Bug 450980 - KDE 5.24 breaks suspend to RAM (really badly); reverting to 5.23 resolves it
Summary: KDE 5.24 breaks suspend to RAM (really badly); reverting to 5.23 resolves it
Status: RESOLVED NOT A BUG
Alias: None
Product: plasmashell
Classification: Plasma
Component: general (show other bugs)
Version: 5.24.2
Platform: Manjaro Linux
: NOR major
Target Milestone: 1.0
Assignee: Plasma Bugs List
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2022-02-28 20:29 UTC by flan_suse
Modified: 2022-03-08 16:55 UTC (History)
1 user (show)

See Also:
Latest Commit:
Version Fixed In:
Sentry Crash Report:


Attachments
Output of "systemd-inhibit --list" when KDE and SDDM are running. (1.19 KB, text/plain)
2022-03-01 15:26 UTC, flan_suse
Details
Output of "systemd-inhibit --list" when KDE and SDDM are *NOT* running. (470 bytes, text/plain)
2022-03-01 15:26 UTC, flan_suse
Details
journalctl when suspending with KDE 5.24 (29.24 KB, text/plain)
2022-03-02 00:01 UTC, flan_suse
Details

Note You need to log in before you can comment on or make changes to this bug.
Description flan_suse 2022-02-28 20:29:32 UTC
Starting with KDE 5.24, when I tell my laptop to suspend to RAM, the following occurs:
* It "appears" to suspend. (Fans stop, external monitor turns off)
* But the CPU is COOKING at maximum heat, without any running fans!
* Pressing keys on my wireless keyboard or mouse does NOT wake up it (as it always has before)
* I need to "wake up" the laptop by pressing on the actual laptop keyboard
* When it "wakes up", the entire laptop feels like it's COOKING. Apparently, the CPU was still running, but no fans were running to keep it cooled.

With 5.23 and earlier:
* The laptop correctly suspends to RAM
* Everything stays cool (CPU is not running)
* Pressing a key on my wireless keyboard or mouse wakes up the laptop
* No issues at all, laptop feels cool

I have an external monitor attached to the laptop, which works properly.

I had to revert back to KDE 5.23, since this bug was literally FRYING my laptop to ungodly levels of heat. (Had I not noticed earlier and reproduced it consistently, I could have risked damaging my components.)

Was there a change done to the "power" side of things from 5.23 -> 5.24?

To reiterate: It's specifically KDE 5.24. Trying a different kernel makes no difference.


SYSTEM INFO:
Operating System: Manjaro Linux
KDE Plasma Version: 5.24.2
KDE Frameworks Version: 5.91.0
Qt Version: 5.15.2
Kernel Version: 5.16.7-1-MANJARO (64-bit)
Graphics Platform: X11
Processors: 8 × 11th Gen Intel® Core™ i7-1165G7 @ 2.80GHz
Memory: 15.4 GiB of RAM
Graphics Processor: Mesa Intel® Xe Graphics

I have a desktop PC with the same software, however, it uses a discrete GPU (Nvidia), which does not have this issue with KDE 5.24.

What other information would be useful?

Thank you.
Comment 1 David Edmundson 2022-03-01 13:43:00 UTC
Please test with "systemctl suspend"

If that behaves the same, the issue isn't our side.

systemd-inhibit --list might also be useful
Comment 2 flan_suse 2022-03-01 15:25:15 UTC
(In reply to David Edmundson from comment #1)
> Please test with "systemctl suspend"
> 
> If that behaves the same, the issue isn't our side.
> 
> systemd-inhibit --list might also be useful

It gets more interesting.

* While I am logged into a KDE session, even using "systemctl suspend" causes the same issue.

AND...

* If I logout of KDE (leaving only the SDDM login screen), then switch to TTY2 and issue "systemctl suspend", it's still affected by the same issue

HOWEVER...

* If I log out of KDE and stop the SDDM display manager ("systemctl stop sddm"), then switch to TTY2 and issue "systemctl suspend", the issue no longer exists: CPU is not running hot, and using my wireless keyboard or mouse properly wakes up from suspend

So *without* KDE nor SSDM running, it behaves correctly again (regardless of which kernel I'm using). If KDE or SDDM is running, then even using "systemctl suspend" reproduces the same issue. Only when I stop SDDM and suspend, it will work correctly

I will attach the outputs from "systemd-inhibit --list".

The first text file is the output when logged into KDE.

The second text file is the output when I stop SDDM.

(I must also note that when I force the laptop to wake up when the bug is occurring, my wallpaper doesn't "stick" anymore. If I switch workspaces, the wallpaper will flicker and disappear.)
Comment 3 flan_suse 2022-03-01 15:26:08 UTC
Created attachment 147216 [details]
Output of "systemd-inhibit --list" when KDE and SDDM are running.

Output of "systemd-inhibit --list" when KDE and SDDM are running.
Comment 4 flan_suse 2022-03-01 15:26:29 UTC
Created attachment 147217 [details]
Output of "systemd-inhibit --list" when KDE and SDDM are *NOT* running.

Output of "systemd-inhibit --list" when KDE and SDDM are *NOT* running.
Comment 5 flan_suse 2022-03-01 19:01:17 UTC
Keep in mind when this bug occurs, my fans STOP. Monitor shut off (no signal). However, for some reason my CPU is heating up rapidly, since it's somehow still running, but without the cooling of the fans. I can attach a journalctl as well, but it appears to believe the suspend was successful.
Comment 6 flan_suse 2022-03-02 00:01:27 UTC
Created attachment 147231 [details]
journalctl when suspending with KDE 5.24

Notice there's a 3-minute gap in the log? That's how long I waited while the system was "suspended". I couldn't wake it up until I used the laptop's keyboard to "wake up" the system. (USB mouse and keyboard did not work.)

Yes, during that 3-minute period, the laptop got REALLY HOT, since the fans stopped, but apparently the CPU was still cooking.

This doesn't happen with KDE 5.23 (doesn't matter which kernel, nor which version of upower.)
Comment 7 flan_suse 2022-03-02 16:14:03 UTC
(In reply to flan_suse from comment #6)
> Notice there's a 3-minute gap in the log?

To emphasize this point, for all intents and purposes the system *did* in fact "suspend".

No activity logged for 3 minutes, until I forced it to "wake up" again.

During this 3-minute period, using the USB mouse and keyboard would *NOT* wake up the laptop, even though it's *supposed* to (as it does in KDE 5.23)

During this 3-minute period (of no logged activity, remember), the laptop's fans are completely off. Yet the CPU quickly gets hotter and hotter! (I can feel it almost hurt my finger when I touch the surface.)

None of the above issue happens if with KDE 5.24:
1) I log out
2) I switch to TTY2
3) I stop the SDDM service
4) I issue "systemctl suspend"

Then it will behave *properly*, as if I was using KDE 5.23.

This is why I think it's related to KDE (and notably, the transition from 5.23 -> 5.24).

I believe it has something to do with local user settings (configured during 5.23 and earlier), which cause a conflict with KDE 5.24's changes in handling external displays, settings, and power management.

Reverting back to KDE 5.23 (regardless of kernel), always, 100%, resolves the issue.
Comment 8 flan_suse 2022-03-03 17:56:05 UTC
This might be upower after all.

I'll keep on eye on things, and see what happens with upstream upower.

(However, I did try with an earlier version of upower, and the same issue persisted.)

https://bbs.archlinux.org/viewtopic.php?id=274292
Comment 9 flan_suse 2022-03-08 16:55:15 UTC
Upstream upower and kernel, after all. (Fixed in kernel 5.17)