[FFmpeg-devel] discussion around the best way to patch the swscale for Blackfin

[Marc Hoffman]

On 6/22/07, Panagiotis Issaris <takis.issaris at uhasselt.be> wrote:
>
> Hi Marc,
>
> Op 22-jun-07, om 13:51 heeft Marc Hoffman het volgende geschreven:
>
> > On 6/22/07, Reimar Doeffinger <Reimar.Doeffinger at stud.uni-
> > karlsruhe.de>
> > wrote:
> >>
> >> Hello,
> >> On Thu, Jun 21, 2007 at 11:00:28PM -0400, Marc Hoffman wrote:
> >>> Does this look like a clean integration for the unscaled variant
> >>> for the
> >>> swscale infrastructure?  Seems reasonably clean to me but another
> >>> set of
> >>> eyes would be great.
> >>
> >> Looks good, except
> >> 1) ff_bfin_get_unscaled_swscale (c); seems to be indented differently
> >> than the other code.
> >> 2) if (ENABLE_BFIN) ff_bfin_get_unscaled_swscale (c); avoids the
> >> #ifdef
> >> and matches current dsputil
> >
> >
> >
> > Hi Reimar,
> >
> >    On #1 I can't see that the indentation is incorrect. I must be
> > missing
> > your point.
> >
> > I'm actually not sure about this either do you mean this?
> >
> >
> > #ifdef ARCH_BFIN
> >     ff_bfin_get_unscaled_swscale (c);
> > #endif
> >
> >
> > converts into the following which to me is a reasonable idea
> >
> >     if (flags & SWS_CPU_CAPS_BFIN)
> >        ff_bfin_get_unscaled_swscale (c);
> >
> > but this would still need to be wrapped inside of an ifdef
> > ARCH_BFIN because
> > the function is defined for compilation only bfin.  Did I miss
> > something?
>
> The ENABLE_BFIN define is set in config.h by the configure script.
> You can therefore use the if (ENABLE_BFIN) ... as Reimar pointed out;
> GCC will optimize the call out in case ENABLE_BFIN is zero.


Thanks I will try this but it seems that if the optimizer is turned off this
would cause the code to generate an undefined reference.  Am I in correct?

Marc