Re: [Xen-devel] performace issue when turn on apicv

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 22.06.15 at 04:48, <jiangyifei@xxxxxxxxxx> wrote:
> We found that all the events are HVM_EMUL_LAPIC_START_TIMER. The number of 
> TRC_HVM_EMUL_LAPIC_STOP_TIMER event is zero.
> 
> Below is the details:
> 
> apicv on:
> TRC_HVM_EMUL_LAPIC_START_TIMER: count=111480
> TRC_HVM_EMUL_LAPIC_STOP_TIMER: count=0
> apicv off:
> TRC_HVM_EMUL_LAPIC_START_TIMER: count=13595
> TRC_HVM_EMUL_LAPIC_STOP_TIMER: count=0
> 
> I want to know what the bad effects of create_periodic_time() being 
> repeatedly called are.

Calling the function the way it is being done here, there is a drift in
the time the next interrupt gets signaled, due to the delay incurred
by the code doing the write emulation. See hvm/rtc.c (and its
history) for further details as well as how the issue got addressed
there. It would certainly be nice if you could see whether a similar
approach can be taken (and would help) here. (A fundamental
question, still unanswered, is over what time period these 111k
timer adjustments happened - said problem matters only when
there are many such updates in a relatively short period of time.)

Jan


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