View Issue Details
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0015067||CentOS-6||kernel||public||2018-07-17 06:55||2018-10-11 06:28|
|Target Version||Fixed in Version|
|Summary||0015067: Kernel panic with EL7 KVM guest on EL6 host on AMD server|
|Description||I have AMD servers running CentOS 6.10 as KVM host and on it guests|
running CentOS 7.5.
Since the latest kernel update on CentOS 6.10,
kernel-2.6.32-754.2.1.el6.x86_64, I can't boot the CentOS 7.5 guests with
Reverting CentOS 6.10 host to kernel-2.6.32-754.el6.x86_64 fixes the problem.
Screendump of the panic is attached, not very helpful I guess.
The CPU models where I could test it are:
cpu family : 15
model : 65
model name : Dual-Core AMD Opteron(tm) Processor 2218
stepping : 3
cpu family : 16
model : 6
model name : AMD Athlon(tm) II Neo N36L Dual-Core Processor
stepping : 3
Didn't test on newer AMD CPU family.
I'm wondering, does someone see the same issue? Does it not happen on
|Steps To Reproduce||On KVM host run CentOS 6.10 with latest kernel-2.6.32-754.2.1.el6.x86_64|
On KVM guest run CentOS 7.5 with latest kernel-3.10.0-862.6.3.el7.x86_64
start CentOS 7.5 guest with "virsh start el7"
It crashes quite early in the boot process.
el7-panic.png (14,391 bytes)
el7-panic.png (14,391 bytes)
I've created this bug for CentOS 6.10 because the regression comes with the upgrade from kernel-2.6.32-754.el6.x86_64 to kernel-2.6.32-754.2.1.el6.x86_64.
However, only CentOS 7.5 crashes on the 6.10 KVM host. I've tested CentOS 6.10, CentOS 5.11 and FreeBSD 11.2 as guests and they are all happy.
So in the end it might also be a bug in CentOS 7.5 kernel-3.10.0-862.6.3.el7.x86_64.
|I suggest you _reinstall_ the kernel and see if that fixes the issue.|
As written above I've verified the issue on two different servers with two different VM guests in two different locations. I'm wondering why you believe reinstalling would fix the issue? Am I missing something obvious?
Anyway I've done as you suggested, reinstalled the latest kernel in both the VM host and guest. It still crashes in the same way.
Sorry for the "noise". :(
I have now heard what looks like a similar case. The workaround in that case was to downgrade 2 packages on the host to:
Correction: those are the latest. You need to downgrade them.
I can confirm that the following works:
kernel-2.6.32-754.el6.x86_64 & qemu-img/qemu-kvm0.12.1.2-2.506.el6_10.1.x86_64
kernel-2.6.32-754.2.1.el6.x86_64 & qemu-img/qemu-kvm-0.12.1.2-2.503.el6_9.6.x86_64
but this doesn't:
kernel-2.6.32-754.2.1.el6.x86_64 & qemu-img/qemu-kvm0.12.1.2-2.506.el6_10.1.x86_64
|OK, this problem was confirmed with RHEL. So this is not a CentOS issue.|
Thanks, is there any more info available somewhere, like a bz# or something else?
BTW, I've just tested with kernel-3.10.0-862.9.1.el7.x86_64 on the CentOS
7 guest and it still crashes the same way.
|I have no further info. Could you file a bug report upstream (Red Hat)? That seems to be the best way to collect more info.|
same issue on kernel-22.6.32-754.3.5.el6.x86_64 & qemu-img/qemu-kvm0.12.1.2-2.506.el6_10.1.x86_64
Is there a solution to the problem?
sorry, Kernel is: kernel-2.6.32-754.3.5.el6.x86_64
kernel-2.6.32-754.el6.x86_64 & qemu-kvm-0.12.1.2-2.503.el6_9.6.x86_64 AND qemu-img-0.12.1.2-2.503.el6_9.6.x86_64
kernel-2.6.32-754.3.5.el6.x86_64 & qemu-kvm-0.12.1.2-2.503.el6_9.6.x86_64 AND qemu-img-0.12.1.2-2.503.el6_9.6.x86_64
This doesn't work:
kernel-2.6.32-754.3.5.el6.x86_64 & qemu-kvm-0.12.1.2-2.506.el6_10.1.x86_64 AND qemu-img-0.12.1.2-2.506.el6_10.1.x86_64
With latest CentOS 7.5 guests:
Upstream solution article"
"RHEL7 guest with AMD cpu fails to boot after upgrading underlying RHEL6 host with AMD cpu to kernel-2.6.32-754.3.5.el6.x86_64"
is it possible to see the content here from https://access.redhat.com/solutions/3601721
You need an active RH subscription to read the content. Basically what it says is that the issue is being tracked. A workaround is already presented in this bug tracker.
Is a solution to this problem in sight, or is the workaround still the current solution?
Good news. The latest kernel (2.6.32-754.6.3) that has just been released fixed the issue.
From the changelog:
- [x86] Add host_initiated check in reading MSR_AMD64_VIRT_SPEC_CTRL (Wei Huang) 
- [x86] KVM: pass host_initiated to functions that read MSRs (Wei Huang) 
thank you so much for posting, that the issue was solved!
I can confirm that the latest kernel-2.6.32-754.6.3.el6 fixes the issue.
|Thanks for the feedback. Closing as 'resolved'.|
|2018-07-17 06:55||simix||New Issue|
|2018-07-17 06:55||simix||File Added: el7-panic.png|
|2018-07-17 06:55||simix||Tag Attached: 2.6.32-754.2.1.el6.x86_64|
|2018-07-17 07:05||simix||Note Added: 0032285|
|2018-07-17 13:27||toracat||Note Added: 0032287|
|2018-07-17 14:05||simix||Note Added: 0032288|
|2018-07-17 15:58||toracat||Note Added: 0032290|
|2018-07-17 16:00||toracat||Note Added: 0032291|
|2018-07-17 17:06||simix||Note Added: 0032295|
|2018-07-17 17:19||toracat||Note Added: 0032296|
|2018-07-18 06:07||simix||Note Added: 0032297|
|2018-07-19 21:39||toracat||Note Added: 0032321|
|2018-08-17 13:54||anderson_thomas||Note Added: 0032516|
|2018-08-17 14:23||anderson_thomas||Note Added: 0032519|
|2018-08-17 15:13||anderson_thomas||Note Added: 0032520|
|2018-08-18 11:33||isvelov||Tag Attached: "Network"|
|2018-08-18 11:33||isvelov||Tag Detached: "Network"|
|2018-09-09 04:59||toracat||Note Added: 0032685|
|2018-09-09 09:44||anderson_thomas||Note Added: 0032686|
|2018-09-10 00:21||toracat||Note Added: 0032688|
|2018-09-10 00:22||toracat||Status||new => acknowledged|
|2018-10-10 03:07||anderson_thomas||Note Added: 0032890|
|2018-10-10 06:15||toracat||Note Added: 0032892|
|2018-10-10 06:32||toracat||Note Added: 0032893|
|2018-10-10 07:53||anderson_thomas||Note Added: 0032895|
|2018-10-11 04:51||simix||Note Added: 0032908|
|2018-10-11 06:28||toracat||Status||acknowledged => resolved|
|2018-10-11 06:28||toracat||Resolution||open => fixed|
|2018-10-11 06:28||toracat||Note Added: 0032910|