plasmashell crashes

I'm having a bit of a problem on my laptop. I run FreeBSD, just upgraded from 14.3 to 15 yesterday. So both systems run kde/plasma 6.5.x on X11 (just upgraded from 6.5.3 to 6.5.4 last night).
or
On my desktop, (Ryzen 7 5800x, 64GB RAM, nVidia RTX3060Ti), everything runs perfectly. The problem occurs on my laptop, which is a Lenovo Thinkpad W520, with the i7-2760QM, 32GB of RAM, and although it has an optimus setup, the intel GPU is enabled and the nVidia discrete GPU is disabled in bios. I also made sure the nvidia drivers were removed.

So what I am seeing is that while I am using the laptop, it is fine. However, when I stop using it, either closing the lid, or running an errand, when the screen blanks and comes back, I am able to move the mouse, but the the mouse buttons and keyboard are inoperative within the X session, and clicking a terminal window, mouse tab, etc., does not shift focus...The system, when it is working as expected, is using about 5MB (16%) of RAM. However, after running with the lid closed over night, idling, and the memory usage goes above 65%.

The problem can be fixed by jumping to one of the VTs and restart the login manager. I have tried switching that between SDDM and SLIM, and it occurs in both. As far as I know, plasmashell continues to run, but most times I get two core dump files in my home directory:

Code:
-rw-------  1 1001 1001 20951040 Dec 14 16:56 drkonqi.core
-rw-------  1 1001 1001 22335488 Dec 14 17:18 plasmashell.core[CODE]

gdb on the plasmashell core shows

[CODE]Core was generated by `/usr/local/bin/plasmashell'.
Program terminated with signal SIGABRT, Aborted.
Sent by thr_kill() from pid 38293 and user 1001.
#0  0x000000085f3c79ea in ?? ()
[Current thread is 1 (LWP 102176)]

and the one on the drkonqi core:

Code:
Core was generated by `/usr/local/lib/libexec/drkonqi --qtversion 6.9.3 --kdeframeworksversion 6.20.0 -'.
Program terminated with signal SIGABRT, Aborted.
Sent by thr_kill() from pid 26747 and user 1001.
#0  0x0000000845e539ea in ?? ()
[Current thread is 1 (LWP 106754)]

Unfortunately, I am a sysadmin, and not a programmer, so I have no experience with debugging core dumps.

Has anyone got any suggestions on a possible cause? Note that I did not see this problem when running 14.2, however when I upgraded to 14.3 and then 15.0, it has been happening with regular frequency.

Thanks
--vr
 
Interesting, I am having the same problem on both of my Thinkpad laptops, An E540, and a T440. Both since upgrading to 14.3 and with fresh installs of 15.0. Sorry I can't be of much help, but am following this to see if someone can shed some light. I can also restart plasma session from a VT. My laptops both have Intel graphics.

X
 
Interesting, I am having the same problem on both of my Thinkpad laptops, An E540, and a T440. Both since upgrading to 14.3 and with fresh installs of 15.0. Sorry I can't be of much help, but am following this to see if someone can shed some light. I can also restart plasma session from a VT. My laptops both have Intel graphics.

X
Hey Xeno,

You know, I have monitoring set up on my homelab, and one thing I noticed is that, for example, if I am using it of an evening, memory is sitting somewhere between 15 and 25% utilization. I close the lid when I am done using it. The next morning, memory is ~70% used even though the system was idling.

Well last night, I killed firefox before closing the lid, and today, memory is at 17.1%. I have not been able to get to the laptop to check it, but I suspect it will be working. So maybe an interaction between firefox and plasmashell?

BTW, how are you restarting plasmashell from the VT? I tried and it said "could not connect to session"
 
I have not been able to get to the laptop to check it, but I suspect it will be working. So maybe an interaction between firefox and plasmashell?
That is a big nope. Opened the lid, and had mouse ball control, but no clicky and no keyboard. Had to restart sddm.

I'm considering a) reinstalling kde, b) creating a new user and seeing if the problem occurs with a different user, or c) going back to 14.2 to see if it still occurs there, then upgrading to 15.0 directly.
 
Hey Xeno,

You know, I have monitoring set up on my homelab, and one thing I noticed is that, for example, if I am using it of an evening, memory is sitting somewhere between 15 and 25% utilization. I close the lid when I am done using it. The next morning, memory is ~70% used even though the system was idling.

Well last night, I killed firefox before closing the lid, and today, memory is at 17.1%. I have not been able to get to the laptop to check it, but I suspect it will be working. So maybe an interaction between firefox and plasmashell?

BTW, how are you restarting plasmashell from the VT? I tried and it said "could not connect to session"
Well I don't use Firefox, I use chromium.... Anyway, I just go to a VT and run htop then kill plasma-session and it automatically take me back to the login screen (SDDM). I'm not too good at understanding crash dump reports, but next time it happens, I'll make a note of what's using high memory in htop.
 
Well I don't use Firefox, I use chromium.... Anyway, I just go to a VT and run htop then kill plasma-session and it automatically take me back to the login screen (SDDM). I'm not too good at understanding crash dump reports, but next time it happens, I'll make a note of what's using high memory in htop.
Yeah, same as I do. I just restart sddm from thr VT. But the fact that you use chromium and I use firefox is a data point, thank you for that.
 
Yeah, same as I do. I just restart sddm from thr VT. But the fact that you use chromium and I use firefox is a data point, thank you for that.
So I think it is kwin that is the culprit. When the screen becomes unresponsive, kwin seems locked up. I cannot kill kwin from a VT using htop. I'm still looking for more info on this sort of problem.

X
 
So I think it is kwin that is the culprit. When the screen becomes unresponsive, kwin seems locked up. I cannot kill kwin from a VT using htop. I'm still looking for more info on this sort of problem.

X
By the way, I'm using X11, not Wayland. Is that true for you as well?
 
So I think it is kwin that is the culprit. When the screen becomes unresponsive, kwin seems locked up. I cannot kill kwin from a VT using htop. I'm still looking for more info on this sort of problem.

X
Interesting. I'm wondering what else is going on. I just checked and kwin is owned by my user.

By the way, I'm using X11, not Wayland. Is that true for you as well?
It is. Wayland has too many broken bits in it for my liking.

Now having said that, I am wondering if KDE has something that is using some of the broken wayland bits internally since they have doubled down on pushing the wayland agenda.

I was actually kicking around the idea of the spinning off a BE and trying to replace xorg with xlibre.

Was also thinking about booting back into my last 14.2 BE, and testing there, then trying to upgrade directly from 14.2 -> 15.0 and see if that makes a difference.
 
Interesting. I'm wondering what else is going on. I just checked and kwin is owned by my user.


It is. Wayland has too many broken bits in it for my liking.

Now having said that, I am wondering if KDE has something that is using some of the broken wayland bits internally since they have doubled down on pushing the wayland agenda.

I was actually kicking around the idea of the spinning off a BE and trying to replace xorg with xlibre.

Was also thinking about booting back into my last 14.2 BE, and testing there, then trying to upgrade directly from 14.2 -> 15.0 and see if that makes a difference.
Yes, I think Plasma 6 was designed with Wayland as its primary target. I'm thinking the problem is the between kwin and some graphics driver bit....but my skills are not yet there to track it down,
 
Yes, I think Plasma 6 was designed with Wayland as its primary target. I'm thinking the problem is the between kwin and some graphics driver bit....but my skills are not yet there to track it down,
Which would make it make sense why it is only occurring on our laptops. My desktop doesn't have intel, I have an nvidia 3060ti, and don't see these problems... So I am also wondering if it may be something with the DRM drivers.
 
Back
Top