[Libvirt-cim] ProcRASD VirtualQuantity Field issue for KVM on F10.
Hi,
The VirtualQuantity field of the ProcRASD is not getting set properly in
case of a *running* KVM guest on F10 with the libvirt-0.5.0-1.fc10.x86_64.
While this is not the case for running KVM guest on F9 with
libvirt-0.4.4-2.fc9.x86_64. Also, there is no problem with a *running*
Xen guest.
The following is the error message in the KVM debug output:
libvir: Domain error : invalid argument in virDomainGetVcpus
libvir: Domain error : invalid argument in virDomainGetVcpus
This is coming from the call to virDomainGetVcpus() from the
domain_vcpu_count() function defined in libxkutil/misc_util.c file.
I verified the signature of the virDomainGetVcpus() with the one given
on libvirt.org and it matches.
The problem is only when the guest is started, else for the defined
guest the VirtualQuantity field of the ProcRASD is set to the vcpu
value used.
_Steps to reproduce:
_1) Apply the patch "Fixing VSMS 09_procrasd_persist.py tc" and run
Or
2) use the following xml:
<domain type="kvm">
<name>procrasd_persist</name>
<on_poweroff>destroy</on_poweroff>
<on_crash>destroy</on_crash>
<uuid>ca793060-a7d0-42d9-9bd2-1db7149f2fe1</uuid>
<os>
<type>hvm</type>
<boot dev="hd"/>
</os>
<currentMemory>131072</currentMemory>
<memory>131072</memory>
<vcpu>1</vcpu>
<devices>
<disk type="file" device="disk">
<source file="/tmp/default-kvm-dimage"/>
<target dev="hda"/>
</disk>
<interface type="network">
<mac address="11:22:33:aa:bb:cc"/>
<source network="default"/>
</interface>
<input type="mouse" bus="ps2"/>
<graphics type="vnc" port="-1" listen="127.0.0.1"
keymap="en-us"/>
</devices>
</domain>
1) Define the guest using the above xml.
2) Verify the VirtualQuantity field of the ProcRASD. It will be set to
the value given in the vcpu.
3) start the guest.
4) Verify the VirtualQuantity field of the ProcRASD. It will be set to 0
and not to the value given in the vcpu.
I am not sure if this is expected behavior for KVM only or if Xen guest
should also behave in the same way, or is it a probable regression in
the latest libvirt library on F10.
I suspect the libvirt library though. Any thoughts ???
--
Thanks and Regards,
Deepti B. Kalakeri
IBM Linux Technology Center
deeptik(a)linux.vnet.ibm.com