[FFmpeg-devel] [PATCH] avcodec/aacenc: Avoid 0 lambda
Michael Niedermayer
michael at niedermayer.cc
Tue Jun 1 11:06:34 EEST 2021
On Mon, May 31, 2021 at 09:22:09AM +0200, Anton Khirnov wrote:
> Quoting Michael Niedermayer (2021-05-28 22:15:51)
> > Fixes: Ticket8003
> > Fixes: CVE-2020-20453
> >
> > Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
> > ---
> > libavcodec/aacenc.c | 3 ++-
> > 1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/libavcodec/aacenc.c b/libavcodec/aacenc.c
> > index aa223cf25f..e80591ba86 100644
> > --- a/libavcodec/aacenc.c
> > +++ b/libavcodec/aacenc.c
> > @@ -28,6 +28,7 @@
> > * TODOs:
> > * add sane pulse detection
> > ***********************************/
> > +#include <float.h>
> >
> > #include "libavutil/libm.h"
> > #include "libavutil/float_dsp.h"
> > @@ -852,7 +853,7 @@ static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
> > /* Not so fast though */
> > ratio = sqrtf(ratio);
> > }
> > - s->lambda = FFMIN(s->lambda * ratio, 65536.f);
> > + s->lambda = av_clipf(s->lambda * ratio, FLT_MIN, 65536.f);
>
> Would FLT_EPSILON not be more appropriate? IIUC FLT_MIN is still
> effectively zero.
yes, i was just trying to eliminate the x/0.
In theory, lambda = 0 or lambda = infinity is not semantically wrong, one
would mean smallest file disregarding distortion the other lowest distortion
disregarding size.
I do not know what the intend of the author of the original code was
But the places where divisions by lambda seem to happen semm all followed
by cliping so i guess the minimum doesnt matter and all reasonable
options will give the same result more or less.
ill change it to FLT_EPSILON
thx
[...]
--
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
Rewriting code that is poorly written but fully understood is good.
Rewriting code that one doesnt understand is a sign that one is less smart
than the original author, trying to rewrite it will not make it better.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 195 bytes
Desc: not available
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20210601/ee2295fc/attachment.sig>
More information about the ffmpeg-devel
mailing list