[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] Big Bug:Time in VM goes slower; foud Solution but demand Judgement! A Interesting Story!



>>> On 13.08.12 at 17:21, "Jan Beulich" <JBeulich@xxxxxxxx> wrote:
> - don't call rtc_timer_update() on REG_A writes when the value didn't
>   change (doing the call always was reported to cause wall clock time
>   lagging with the JVM under Windows)
> - in the same spirit, don't call rtc_timer_update() or
>   alarm_timer_update() on REG_B writes when the respective RTC_xIE bit
>   didn't change

Actually, this didn't go far enough yet: REG_B writes should
never cause any timers to get updated when merely one of the
xIE bits changes, as those bits shouldn't control the timers'
activity (and as a result, the eventual setting of the xF bits in
REG_C).

> - raise the RTC IRQ not only when RTC_UIE gets set while RTC_UF was
>   already set, but generalize this to alarm and periodic interrupts as
>   well
> - properly handle RTC_PF when the guest is not also setting RTC_PIE

In line with the above, this ought to also be done for AF and UF
(it may be the case for UF already).

Jan

> - also handle the two other clock bases




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


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.