[Bug 1129428] New: Oops'es and segfaults with kernel 4.12.14 and 5.0.1 on i7-7700K
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428 Bug ID: 1129428 Summary: Oops'es and segfaults with kernel 4.12.14 and 5.0.1 on i7-7700K Classification: openSUSE Product: openSUSE Distribution Version: Leap 15.0 Hardware: Other OS: Other Status: NEW Severity: Normal Priority: P5 - None Component: Kernel Assignee: kernel-maintainers@forge.provo.novell.com Reporter: jdsn@suse.com QA Contact: qa-bugs@suse.de Found By: --- Blocker: --- Created attachment 800236 --> http://bugzilla.opensuse.org/attachment.cgi?id=800236&action=edit core dump of mkinitrd On openSUSE Leap 15.0 with kernel 4.12.14-lp150.12.48-default I saw several segfaults in userspace programs. So I checked the journal (/var/log/messages) and even found 'Oops'es. The first Oops said: # Fixing recursive fault but reboot is needed So I rebooted (with the same kernel). After reboot I wanted to make sure that I have all latest fixes I installed all latest updates with # zypper ref # zypper up The latter crashed for some packages. As the machine is pretty new (compared to kernel 4.12) I thought about trying the latest kernel. So I added the repo # zypper ar http://download.opensuse.org/repositories/Kernel:/stable/standard/ kernel-stable # zypper ref # zypper in kernel-default-5.0.1 ... the latter failed repeatedly with segfaults. I then did # wget http://downloa.opensu....rpm # rpm -i kernel-default-5.0.1-3.1.g8c6a826.x86_64.rpm ... and this failed with segfaults too. Running it again like this # strace rpm -i kernel-default-5.0.1-3.1.g8c6a826.x86_64.rpm ... installed the kernel, but the creation of the initrd segfaulted reproducibly. So I enabled coredumps # echo /tmp/core > /proc/sys/kernel/core_pattern and again ran # mkinitrd ... it segfaulted and created the "core" that I attach to this bug. Running it again like this however # taskset 1 mkinitrd ... worked and created a new initrd Then I booted with the new kernel and retried # mkinitrd ... this worked (note, this was without 'taskset 1') But I still see segfaults in userspace. Most of them are in Firefox and Akonadi (KDE PIM backend). Short summary: With kernel 4.12.14-lp150.12.48-default - I see many userspace segfaults - "mkinitrd" segfaults - "taskset 1 mkinitrd" works With kernel 5.0.1-3.g8c6a826-default - the mkinird segfaults are gone - I still see userspace segfaults from * apparmor_parser * Web Content (firefox) * akonadi - Firefox does not segfault when run like # taskset 1 firefox Is there any way I can further help to chase down this issue? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c1
--- Comment #1 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c2
--- Comment #2 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c3
--- Comment #3 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c4
--- Comment #4 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c5
--- Comment #5 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c6
--- Comment #6 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c7
--- Comment #7 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c8
--- Comment #8 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c9
--- Comment #9 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c10
--- Comment #10 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c11
--- Comment #11 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c12
J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c13
Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c15
--- Comment #15 from J. Daniel Schmidt
Looks like pagetable corruption to me. How reproducible is this?
I have to run all desktop applications with taskset 1 or 2 in to be able to work. I was not able to submit this bug report due to too many crashes of firefox (tried 2 times).
Can you try with the latest KOTD from kernel.suse.com?
Sure. I am now running 5.0.2-1.g815c1bc-default from Kernel:HEAD and I still see segfaults. Will attach another journalctl output. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c16
--- Comment #16 from J. Daniel Schmidt
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428
http://bugzilla.opensuse.org/show_bug.cgi?id=1129428#c19
--- Comment #19 from J. Daniel Schmidt
So is there a kernel which *doesn't* show such segfaults and corruptions?
IOW, is there a configuration where that box works fine?
Oh, thanks for reminding me. I wanted to mention this in the initial description. All this started on 2019-02-27 11:49:52 when I updated to kernel-default-4.12.14-lp150.12.48.1.x86_64 from 4.12.14-lp150.11.something This update was done after a system freeze, but unfortunately I do not have any logs from that. On 2019-02-27 I saw several segfaults in journalctl. Then they reappeared after about two weeks of silence. Here are the "segfault" statistics (per day): 12 2019-02-27 1 2019-03-10 21 2019-03-11 60 2019-03-12 11 2019-03-13 1 2019-03-14 99 2019-03-15 -- You are receiving this mail because: You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply＠novell.com