tor 2024-05-30 klockan 00:24 +0200 skrev Andreas Rheinhardt:
> Tomas Härdin:
> > static av_always_inline av_const int av_clip_intp2_c(int a, int p)
> > {
> > - if (((unsigned)a + (1 << p)) & ~((2 << p) - 1))
> > + if (((unsigned)a + (1U << p)) & ~((2U << p) - 1))
> > return (a >> 31) ^ ((1 << p) - 1);
> > else
> > return a;
>
> This will support p == 30 (but not 31); but the first change is not
> UB
> in this range.
p=31 is most definitely UB before this change. 1<<31 is signed overflow
with 32-bit int. The compiler has therefore been allowed to do whatever
for p=31 up until this point. To me it seems the intent of the code is
preserved
Personally I dislike bithacks because they are difficult to verify. A
good enough compiler will gain peephole optimizations for them sooner
or later anyway
/Tomas
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
