Bug 107351

Summary:	At higher load, system sometimes freezes completely
Product:	Other	Reporter:	remus (remus)
Component:	Other	Assignee:	other_other
Status:	NEW ---
Severity:	high	CC:	remus
Priority:	P1
Hardware:	x86-64
OS:	Linux
Kernel Version:	4.2.2-1 and higher	Subsystem:
Regression:	No	Bisected commit-id:
Attachments:	Output of `journalctl -b -1` after restart Output of `journalctl -b -1` after restart (#2)

Description remus 2015-11-06 09:43:54 UTC

Created attachment 192231 [details]
Output of `journalctl -b -1` after restart

I've already described the problem here:

https://bbs.archlinux.org/viewtopic.php?pid=1575711#p1575711

After a recent Kernel update (to 4.2.5-1, at the time), my system started to (more or less reliably) freeze after a few minutes, when under higher load (i.e., multiple programs open and starting some more programs/services). The freeze is complete, I can neither use the mouse nor keyboard or switch over to another virtual terminal with Ctrl+F*. The sound played in the last two seconds keeps looping.

After reading the linked thread, I downgraded to 4.2.2-1 where I didn't have the problem for a few days (so it seems it's way stabler), but today it occurred again, same as the other times.

The last times, there was nothing interesting in journalctl afterwards, but this time it seems there could be something. Can't be sure whether it's exactly from the time of the freeze, though, or a few minutes earlier. If the problem occurs again, I'll report back with (hopefully) more information. Please also tell me what else could be helpful in pinning this down.

I'm aware that such a problem is very hard to pinpoint and solve, but I thought I'd still better report it. I've never compiled the Kernel myself, though, so that might be hard for me if it's required to solve this. But I'd be open to try, if someone has a suggestion/patch.


System:
Distro: Arch Linux
CPU: AMD Phenom II X4 965

Comment 1 remus 2015-11-06 18:22:50 UTC

Created attachment 192301 [details]
Output of `journalctl -b -1` after restart (#2)

Happened again, this is the journal from this time.
Here, there are actually kernel errors at the end, which looks more promising, but the last one occurred 19 seconds before the freeze. Also, since they weren't there in the logs of the last freezes, I don't know whether that can really be related.

Anyways, I'm now going back to Kernel 4.1.12-1-lts, so (hopefully) won't have any more data points to report (or maybe one now, before the restart). If there's still anything I can do to help solve this, please tell me!