Description of problem: Only 1 processor can be detected in "systeminfo" of win2k3 guest with -smp 2 and ACPI multiprocessor HAL. Version-Release number of selected component (if applicable): # cat /etc/redhat-release Red Hat Enterprise Virtualization Hypervisor release 5.4-2.0.99 (8.2) kvm-83-81.el5 How reproducible: 100% Steps to Reproduce: 1.Start guest by: /usr/libexec/qemu-kvm -drive file=win2003-32-virtio.qcow2,if=ide -net nic,macaddr=20:20:20:00:16:75,model=virtio -net tap,script=/etc/qemu-ifup -rtc-td-hack -no-hpet -usbdevice tablet -m 1G -smp 2 -vnc :11 2.Issue "systeminfo" in cmd. Actual results: output of "systeminfo" says: 1 Processor(s) Installed. [01]: x86 Family 6 Model 3 Stepping 3 GenuineIntel ~2660 Mhz Expected results: output of "systeminfo" says: 2 Processor(s) Installed. Additional info: Tried 2 intel hosts and 1 one amd host, this problem only happened in intel hosts. 1.host1 cpuinfo processor : 3 vendor_id : GenuineIntel cpu family : 6 model : 23 model name : Intel(R) Core(TM)2 Quad CPU Q9400 @ 2.66GHz 2.host2 cpuinfo processor : 7 vendor_id : GenuineIntel cpu family : 6 model : 26 model name : Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz
Interesting. At least on my physical laptop (Core2 Duo CPU 9600), it also shows a single processor. What does the Task Manager say?
(In reply to comment #1) > Interesting. At least on my physical laptop (Core2 Duo CPU 9600), it also shows > a single processor. What does the Task Manager say? In windows task manager, i can see 1 cpu usage history, but in device manager, it shows 2 processors. I tried 32bitwin23k in physical machine (2 Core2 Duo CPU E8200), it shows 2 processors installed in output of "systeminfo".
(In reply to comment #2) > (In reply to comment #1) > > Interesting. At least on my physical laptop (Core2 Duo CPU 9600), it also shows > > a single processor. What does the Task Manager say? > > In windows task manager, i can see 1 cpu usage history, but in device manager, > it shows 2 processors. Just making sure - in Task Manager, did you select View -> CPU History -> One Graph per CPU? > I tried 32bitwin23k in physical machine (2 Core2 Duo CPU E8200), it shows 2 > processors installed in output of "systeminfo".
(In reply to comment #3) > (In reply to comment #2) > > (In reply to comment #1) > > > Interesting. At least on my physical laptop (Core2 Duo CPU 9600), it also shows > > > a single processor. What does the Task Manager say? > > > > In windows task manager, i can see 1 cpu usage history, but in device manager, > > it shows 2 processors. > > Just making sure - in Task Manager, did you select View -> CPU History -> One > Graph per CPU? Yes, sure, "One Graph per CPU" is selected(there is no option of "One Graph,All CPUs"). > > > I tried 32bitwin23k in physical machine (2 Core2 Duo CPU E8200), it shows 2 > > processors installed in output of "systeminfo".
This is caused by https://bugzilla.redhat.com/show_bug.cgi?id=504628 : try with -cpu ...,model=2 (the default was changed to model=3 in KVM-83-31).
Andrea sent a patch changing the default to model=6. Could you test with -cpu ...,model=6 too?
(In reply to comment #9) > Andrea sent a patch changing the default to model=6. Could you test with -cpu > ...,model=6 too? That worked well under KVM-87.
Paste the result of other wins guest for bug verification when patch is applied. guest -smp Actual vcpu 32bitvista 4 1 64bitvista 4 2 32bitwin23k 2 1 64bitwin23k 4 1 32bitwin28k 2 1 64bitwin28k 4 4 32bitwinxp 4 2 (for Linux guest(RHEL/Fedora) /proc/cpuinfo show the correct infomation)
Hi lihuang, what does that mean? Is windows not capable of seeing all vcpus because of a windows limitation (like installing itself with UP kernel or similar) or is this because it is using quirks/workarounds depending on cpu model? Until now I understood this bugreport was about windows seeing 2 vcpu but using only one with model=3. It's also hard to imagine this as a regression compared to the old KVM cpu model = 2. Also can you please try with "-cpu core2duo" on intel host to see if there's any difference? Thanks!
(In reply to comment #15) > Paste the result of other wins guest for bug verification when patch is > applied. > guest -smp Actual vcpu > 32bitvista 4 1 > 64bitvista 4 2 > 32bitwin23k 2 1 > 64bitwin23k 4 1 > 32bitwin28k 2 1 > 64bitwin28k 4 4 > 32bitwinxp 4 2 > > > (for Linux guest(RHEL/Fedora) /proc/cpuinfo show the correct infomation) The last line is irrelevant - on Windows, you do see all CPUs in device manager. Yet, when looking everywhere else, it uses a single CPU.
(In reply to comment #15) > Paste the result of other wins guest for bug verification when patch is > applied. > guest -smp Actual vcpu > 32bitvista 4 1 > 64bitvista 4 2 > 32bitwin23k 2 1 > 64bitwin23k 4 1 > 32bitwin28k 2 1 > 64bitwin28k 4 4 > 32bitwinxp 4 2 > > > (for Linux guest(RHEL/Fedora) /proc/cpuinfo show the correct infomation) How did you apply the patch? Did you compile it using brew yourself?
(In reply to comment #18) > (In reply to comment #15) > > Paste the result of other wins guest for bug verification when patch is > > applied. > > guest -smp Actual vcpu > > 32bitvista 4 1 > > 64bitvista 4 2 > > 32bitwin23k 2 1 > > 64bitwin23k 4 1 > > 32bitwin28k 2 1 > > 64bitwin28k 4 4 > > 32bitwinxp 4 2 > > > > > > (for Linux guest(RHEL/Fedora) /proc/cpuinfo show the correct infomation) > > How did you apply the patch? Did you compile it using brew yourself? the data is copied from test log. sorry . if the unclear description misled you. this is the behavior of the pre-patched kvm ( kvm-83-83.el5 ) the "-smp" column is the vcpu number in command line. the "Actual vcpu" column is the vcpu number reported from "systeminfo" in guest.
Verified this bug with the following guests in kvm-83-87, guest -smp vcpu seen in task manager 32bitwin23k 8 8 64bitwin23k 8 8 32bitwin28k 8 8 64bitwin28k 8 8 32bitwinxp 8 2 32bitvista 8 2 32bitvista 4 2 64bitvista 8 2 32bitwinxp has a bug-- Bug 508040 - Windows XP not using all CPUS. michen-->Andrea Arcangeli, Just want to confirm with you that if vista has the same problem with xp, thanks.
Confirmed with Gleb, cpu problem of vista and xp will be verified in bug 508040, and according to comment#25, change bug status to "verified".
The new model=6 default has a side-effect on win2k3-32bit, reported on bug #516762.
Release note added. If any revisions are required, please set the "requires_release_notes" flag to "?" and edit the "Release Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. New Contents: Windows 2003 32-bit guests with more than 4GB of RAM may crash on reboot. This can be worked around by setting a different CPU model (model=3) on the management interface. [instructions to do this on RHEV-M are needed]
Oops, added rel notes to the wrong bug, sorry.
Deleted Release Notes Contents. Old Contents: Windows 2003 32-bit guests with more than 4GB of RAM may crash on reboot. This can be worked around by setting a different CPU model (model=3) on the management interface. [instructions to do this on RHEV-M are needed]
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-2009-1272.html