i915 bug: GPU HANG. Kernel: 5.7.10-1
This is another one bug on the same system as was #2187 (closed) but under another conditions:
-) kernel version now is 5.7.8-8-MANJARO.
-) again current app is opera and there was playing video youtube tab, opened new youtube homepage tab and mouse double click on a link to open a link in a new window (the same ability as press on scroll wheel, it is typical setting for that double (left+right) mouse btn click in KDE: Settings -> Input devices -> Mouse -> Press left and right buttons for middle-click). I strongly believe that it is not the root problem: recently I had many freezes without mouse clicks.
All stuff below is only copy-paste-and-edit according to the current case. Template taken from the sort of the same issue linked above (# 2187).
Further ticket is #2189 (closed).
Key journalctl -b-1 -r
lines
...
Jul 16 04:38:58 pc kernel: GPU crash dump saved to /sys/class/drm/card0/error
Jul 16 04:38:58 pc kernel: The GPU crash dump is required to analyze GPU hangs, so please always attach it.
Jul 16 04:38:58 pc kernel: drm/i915 developers can then reassign to the right component if it's not a kernel issue.
Jul 16 04:38:58 pc kernel: Please see https://gitlab.freedesktop.org/drm/intel/-/wikis/How-to-file-i915-bugs for details.
Jul 16 04:38:58 pc kernel: Please file a _new_ bug report at https://gitlab.freedesktop.org/drm/intel/issues/new.
Jul 16 04:38:58 pc kernel: GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
Jul 16 04:38:58 pc kernel: i915 0000:00:02.0: GPU HANG: ecode 9:1:87f9ffff, in opera [1215]
Jul 16 04:38:58 pc kernel: i915 0000:00:02.0: opera[1215] context reset due to GPU hang
Jul 16 04:38:58 pc kernel: i915 0000:00:02.0: Resetting rcs0 for preemption time out
...
Platform
Manjaro Linux 20.0.3 64 bit KDE, unstable branch (it has latest updates in Manjaro project).
Kernel: uname -r
: 5.7.8-8-MANJARO.
Hardware briefly
1-year old mini-pc with the only GPU which is CPU on-die (on the same substrate as CPU is: Intel UHD Graphics 620). No any other GPU present in PC (ever).
Kernels affected
looks like all current of the Manjaro Linux OS. They are (at least)
4.19 LTS family and latest 4.19.132-1 LTS also,
5.4 LTS family (probably latest 5.4.51-1 also),
5.6 family (probably latest 5.6.19-2 also),
5.6-real-time family (probably latest 5.6.17_rt9-2 also),
5.7 family and latest 5.7.8-8 also,
5.8rc family and latest 5.8rc5 also.
Kernel. Usual behaviour
4.19: restore state of graphics by resetting graphics (see 'More info by me also located on' link below),
5.7: hangs up but only picture (blinking picture with tens and hundreds colour rectangles in the black backround), sound is playing back still, second terminal (Ctrl+Alt+F2) is available and working. Here that current ticket is then 5.7.x kernel remaining 'alive' which is usual and info gathered via second terminal session.
5.8rc (attention: not 'rt'): freezes completely OS: picture like still picture/photo, sound stops, no any terminals available: pc no responds.
Additionally about kernels.
!!!Could be important!!!
( More info by me also located on: https://forum.manjaro.org/t/random-freezing-with-resetting-rcs0-for-hang-on-rcs0/119313/34 )
It contains logs and GUI messages how another (not that current kernel version but) v4.19 LTS in KDE prevents graphics from crash and resets graphics and continue to let user have his/her working session.
Moreover: Manjaro Linux LiveCD v2003 (USB flash stick, was downloaded about 2 weeks ago) hangs on also during work as LiveCD. It has 5.6.15-1-MANJARO kernel built-in.
Features: looks like software-independent bug: software mentioned in "General steps to reproduce an issues" item below.
General steps to reproduce an issues
I have tons of such issues. They based on randomly time and actions manner. Can't select the exact user behaviour, spontaneous error: it could be on logon screen before user input or after 12 hours of work. Usually after 30-120 minutes of very typical user actions in text editor, media player, web browser, desktop icons selection, selecting item in context menu on system tray icon, etc.
How often does the steps listed above trigger the issue
Issue occurred about 50-60 times for about last 2.5 weeks period. OS installation was 3-4 month ago and before that 2.5 weeks ago works well. May be software update was the cause of the problem, can't trace this now. Never saw such hard to catch bug and recently it became very very frequently several times per day at very least.
Display connector: (such as HDMI, DP, eDP, ...)
Monitor port: DVI-D.
PC port: HDMI (probably 1.4a).
Single cable (factory-made: without any adapters): one side has HDMI connector (PC) - other side has DVI-D connector (monitor).
Screenshot or photo (a picture is worth a thousand words)
see some of them by 'More info by me also located on' link to the manjaro forum (located above).
Attached files
sudo cp /sys/class/drm/card0/error Desktop/error
inxi -Fxxxz > inxi.txt
sudo dmidecode > dmidecode.txt
sudo dmesg > dmesg.txt
mhwd -li > mhwd.txt mhwd -la >> mhwd.txt
(than was shutdown now -r
and later:)
xrandr --verbose > xrandr.txt
journalctl -b-1 -r > journalctl.txt
KDE. System settings -> System information
Operating System: Manjaro Linux
KDE Plasma Version: 5.19.3
KDE Frameworks Version: 5.72.0
Qt Version: 5.15.0
Kernel Version: 5.7.8-8-MANJARO
OS Type: 64-bit
Processors: 4 × Intel Core i5-8250U CPU @ 1.60GHz
Memory: 31.1 GiB of RAM
Graphics Processor: Mesa Intel UHD Graphics 620