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

Re: [Xen-devel] Bug: Windows 2003 fails to install on xen-unstable tip

At 12:07 +0100 on 29 Apr (1367237252), Ian Campbell wrote:
> On Mon, 2013-04-29 at 11:29 +0100, Tim Deegan wrote:
> > > > >> Now it is obvious that the combination of that flag and proper
> > > > >> RTC emulation can't work together,
> > > > >
> > > > > So does the flag actually require *improper* behaviour from the
> > > RTC
> > > > > (emulated or otherwise)?
> > 
> > Specifically emulated: that's what the 'E' in WAET stands for.  This
> > flag is to tell Windows that it needn't ack every RTC interrupt with an
> > IO read, so avoiding a bunch of VMEXITs (yay!).
> Ah, so this is actually a viridian thing? HVM domains do have a viridian
> mode already, can we (sanely) gate on that?
> And knowing what E stands for now I can now read "GOOD" as "good, if you
> are virtualising", which makes more sense.
> Don't you get an equivalent number of VMEXITs from the unnecessary
> interrupt injections? I'm clearly missing something...

Well, if you are using the RTC, it's a good thing. :)  And if you're not
using the RTC (and disable it by turning off the divider) it makes no

But if you enable the RTC (which is the default, and e.g. that's how
SeaBIOS leaves it) but aren't using it, it's very bad.  

I suppose another option for Xen would be to disable the RTC's vpt if
IRQ 8 isn't plumbed through the APICs?

> >   But clearly you can't
> > have that behaviour and also Jan's change to skip all the interrupts
> > entirely if the guest's not using them.
> Yes.
> > My inclination is:
> > 
> > - for 4.3, revert.  Even if we keep the various emulation changes, we
> >   should go back to having the VPT code inject the interrupts, and
> >   drop the idea of suppressing the timer when the interrupt's unused. 
> Should perhaps investigate the impact of not setting that specific bit?

Sure, that seems like a good idea, especially if it makes Windows DTRT
and uses a different time source.  But we'll need to handle guests that
have seen this bit set anyway (as it was set in 4.2).  So for now I
think we should drop the RTC changes for 4.3 unless someone's
volunteering to do some _serious_ testing on them.


Xen-devel mailing list



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