Bug 198263 - 4.14.9: INFO: rcu_sched detected stalls on CPUs/tasks
Summary: 4.14.9: INFO: rcu_sched detected stalls on CPUs/tasks
Status: RESOLVED CODE_FIX
Alias: None
Product: Other
Classification: Unclassified
Component: Other (show other bugs)
Hardware: x86-64 Linux
: P1 blocking
Assignee: other_other
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2017-12-25 20:26 UTC by Stijn Tintel
Modified: 2021-12-08 09:52 UTC (History)
2 users (show)

See Also:
Kernel Version: 4.14.9
Subsystem:
Regression: No
Bisected commit-id:


Attachments
serial console output of a KVM guest with the problem (21.21 KB, text/plain)
2017-12-25 20:27 UTC, Stijn Tintel
Details
serial console output of a KVM guest with the problem (98.48 KB, text/plain)
2017-12-25 22:56 UTC, Stijn Tintel
Details
serial console output of a KVM guest with the problem with CONFIG_UNWINDER_FRAME_POINTER (24.16 KB, text/plain)
2017-12-26 01:04 UTC, Stijn Tintel
Details

Description Stijn Tintel 2017-12-25 20:26:25 UTC
Since kernel 4.14.9, I have several machines that no longer boot. Both are showing rcu stalls. This problem occurs on a bare metal Intel i5 2500 system running Gentoo hardened with glibc, and also on KVM guests running Gentoo hardened musl.

Downgrading to 4.14.8 makes the problem go away on both physical and virtual.
Comment 1 Stijn Tintel 2017-12-25 20:27:27 UTC
Created attachment 273299 [details]
serial console output of a KVM guest with the problem
Comment 2 Stijn Tintel 2017-12-25 20:29:54 UTC
I'm rebooting the machine again with 4.14.9 to see if it will produce some more useful backtrace when giving it some more time before rebooting it.
Comment 3 Stijn Tintel 2017-12-25 22:56:16 UTC
Created attachment 273305 [details]
serial console output of a KVM guest with the problem

Unfortunately nothing useful yet after >2h "uptime"
Comment 4 Thomas Mann 2017-12-26 00:11:16 UTC
same for me 4.14.8 boots without a problem

i use an Intel(R) Core(TM) i5-2540M cpu
Comment 5 Stijn Tintel 2017-12-26 01:04:45 UTC
Created attachment 273307 [details]
serial console output of a KVM guest with the problem with CONFIG_UNWINDER_FRAME_POINTER

Previous logs were with CONFIG_UNWINDER_ORC=y. Still doesn't look very useful though, to me.
Comment 6 Stijn Tintel 2017-12-27 14:02:35 UTC
Reported this downstream @ https://bugs.gentoo.org/642268 and while debugging further, this might actually be a compiler bug. 4.14.9 compiled with GCC 6.4.0 results in this problem. The same version and config compiled with GCC 5.4.0 boots fine.
Comment 7 Alexander Tsoy 2017-12-27 23:01:50 UTC
This bug is only reproducible with CONFIG_MCORE2=y
Comment 8 Stijn Tintel 2021-12-08 09:52:11 UTC
This has been fixed a long time ago.

Note You need to log in before you can comment on or make changes to this bug.