[bug] Slackware 15.0 : power-saving transistions cause the Linux kernel to crash

BernieK · 07-22-2022, 01:17 PM

Linux kernel 5.15.38 (x86_64) intermittently crashes during power-saving transitions when Xorg 1.20.14 is running with DPMS. The problem is most pronounced during transitions triggered by the "Offtime" setting in /etc/X11/xorg.conf

Code:

# in the Monitor section
Option  "DPMS"  "true"   # the default

# in the ServerFlags section
BlankTime     3
StandbyTime   5   # only uncommented when the DPMS option is "true"
SuspendTime  10   # only uncommented when the DPMS option is "true"
Offtime      30   # only uncommented when the DPMS option is "true"

There are no inactivity/power-management related kernel crashes when the X server is not running.

business_kid · 07-22-2022, 01:47 PM

Personally, I'd comment out the standby, which leaves 90% of power usage. I'd comment out the off time unless you're on a battery only laptop. There's usually a power manager in the window manager, and I wouldn't leave power management to the X server. So really, all those server flags could go.

Don't have us guessing blind. What window manager? You can leverage opinions of people using a similar setup.

BernieK · 07-22-2022, 02:10 PM

Quote:

Originally Posted by business_kid

What window manager?

fvwm

business_kid · 07-23-2022, 10:03 AM

Quote:

Originally Posted by BernieK

fvwm

Last time I used that it was pretty dumb. Try working with the suspend.

I have xfce, and it's got everything linked off the menu: Logout, suspend, hibernate,restart & shutdown.

BernieK · 08-13-2022, 07:10 PM

[ correction per business_kid (thanks!) ]
It appears upowerd is the daemon between the X server (Xorg) and the Linux kernel proper which is causing the kernel to crash during power-saving transitions which occur while the X server is running. Associated extracts from /var/log/messages* follow:

messages.2:May 17 23:51:09 luna kernel: upowerd[7764]: segfault at 28 ip 000055fb7f138a2e sp 00007fff5b96cd20 error 4 in upowerd[55fb7f135000+1d000]
messages.2:May 23 19:05:55 luna kernel: upowerd[4666]: segfault at 28 ip 000055accc47aa2e sp 00007ffc2f26fa40 error 4 in upowerd[55accc477000+1d000]
messages.2:May 24 19:49:09 luna kernel: upowerd[4348]: segfault at 28 ip 00005605e0ec6a2e sp 00007ffe28f483a0 error 4 in upowerd[5605e0ec3000+1d000]
messages.2:May 25 19:53:21 luna kernel: upowerd[2382]: segfault at 28 ip 0000558adb044a2e sp 00007ffcf0c1f300 error 4 in upowerd[558adb041000+1d000]
. . .
. . .
. . .
messages:Aug 21 17:13:43 luna kernel: upowerd[3543]: segfault at 28 ip 000055fbd5002a2e sp 00007ffc2166b3f0 error 4 in upowerd[55fbd4fff000+1d000]
messages:Aug 22 17:48:38 luna kernel: upowerd[8079]: segfault at 28 ip 00005585fe7a5a2e sp 00007ffe6898c9c0 error 4 in upowerd[5585fe7a2000+1d000]
messages:Aug 23 17:55:09 luna kernel: upowerd[3804]: segfault at 28 ip 000055cecae2ba2e sp 00007ffe8836ae50 error 4 in upowerd[55cecae28000+1d000]
messages:Aug 26 19:09:08 luna kernel: upowerd[7630]: segfault at 28 ip 00005643dbde1a2e sp 00007ffd41151f40 error 4 in upowerd[5643dbdde000+1d000]

business_kid · 08-14-2022, 03:55 AM

Things ending in 'd' are usually daemons, or servers. It has it's own man page. Dbus owns & starts it.

Error 4 is an 'interrupted system call.' Upowerd segfaulting sounds like an interruption to me. I'd upgrade or reinstall dbus and see if that helps. Maybe the kernel too.