Bug 1109592 - Machine check on boot with 3.14.6 or 3.14.7
Summary: Machine check on boot with 3.14.6 or 3.14.7
Keywords:
Status: CLOSED WORKSFORME
Alias: None
Product: Fedora
Classification: Fedora
Component: kernel
Version: 20
Hardware: i686
OS: Linux
unspecified
high
Target Milestone: ---
Assignee: Kernel Maintainer List
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2014-06-15 18:56 UTC by Peter Lister
Modified: 2014-09-03 12:18 UTC (History)
7 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2014-09-03 12:18:53 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)
Output of /proc/cpuinfo (1.84 KB, text/plain)
2014-06-15 18:56 UTC, Peter Lister
no flags Details

Description Peter Lister 2014-06-15 18:56:10 UTC
Created attachment 908934 [details]
Output of /proc/cpuinfo

Description of problem:

On booting, the kernel immediately reports a Machine Check Exception and hangs, then reboots after 30 seconds. 

Versions:
kernel-3.14.6-200 and 3.14.7-200
kernel-PAE-3.14.6-200 and 3.14.7-200

This does NOT occur with versions 3.14.5 or 3.14.4

How reproducible:

100% failure with 3.14.[67]
100% success with 3.14.[45]

Steps to Reproduce:

1. Upgrade to kernel 3.14.7-200
2. Boot from 3.14.7-200 kernel.
3. Immediate failure as described.

Actual results:

Machine check exception reported and kernel does not continue boot, and then attempts reboot after 30 seconds

Expected results:

Normal boot sequence of kernel

Additional info:

The system is a Gigabyte J1800 D2H. Details from /proc/cpuinfo attached.

I would like to attach the actual MCE, but I haven't been able to extract the info from mcelog. I might need some guidance on doing this.

I can boot from 3.14.5 just fine, so "high" priority, rather than urgent.

Comment 1 Peter Lister 2014-06-20 17:58:56 UTC
Problem also seen with 3.14.8-200

Comment 2 Peter Lister 2014-09-03 11:52:06 UTC
Well, I'm now on to 3.15.10 and the previous few fc20 kernel updates have been fine.

So this problem has "gone away". Has someone fixed it? Please reply if so and formally close the bug.

Or is there still some latent gremlin waiting to jump out at us again?

Comment 3 Josh Boyer 2014-09-03 12:18:53 UTC
It may well be fixed.  It may also just be hidden due to a different address layout or something similar.  There are around 10000 commits per major kernel release and we can't possibly track them all.  For now, we'll close this as WORKSFORME and if you see it again please reopen the bug.


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