Description of problem: On ia64 full virt Xen guest: # virt-what # echo $? 0 # The side result of this is that system is detected as physical system in entitlement webUI. Version-Release number of selected component (if applicable): virt-what-1.9-1.3.el5 How reproducible: Always
Paolo, the problem is that Xen IA64 HVM guest does not see the CPUID leaf, ie: # virt-what-cpuid-helper prints nothing (we would expect XenVMMXenVMM in an ordinary Xen HVM guest). Any ideas?
There are some Xen-specific devices. We might base detection on the presence of these, but I'd prefer to have pbonzini's opinion on the subject. # find /sys/ -name '*xen*' /sys/module/xen_vnif /sys/module/xen_vbd /sys/module/xen_platform_pci /sys/bus/xen /sys/bus/pci/drivers/xen-platform-pci /sys/devices/xen
Testing those devices is always effective, but it only detects guests which have PV-on-HVM drivers loaded. You can do that before the CPUID test, but it is not 100% effective (on either x86 or ia64).
Patches posted upstream: https://www.redhat.com/archives/virt-tools-list/2011-May/msg00068.html
(In reply to comment #6) > Patches posted upstream: > https://www.redhat.com/archives/virt-tools-list/2011-May/msg00068.html I forgot to mention ... In the specific corner case where we fall back to the timing attack method, we can't really tell what sort of virtualization is present, just that virtualization is there. So we print: # virt-what virt This may or may not require a change to subscription-manager. Can someone point me to the relevant piece of code?
I've pushed this upstream. The patches as posted contained a very small mistake, so there is an additional commit required to correct that. The full set of patches required to fix this is below. Included in virt-what 1.11: http://git.annexia.org/?p=virt-what.git;a=commitdiff;h=3ea952163f0c28cd88f8473bb015df892c4247bf http://git.annexia.org/?p=virt-what.git;a=commitdiff;h=2d9d8b9e404d831f7aecbea729d87d0db62cec23 Pushed after virt-what 1.11: http://git.annexia.org/?p=virt-what.git;a=commitdiff;h=d87e9cbae34b22c89e6094263ae44567f5b5bd90
Ping PMs ...
Host hardware: HP BL860c IA64 host, Host kernel: 2.6.18-238.12.1.el5xen Host xen: xen-3.0.3-120.el5 Verification steps: 1. Created one ia64 rhel5 hvm guest 2. install virt-what-1.10-1.el5.ia64.rpm in guest before the fix, run virt-what and return nothing 3. then update virt-what to latest virt-what-1.11-2.el5.ia64.rpm, run virt-what again, and this time it return correct info "xen xen-hvm" 4. In host, install virt-what and run virt-what, it return correct info "xen xen-dom0". Bug verified. Please check attached logs
Created attachment 503650 [details] ia64-host log
Created attachment 503651 [details] ia64-hvm-guest log
also tried with ia64 pv guest, both virt-what-1.10-1 old build and virt-what-1.11-2 new build works file. Both give correct result "xen xen-domU“
Created attachment 503655 [details] ia64-pv-guest
(In reply to comment #15) > Created attachment 503651 [details] > ia64-hvm-guest log This is correct, because the guest has PV-on-HVM drivers installed (notice /proc/sys/bus/xen directory in the guest). If you didn't have PV-on-HVM drivers installed, then you would see a slightly different output: virt The test results are still verified.
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on therefore solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHEA-2011-1015.html