Bug 463379 - Double Monitor: Lock-screen, after some indefinite amount of time, will fail to actually display when the machine is taken out of sleep due to kwin_x11 abort
Summary: Double Monitor: Lock-screen, after some indefinite amount of time, will fail ...
Status: RESOLVED FIXED
Alias: None
Product: kwin
Classification: Plasma
Component: general (show other bugs)
Version: 5.26.5
Platform: Arch Linux Linux
: VHI critical
Target Milestone: ---
Assignee: KWin default assignee
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2022-12-23 06:00 UTC by Maluchi Ugokwe Jr.
Modified: 2023-03-13 02:11 UTC (History)
5 users (show)

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


Attachments
Picture shows the problem. (233.14 KB, image/png)
2022-12-23 06:04 UTC, Maluchi Ugokwe Jr.
Details
Stack trace of crashing kwin_x11 (3.10 MB, image/jpeg)
2023-03-03 16:05 UTC, Louis Moureaux
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Maluchi Ugokwe Jr. 2022-12-23 06:00:22 UTC
SUMMARY
***
Lock-screen, after some indefinite amount of time, will fail to actually display when the machine is taken out of sleep.

And, yes, this is a grave matter. This is a UX nightmare, with all due respect because I love KDE. 

Ostensibly, the lock-screen works as normal at the very start, with laptop lock buttons working expectedly, ie: in accordance with the function of locking.

Now, after keeping the computer on lock while being indiscriminate with the amount of time locked, trying to open the machine (laptop in my case) again will have you greeted with only a cursor and both screens activated. However, both screens will be black and your cursor will be trapped on one screen.

This happens on a laptop and external monitor setup. Disconnecting the external monitor might actually show me the lock-screen and allow me to login in. I should not have to do this.

This kind of issue happens after some unknown time elapsed. It can be a matter of screen-locking from somewhere 30 seconds all up to 6 hours. For instance I tried 5 minutes and I could replicate this issue.
***


STEPS TO REPRODUCE
1.  Have a laptop with an external monitor. (I do not know about other setups; this is mine)
2.  Lock the screen (sleep mode). Wait 5 minutes.
3.  Take the laptop out of sleep mode. You should see the issue.

OBSERVED RESULT
Black screen with cursor shown. No lock-screen elements or way to actually put in the password needed to get back to work/play/whatever.

EXPECTED RESULT
Functioning lock-screen.

SOFTWARE/OS VERSIONS
Windows:  N/A
macOS:  N/A
Linux/KDE Plasma: Arch Linux, Linux-zen 6.0.12, Plasma X11 
KDE Plasma Version: 5.26.4
KDE Frameworks Version: 5.101.0
Qt Version: 5.15.7

ADDITIONAL INFORMATION
Intel Core i7, NVIDIA GeForce RTX 2060, MSI GE75 Raider 10SE
Comment 1 Maluchi Ugokwe Jr. 2022-12-23 06:04:03 UTC
Created attachment 154772 [details]
Picture shows the problem.
Comment 2 Maluchi Ugokwe Jr. 2022-12-23 17:57:03 UTC
To note: this is seemingly at random. Perhaps not even based off of time but some kind of stability. I don't know.

I have tried to unlock my computer from sleep (after hours of my own sleep)... and it worked perfectly.

Will return with more comments if need be.
Comment 3 Nate Graham 2022-12-24 22:19:04 UTC
Possibly a GPU issue; see similar bug 457284.

Marking as VHI critical because this can cause data loss due to the inability to unlock the screen unless you're an expert who knows how to unlock the screen using a virtual terminal.
Comment 4 David Edmundson 2022-12-24 23:26:41 UTC
If you disable lock after suspend do you still have issues waking up from sleep?

Another test:

Do you reliably get issues if you lock the screen (without suspend)
control+alt+f4 (to switch to another VT)
control+alt+f1 (to switch back)

That should trigger an nvidia context loss which is what you'll get sometimes with sleep.
Comment 5 Maluchi Ugokwe Jr. 2022-12-25 19:37:19 UTC
(In reply to Nate Graham from comment #3)
> Possibly a GPU issue; see similar bug 457284.
> 
> Marking as VHI critical because this can cause data loss due to the
> inability to unlock the screen unless you're an expert who knows how to
> unlock the screen using a virtual terminal.

Well, to provide info. I don't know what else to say but that the bug just stopped happening. It just simply demanifested.

If anything happens, I'll tell you guys.
Comment 6 Maluchi Ugokwe Jr. 2022-12-25 20:58:03 UTC
(In reply to Nate Graham from comment #3)
> Possibly a GPU issue; see similar bug 457284.
> 
> Marking as VHI critical because this can cause data loss due to the
> inability to unlock the screen unless you're an expert who knows how to
> unlock the screen using a virtual terminal.

Nate. The issue reoccurred. To elaborate:
The screen will flicker black and it will display pure black on both monitors.
Comment 7 Maluchi Ugokwe Jr. 2022-12-25 21:03:28 UTC
(In reply to David Edmundson from comment #4)
> If you disable lock after suspend do you still have issues waking up from
> sleep?
> 
> Another test:
> 
> Do you reliably get issues if you lock the screen (without suspend)
> control+alt+f4 (to switch to another VT)
> control+alt+f1 (to switch back)
> 
> That should trigger an nvidia context loss which is what you'll get
> sometimes with sleep.

I did the lock without suspend test you prompted me to here. I locked the screen and switched ttys. It flickers a bit but it will invariably return to the correct lock-screen prompt and I am able to unlock from there.
Comment 8 Nate Graham 2022-12-25 21:35:57 UTC
Thanks for the info.
Comment 9 Maluchi Ugokwe Jr. 2022-12-27 23:19:05 UTC
TO ADD ON:

Waking up of sleep+suspension has the bug detailed in my OP, but I can get a lock-screen by unplugging my HDMI and using my built in to log in...

Unfortunately this is not really a solution, and it makes my external monitor unusable, given that when plugging in the external monitor again, all screens will shut off and be pitch black.

This is beginning to make using KDE unmanageable for my work. If I'm in the middle of programming, I cannot afford to have a lock-screen glitch make me lose all of my progress.
Comment 10 Maluchi Ugokwe Jr. 2023-01-05 21:14:02 UTC
This issue has not occurred since I have updated to 5.26.5.

Lockscreen looks good on two tests: one long term, one short term.

If after a day's worth of sporadic testing with varying times I no longer experience the issue, I will mark it as fixed in 5.26.5.
Comment 11 Nate Graham 2023-01-05 21:16:53 UTC
Aha, now I know what this bug is: Bug 457284. Other people in there are reporting that it's also fixed for them with 5.26.5.

*** This bug has been marked as a duplicate of bug 457284 ***
Comment 12 Maluchi Ugokwe Jr. 2023-01-06 05:34:42 UTC
(In reply to Nate Graham from comment #11)
> Aha, now I know what this bug is: Bug 457284. Other people in there are
> reporting that it's also fixed for them with 5.26.5.
> 
> *** This bug has been marked as a duplicate of bug 457284 ***

Yup. It's fixed.
Good work Nate and everyone else at KDE.
Comment 13 Maluchi Ugokwe Jr. 2023-01-10 02:33:25 UTC
The issue occurred again at 9:27 PM.

Same description. Put the computer to sleep. Reopen it. Pitch black and only a cursor is shown.
Comment 14 Maluchi Ugokwe Jr. 2023-01-10 04:08:58 UTC
(In reply to Maluchi Ugokwe Jr. from comment #13)
> The issue occurred again at 9:27 PM.
> 
> Same description. Put the computer to sleep. Reopen it. Pitch black and only
> a cursor is shown.

To continue, maybe this was a once in a blue moon kind of rarity for the bug I described to reoccur after I thought it was patched.

I don't know if the screenlocker crashed or if the bug was actually not completely fixed.
Comment 15 Nate Graham 2023-01-10 22:07:09 UTC
Darn. `coredumpctl --reverse` will tell you if there are any kscreenlocker_greet crashes. Do you see any?
Comment 16 SnDream 2023-01-13 12:01:21 UTC
I had a similar problem with my computer.
If I use the lock screen, the computer lags. When unlocked it becomes a black screen and only a cursor is visible.
If I just use power management to turn off the screen automatically, there is no problem.
My computer only has a single screen and no NVIDIA graphics card installed.

---

Operating System: KDE neon 5.26
KDE Plasma Version: 5.26.5
KDE Frameworks Version: 5.101.0
Qt Version: 5.15.8
Kernel Version: 5.15.0-57-generic (64-bit)
Graphics Platform: X11
Processors: 6 × Intel® Core™ i5-8500 CPU @ 3.00GHz
Memory: 15.3 GiB of RAM
Graphics Processor: Mesa Intel® UHD Graphics 630
Comment 17 Maluchi Ugokwe Jr. 2023-01-14 02:59:57 UTC
(In reply to Nate Graham from comment #15)
> Darn. `coredumpctl --reverse` will tell you if there are any
> kscreenlocker_greet crashes. Do you see any?

No. Only a signal abort on /usr/bin/kwin_x11 and then a subsequent signal segmentation fault on /usr/bin/kwin_x11

So something seriously wrong is happening with kwin for an ABRT to happen, which, if I had to guess, is what's causing this bug.
Comment 18 Maluchi Ugokwe Jr. 2023-02-17 03:39:40 UTC
Has not occurred on 5.27.0. Closing for now.
Comment 19 Maluchi Ugokwe Jr. 2023-02-23 21:19:13 UTC
This is still a bug. Issue is probably with Kwin as well as the screen-locker crashing.
Comment 20 Louis Moureaux 2023-03-01 16:58:58 UTC
I ran into a similar issue today that eventually degenerated into a total breakdown of the system. I was able to get a stack trace but it was lost when I had to reboot. It all started with a wakeup from sleep with an external screen. Restarting kwin_x11 seemed to have solved the issue, but one day later it started again.

The main symptom was kwin_x11 exiting with code 1 (hence no DrKonqi like in the original report). I recall the stack trace being in the default X11 error handler and seeing error messages about the X11 connection being lost. X11 was still running and other applications working properly. I also recall seeing effect classes and GL functions in the last levels of the trace (I know this is very unspecific but that's what I have).

At first the issue seemed to be related to a window being activated in fullscreen (both my own Qt app and Zoom). Eventually even notifications coming on top of a fullscreen window appeared to trigger a crash. It even crashed on alt-tabbing. Towards the end (after many kwin restarts) the system also prompted me to increase the inotify limit, but this may be unrelated.

I tried upgrading KWin and other packages on the unstable system but it didn't seem to help. I'm now running 5.27.2.

System info (after upgrade):
=======================

Operating System: KDE neon 5.27
KDE Plasma Version: 5.27.2
KDE Frameworks Version: 5.103.0
Qt Version: 5.15.8
Kernel Version: 5.19.0-32-generic (64-bit)
Graphics Platform: X11
Processors: 8 × Intel® Core™ i5-10310U CPU @ 1.70GHz
Memory: 15.3 Gio of RAM
Graphics Processor: Mesa Intel® UHD Graphics
Manufacturer: Dell Inc.
Product Name: Latitude 7410
Comment 21 Louis Moureaux 2023-03-03 16:05:30 UTC
Created attachment 156960 [details]
Stack trace of crashing kwin_x11

So it happened again and I was able to gdb in and grab a stack trace of crashing kwin_x11. It looks completely different from the symptoms I described yesterday, not sure why.
Comment 22 Maluchi Ugokwe Jr. 2023-03-06 09:37:03 UTC
(In reply to Louis Moureaux from comment #21)
> Created attachment 156960 [details]
> Stack trace of crashing kwin_x11
> 
> So it happened again and I was able to gdb in and grab a stack trace of
> crashing kwin_x11. It looks completely different from the symptoms I
> described yesterday, not sure why.

The issue that I reported seems to no longer be happening to me. Here are my specifications:

Operating System: Arch Linux 
KDE Plasma Version: 5.27.2
KDE Frameworks Version: 5.103.0
Qt Version: 5.15.8
Kernel Version: 6.0.12-zen1-1-zen (64-bit)
Graphics Platform: X11
I am using NVIDIA GeForce RTX 2060