 352.55-17.1 from openSUSE Leap 42.1 repo

Hmm?
Drivers hmm...and you're using the Leap repo version. I'm using TW and an Nvidia beta...hmm?
The crazy part here is I removed the systray and installed widgets in the panel to replace what I had in the systray, now my PC's fine.
What weird and crazy machines computers are.
That said if you're not having any more freezes and I'm not, all I can say is thank you. Thank you Hardyheroin for showing me I wasn’t the only one and for teaching me some new places to look if this happens again.