On Fri, 5 Apr 2013, Gerd Hoffmann wrote:
> >> -#else
> >> ; /* nothing */
> >> -#endif
> >> } else {
> >> /* we don't get update notifications, thus use the
> >> * sledge hammer approach ... */
> >
> > You might as well remove the if () nothing; case.
>
> Yep, will do.
>
> >> @@ -785,6 +755,20 @@ static void xenfb_update(void *opaque)
> >> xenfb->up_fullscreen = 0;
> >> }
> >>
> >> +static void xenfb_update_interval(void *opaque, uint64_t interval)
> >> +{
> >> + struct XenFB *xenfb = opaque;
> >> +
> >> + if (xenfb->feature_update) {
> >> +#ifdef XENFB_TYPE_REFRESH_PERIOD
> >> + if (xenfb_queue_full(xenfb)) {
> >> + return;
> >> + }
> >> + xenfb_send_refresh_period(xenfb, interval);
> >
> > Shouldn't we be updating xenfb->refresh_period here? And shouldn't we
> > call xenfb_send_refresh_period only if interval !=
> > xenfb->refresh_period?
>
> > On the other hand if refresh_period is not useful anymore, shouldn't
> > we remove it from struct XenFB?
>
> xenfb_update_interval is only called when interval changes, which I
> think means we don't need refresh_period any more, correct?
that's right