[Users] can't add hosts with latest ovirt nightly (3.2)

Alon Bar-Lev alonbl at redhat.com
Thu Dec 6 07:15:47 UTC 2012



----- Original Message -----
> From: "Cristian Falcas" <cristi.falcas at gmail.com>
> To: dougsland at redhat.com
> Cc: users at ovirt.org
> Sent: Wednesday, December 5, 2012 10:25:07 PM
> Subject: Re: [Users] can't add hosts with latest ovirt nightly (3.2)
> 
> 
> 
> 
> 
> 
> 
> On Wed, Dec 5, 2012 at 8:05 PM, Douglas Landgraf <
> dougsland at redhat.com > wrote:
> 
> 
> Hi Christian,
> 
> 
> On 12/05/2012 09:54 AM, Cristian Falcas wrote:
> 
> 
> Hi,
> 
> The latest otopi installer is failing with the following error:
> 
> virtualization support AuthenticAMD (cpu: True, bios: False)
> and
> RuntimeError: Hardware does not support virtualization
> 
> Running rdmsr also says that virtualization is not enabled in bios:
> ./rdmsr --bitfield 4:4 0xc0010114
> 0
> 
> Is the bios flag mandatory for virtualization? On the same machine
> virt-manager works and can start virtual machines.
> 
> oVirt is based on KVM which requires full virtualization (bios flag
> enabled).
> In your case, looks like virt-manager is using pure qemu to emulate.
> 
> 
> 
> 
> I can't check now what is in bios, but I will update with the status
> when I have a chance to check.
> 
> Sure.
> 
> 
> 
> --
> Cheers
> Douglas
> 
> 
> 
> I checked and I had virtualization enabled in bios.
> 
> Also, I was mistaken and the result from rdmsr means bios is ok:
> result is supposed to be 0 (based on kvm-ok script).
> 
> I did the test with and without virtualization enabled in bios and I
> got this:
> - enabled = 00000100
> - disabled= 00011000
> 
> From internet: Bit 4 of MSR 0xc0010114 returns 0 means virtualization
> enabled
> 
> Based on this, I think that the check in
> /usr/share/ovirt-host-deploy/plugins/ovirt-host-deploy/vdsm/hardware.py
> it should not be != 0, but:
> def _svm_enabled_by_bios(self):
> ....
> (vm_cr & (1 << SVM_VM_CR_SVM_DISABLE)) == 0
> 
> I think the above means (8 & (1 << 4)) which is 0?

Right!
Thank you so much.
It was a typo while converting the legacy code to better syntax.
Had no AMD to test... :)

Fixed.

http://gerrit.ovirt.org/#/c/9758/

commit a271c2a8d53cbf13d639188bc5c3c0c9779cdebc
Author: Cristian Falcas <cristi.falcas at gmail.com>
Date:   Thu Dec 6 09:12:35 2012 +0200

    vdsm: hardware: fix typo in amd virtualization detection
    
    Change-Id: I2deb8f7ab0f306b15edc7b8238325ef6c359e809
    Signed-off-by: Alon Bar-Lev <alonbl at redhat.com>

diff --git a/src/plugins/ovirt-host-deploy/vdsm/hardware.py b/src/plugins/ovirt-host-deploy/vdsm/hardware.py
index c4685aa..275d1c8 100644
--- a/src/plugins/ovirt-host-deploy/vdsm/hardware.py
+++ b/src/plugins/ovirt-host-deploy/vdsm/hardware.py
@@ -111,7 +111,7 @@ class Plugin(plugin.PluginBase):
         MSR_VM_CR = 0xc0010114
 
         vm_cr = self._prdmsr(0, MSR_VM_CR)
-        return (vm_cr & (1 << SVM_VM_CR_SVM_DISABLE)) != 0
+        return (vm_cr & (1 << SVM_VM_CR_SVM_DISABLE)) == 0
 
     def _check_kvm_support_on_power(self):
         with open('/proc/cpuinfo', 'r') as f:

> 
> Best regards,
> 
> _______________________________________________
> Users mailing list
> Users at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
> 



More information about the Users mailing list