[FFmpeg-devel] [PATCH 01/12] Make av_d2q() manage the case where d is infinite.

Stefano Sabatini stefano.sabatini-lala
Sat Oct 2 11:54:51 CEST 2010


On date Saturday 2010-10-02 05:26:56 +0200, Michael Niedermayer encoded:
> On Fri, Oct 01, 2010 at 11:56:41PM +0200, Stefano Sabatini wrote:
> > On date Friday 2010-10-01 21:44:44 +0200, Michael Niedermayer encoded:
> > > On Fri, Oct 01, 2010 at 08:55:08PM +0200, Stefano Sabatini wrote:
> > > > On date Friday 2010-10-01 20:36:10 +0200, Michael Niedermayer encoded:
> > > > > On Fri, Oct 01, 2010 at 06:08:49PM +0200, Stefano Sabatini wrote:
> > > > > > ---
> > > > > >  libavutil/rational.c |    2 ++
> > > > > >  libavutil/rational.h |    2 ++
> > > > > >  2 files changed, 4 insertions(+), 0 deletions(-)
> > > > > > 
> > > > > > diff --git a/libavutil/rational.c b/libavutil/rational.c
> > > > > > index 0e0571e..b5af28e 100644
> > > > > > --- a/libavutil/rational.c
> > > > > > +++ b/libavutil/rational.c
> > > > > > @@ -98,6 +98,8 @@ AVRational av_d2q(double d, int max){
> > > > > >  #define LOG2  0.69314718055994530941723212145817656807550013436025
> > > > > >      int exponent= FFMAX( (int)(log(fabs(d) + 1e-20)/LOG2), 0);
> > > > > >      int64_t den= 1LL << (61 - exponent);
> > > > > > +    if (isinf(d))
> > > > > > +        return (AVRational){1,0};
> > > > > 
> > > > > if iam not mistaken then the int cast above has undefined behavior in this case
> > > > 
> > > > Why?
> > > 
> > > well i thought casting infinity to int can at least trigger exceptions on
> > > some platforms ...
> > > nan aint better of course
> > 
> > ...
> > -- 
> > FFmpeg = Fierce Foolish Maxi Portentous Extensive God
> 
> >  rational.c |    6 ++++--
> >  1 file changed, 4 insertions(+), 2 deletions(-)
> > 8fa431a6aa605f47f7dc500bffff31b7630e5800  0001-Avoid-cast-of-double-nan-to-int.patch
> > From 5aecbff4474037496828b7aefac65fd719a3a8d6 Mon Sep 17 00:00:00 2001
> > From: Stefano Sabatini <stefano.sabatini-lala at poste.it>
> > Date: Fri, 1 Oct 2010 18:00:31 +0200
> > Subject: [PATCH 01/13] Avoid cast of double nan to int.
> >  It may cause exceptions on some platform.
> 
> ok

Applied.

> [...]
> 
> >  rational.c |    2 ++
> >  rational.h |    2 ++
> >  2 files changed, 4 insertions(+)
> > d11ea8d0d46128505f8bee4b37679d4e9b8c12fe  0002-Make-av_d2q-manage-the-case-in-which-the-value-to-co.patch
> > From ae6bc76a5b29ab4ac799eba609957cd32094be8e Mon Sep 17 00:00:00 2001
> > From: Stefano Sabatini <stefano.sabatini-lala at poste.it>
> > Date: Fri, 1 Oct 2010 23:52:15 +0200
> > Subject: [PATCH 02/13] Make av_d2q() manage the case in which the value to convert is inf.
> > 
> > ---
> >  libavutil/rational.c |    2 ++
> >  libavutil/rational.h |    2 ++
> >  2 files changed, 4 insertions(+), 0 deletions(-)
> > 
> > diff --git a/libavutil/rational.c b/libavutil/rational.c
> > index efed674..956b939 100644
> > --- a/libavutil/rational.c
> > +++ b/libavutil/rational.c
> > @@ -100,6 +100,8 @@ AVRational av_d2q(double d, int max){
> >      int64_t den;
> >      if (isnan(d))
> >          return (AVRational){0,0};
> > +    if (isinf(d))
> > +        return (AVRational){1,0};
> 
> have you tried -inf?

Patch updated.
-- 
FFmpeg = Fantastic and Free Magic Proud Ermetic Gnome



More information about the ffmpeg-devel mailing list