Registered Member
|
Hi,
After a while, plasmashell will have 25% CPU load, which should correspond to one core (2 cores and HT). It would stay like that for a long time, but every now and then it may use less CPU. Plasma becomes really unresponsive during this time. I normally fix it by killing plasmashell and restarting it. I can't figure out which part of plasma is responsible for this, I tried desktop and folder view, tried removing different widgets, but it's still the same. Also, I can't figure out any particular action to trigger this. This is probably a bug and should be reported, but because I have so little information, I want to ask here first. I'm running gentoo, with Qt 5.3.0 and compiled kde-frameworks and kde-workspaces from git a couple of days ago. I have had this issue from the first day I'm Plasma Next, so bisecting wouldn't help. |
Registered Member
|
Hello,
How you are starting Plasma Next session? Can you provide ~/.xsession-errors or console output depending upon how you are starting plasma next session? Thanks |
Registered Member
|
I'm using the file /usr/bin/startkde to start my session. The file .xsession errors does not exist, but I can manually launch plasmashell from the command line. I did that before and I couldn't see anything suspicious. This is the output until just after the launch:
I will post more output when the problem occurs again. edit: this is the full output. The problem occured again, I was watching a movie when it did. I only censored some passwords and ESSIDs
|
Registered Member
|
This time it happened really quickly after launching plasmashell. The output is really short, it should be easier to find anything relevant:
|
Administrator
|
If you have gdb and debug symbols you may want to attach to the plasmashell process and check if you can get a backtrace when this issue occurs.
"Violence is the last refuge of the incompetent."
Plasma FAQ maintainer - Plasma programming with Python |
Registered Member
|
Please check, whether it is caused by "active" systray items (e.g. spinning plasma-nm connection indicator or the file operation in progress indicator when copying files).
There's a bug for that: https://bugs.kde.org/show_bug.cgi?id=336274 |
Registered Member
|
While the spinner does cause high CPU load, plasmashell remains usable, unlike the problem that I am describing here. So they may be the same problem, but I can't say that for sure. Since my computer tries to connect to Ethernet when there is no cable plugged in after being suspended to RAM a couple of times, it's very well possible that the high CPU load seemed to occur randomly.
edit: nope, I'm also getting high CPU load when there's nothing going on in the notification area. So it seems to be a different bug. Also, while this is happening, I can remove pretty much any applet (it takes time and patience, because the shell is rather unresponsive) without anything changing. |
Registered Member
|
I kind of want to fix this problem. Can you tell me how exactly I have to run gdb to get relevant output? simply typing "gdb plasmashell" in the terminal just prompts me for more input and I don't know what to do with it.
edit: so when the problem occurs I start gdb and type "attach <pid>" and open a bug report? |
Administrator
|
When plasmashell is on full CPU rampage, note its PID, then run gdb, then issue "attach <PID>". Afterwards type "bt" to get a backtrace: copy and paste that to a bug report along with additional information (how to reproduce, etc).
"Violence is the last refuge of the incompetent."
Plasma FAQ maintainer - Plasma programming with Python |
Administrator
|
Did the 100% CPU usage occur not at startup, but afterwards? A very nasty bug was just fixed this morning and may be the cause of this.
"Violence is the last refuge of the incompetent."
Plasma FAQ maintainer - Plasma programming with Python |
Registered Member
|
I'm running 4.98.0 now. I have already had the bug with this version (afair) but now I haven't had it for more than day. It happened before that it didn't occur for hours or even days and then three times within an hour, so it may be something that I'm doing. I don't really want to upgrade to live right now, so I guess I will just post the backtrace here anyway if the problem occurs again.
|
Registered Member
|
It's been a couple of days, I haven't restarted my session and I haven't had the problem again. It seems fixed.
edit: false alarm. After restarting my session the problem returned. I will wait for the release tomorrow, I don't have time to test code from git right now. |
Registered Member
|
Nope, I still have the problem with 5.0.0. Only now it's periodically 50% (probably two cores) instead of constant 25% (1 core).
|
Administrator
|
If so please try to attach to plasmashell using gdb and let's see where it's stuck (by looking at the backtrace).
"Violence is the last refuge of the incompetent."
Plasma FAQ maintainer - Plasma programming with Python |
Registered Member
|
|
Registered users: Bing [Bot], Evergrowing, Google [Bot], rblackwell