>Thanks Xin, for the explaination. I am only testing the 32bit PAE and
>64bit XEN Hypervisors because we are interested in support for RHEL4
>32bit SMP kernels. We recently changed our 32bit hypervisor build to
>enable PAE so we were testing a valid combination.
>Are you close to submitting your PS support patch? I'm very interested
>in seeing 32bit SMP guests working on 64bit hypervisors.
Xiaohui is finishing the patch, hope to send it out very soon, and RHEL4
IA32 SMP kernel runs fine :-)
>Are there release notes indicating what is supported in the 3.0.2 test
>branch and what is planned to be supported in the next release?
"xm info" can report the info.
>From: Li, Xin B [mailto:xin.b.li@xxxxxxxxx]
>Sent: Thursday, May 04, 2006 11:38 AM
>To: Ed Smith; xen-devel@xxxxxxxxxxxxxxxxxxx
>Subject: RE: [Xen-devel] Testing status of fully virtualized guests
>(Intel VT) on 64bit XEN unstable
>Thanks a lot for your tests on HVM SMP guests support, here I think we
>need make it more clear, actually on x86 we have 3 types of xen
>1) 32bit none-PAE xen hypervisor;
>2) 32bit PAE xen hypervisor;
>3) 64bit xen hypervisor,
>and 3 types of HVM guests:
>a) 32bit none-PAE HVM guests;
>b) 32bit PAE HVM guests;
>c) 64bit HVM guests.
>On 1), we only support a)
>On 2), we support a), and we are planning to support b), yet
>this is not
>done yet :-( On 3), we support a), b) and c).
>>1. 32bit UP guest fail ltp gettimeofday02:
>> "Time is going backwards"
>>2. 32bit and 64bit guests fail first boot attempt after booting XEN:
>> "(XEN) Failed vm entry, domain_crash_sync called from vmx.c"
>>3. 32bit SMP guest hangs on boot after:
>> "Uncompressing Linux... Ok, booting the kernel."
>The default Redhat 32bit SMP kernel is PAE enabled, so if you
>to run it on 1), you got such results :-(, for this kind of
>I suggest you compile a none PAE 32bit SMP kernel by your own.
>If you are running it on 64bit xen hypervisor, only FC5 32bit
>can successfully boot currently. The reason is on 2 sides, one is
>because PS support is still not complete, and the other is old version
>(2.6.12-) Linux kernel can not handle such cases, SMP code and PAE code
>are mixed there.
>We have a patch in hand to complete PS support on x86_64, and
>RHEL4 32bit SMP can successfully boot :-)
>Hope my explaination is clear for you :-)
>>4. 64bit SMP guest fails testing:
>> domU hangs under load
>>5. 64bit SMP guest report these messages the XEN console:
>> "(XEN) spurious IRQ irq got=-1"
>>6. 64bit SMP guest report these messages the XEN console during
>> "(XEN) <vlapic_accept_irq>level trig mode repeatedly for vector
Xen-devel mailing list