Re: [Xen-devel] [PATCH, RFC v2] x86/HVM: assorted RTC emulation adjustments (was Re: Big Bug:Time in VM goes slower...)

[tupeng212 <tupeng212@xxxxxxxxx>]

Dear Jan:

I didn't check the flipping in JVM, but from the printing period in xen, this setting happened(unit: ns):

976562, 976562, 976562, 976562, 15625000, 976562, 976562, 976562, 976562, 15625000 .....

and someone told me windows use these two rate/period only.

 

besides, I checked my former simple tester this morning after it ran for a whole night, it lagged much.

so there exists difference between virtualization and reality.

---

tupeng212

 

From: Jan Beulich

Date: 2012-08-16 16:22

To: tupeng212

CC: Yang Z Zhang; xen-devel; Keir Fraser; Tim Deegan

Subject: Re: [Xen-devel] [PATCH, RFC v2] x86/HVM: assorted RTC emulation adjustments (was Re: Big Bug:Time in VM goes slower...)

>>> On 15.08.12 at 16:12, tupeng212 <tupeng212@xxxxxxxxx> wrote:

> The results for me were these:

> 1 In my real application environment, it worked very well in the former 

> 5mins, much better than before,

>  but at last it lagged again. I don't know whether it belongs to the two 

> missed functions. I lack the 

>  ability to figure them out.

 

Did you check whether possibly the guest kernel started flipping

between to rate values? If so, that is something that we could

deal with too (yet it might be a bit involved, so I'd like to avoid

going that route if it would lead nowhere).

 

Jan

 

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel