On Sun, Nov 06, 2022 at 09:51:57AM +0100, Paul B Mahol wrote:
> On 11/5/22, Michael Niedermayer <michael@niedermayer.cc> wrote:
> > Fixes: signed integer overflow: -2889074 * 2048 cannot be represented in
> > type 'int'
> > Fixes:
> > 51363/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_BONK_fuzzer-5660734784143360
> >
> > Found-by: continuous fuzzing process
> > https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
> > Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
> > ---
> >  libavcodec/bonk.c | 7 ++++++-
> >  1 file changed, 6 insertions(+), 1 deletion(-)
> >
> > diff --git a/libavcodec/bonk.c b/libavcodec/bonk.c
> > index 471e09fe14..1695229dbd 100644
> > --- a/libavcodec/bonk.c
> > +++ b/libavcodec/bonk.c
> > @@ -363,12 +363,17 @@ static int bonk_decode(AVCodecContext *avctx, AVFrame
> > *frame,
> >              return ret;
> >
> >          for (int i = 0; i < samples_per_packet; i++) {
> > +            int64_t t64;
> >              for (int j = 0; j < s->down_sampling - 1; j++) {
> >                  sample[0] = predictor_calc_error(s->k, state, s->n_taps,
> > 0);
> >                  sample++;
> >              }
> >
> > -            sample[0] = predictor_calc_error(s->k, state, s->n_taps,
> > s->input_samples[i] * quant);
> > +            t64 = s->input_samples[i] * (int64_t)quant;
> > +            if ((int32_t)t64 != t64)
> > +                return AVERROR_INVALIDDATA;
> > +
> > +            sample[0] = predictor_calc_error(s->k, state, s->n_taps, t64);
> >              sample++;
> >          }
> >
> 
> NAK, using int64_t and thus slowing things down.

this code has little speed relevance, a single interation of this loop
takes over 1300 cpu cycles already

and it is faster, why, no clue, i guess it maybe reshuffles some
speed relevant bits

int
13145422 decicycles in QUA,    8192 runs,      0 skips
13175400 decicycles in QUA,    8192 runs,      0 skips
13260076 decicycles in QUA,    8192 runs,      0 skips

int64
13049729 decicycles in QUA,    8192 runs,      0 skips
13049418 decicycles in QUA,    8192 runs,      0 skips
13038855 decicycles in QUA,    8192 runs,      0 skips 



[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Old school: Use the lowest level language in which you can solve the problem
            conveniently.
New school: Use the highest level language in which the latest supercomputer
            can solve the problem without the user falling asleep waiting.