Re: [Xen-devel] Time Skewing on Windows XP

[Tim Deegan <tim@xxxxxxx>]

At 08:23 +0000 on 15 Mar (1363335830), Jan Beulich wrote:
> >>> On 14.03.13 at 21:23, Suravee Suthikulanit 
> >>> <suravee.suthikulpanit@xxxxxxx> wrote:
> > So far, this patch allows me to get the correct timing on the XP.
> 
> At the price of never stopping the periodic time - other than said
> yesterday, _that_ was the main point of the flow change. I.e.
> making the host wake up more frequently even on an otherwise
> idle system.
> 
> It ought to be possible to keep the timer off when unused by
> the guest _and_ have the time remain stable...

Sure, but right now, timekeeping in XP is totally borked.  So we should
probably revert 620d5dad54008e40798c4a0c4322aef274c36fa3 until a proper
fix can be found.

Cheers,

Tim.

> > --- a/xen/arch/x86/hvm/rtc.c
> > +++ b/xen/arch/x86/hvm/rtc.c
> > @@ -76,6 +76,15 @@ void rtc_periodic_interrupt(void *opaque)
> >       spin_unlock(&s->lock);
> >   }
> > 
> > +static void rtc_periodic_cb(struct vcpu *v, void *opaque)
> > +{
> > +    RTCState *s = opaque;
> > +
> > +    spin_lock(&s->lock);
> > +    s->hw.cmos_data[RTC_REG_C] |= RTC_PF | RTC_IRQF;
> > +    spin_unlock(&s->lock);
> > +}
> 
> And if we indeed revert that part of the original patch, then
> properly - this function then ought to replace the now dead
> rtc_periodic_interrupt().
> 
> Jan
> 
> > +
> >   /* Enable/configure/disable the periodic timer based on the RTC_PIE and
> >    * RTC_RATE_SELECT settings */
> >   static void rtc_timer_update(RTCState *s)
> > @@ -98,7 +107,7 @@ static void rtc_timer_update(RTCState *s)
> >           {
> >               period = 1 << (period_code - 1); /* period in 32 Khz cycles */
> >               period = DIV_ROUND(period * 1000000000ULL, 32768); /* in ns */
> > -            create_periodic_time(v, &s->pt, period, period, RTC_IRQ, 
> > NULL, s);
> > +            create_periodic_time(v, &s->pt, period, period, RTC_IRQ, 
> > rtc_periodic_cb, s);
> >               break;
> >           }
> >           /* fall through */
> > @@ -619,7 +628,6 @@ static uint32_t rtc_ioport_read(RTCState *s, 
> > uint32_t addr)
> >           s->hw.cmos_data[RTC_REG_C] = 0x00;
> >           check_update_timer(s);
> >           alarm_timer_update(s);
> > -        rtc_timer_update(s);
> >           break;
> >       default:
> >           ret = s->hw.cmos_data[s->hw.cmos_index];
> > diff --git a/xen/arch/x86/hvm/vpt.c b/xen/arch/x86/hvm/vpt.c
> > index 46d3ec6..9c1dbfb 100644
> > --- a/xen/arch/x86/hvm/vpt.c
> > +++ b/xen/arch/x86/hvm/vpt.c
> > @@ -259,8 +259,6 @@ int pt_update_irq(struct vcpu *v)
> > 
> >       if ( is_lapic )
> >           vlapic_set_irq(vcpu_vlapic(v), irq, 0);
> > -    else if ( irq == RTC_IRQ && pt_priv )
> > -        rtc_periodic_interrupt(pt_priv);
> >       else
> >       {
> >           hvm_isa_irq_deassert(v->domain, irq);
> 
> 
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxx
> http://lists.xen.org/xen-devel

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