Re: [Qemu-devel] Re: [PATCHv2] Add rtc reset function.

[Dor Laor]

Gleb Natapov wrote:
> On Tue, Jun 09, 2009 at 12:23:21PM +0200, Jan Kiszka wrote:
>   
> > > +static void rtc_reset(void *opaque)
> > > +{
> > > +    RTCState *s = opaque;
> > > +
> > > +    s->cmos_data[RTC_REG_B] &= ~(REG_B_PIE | REG_B_AIE | REG_B_SQWE);
> > > +    s->cmos_data[RTC_REG_C] &= ~(REG_C_UF | REG_C_IRQF | REG_C_PF | REG_C_AF);
> > > +
> > > +    qemu_irq_lower(s->irq);
> > > +
> > > +#ifdef TARGET_I386
> > > +    if (rtc_td_hack)
> > > +           s->irq_coalesced = 0;
> > >       
> > Sorry, missed that on first run: Isn't it safer, clearer, more beautiful
> > to reset this unconditionally?
> >
> >     
> It doesn't really matter. From safety point of view the code is safe.
>
>   
Actually we should drop this rtc-td-hack flag entirely.
The time drift fix is now only compiled for x86 so it will be harmless 
for other arch.
Users are affected by not calling it although it is a must for every OS 
that uses rtc clock source
and cannot deal with drifts (aka all windows, and some of the Linux guests).

Anthony, will you accept such a patch to remove rtc-td-hack or at least 
change it's default?
> --
>                         Gleb.
>
>
>