[FFmpeg-devel] [PATCH 3/4] lavfi/nlmeans: Checking number precision when computing integral images

[mypopy at gmail.com]

On Wed, Mar 6, 2019 at 9:15 PM Carl Eugen Hoyos <ceffmpeg at gmail.com> wrote:
>
> 2019-03-06 12:31 GMT+01:00, mypopy at gmail.com <mypopy at gmail.com>:
> > On Wed, Mar 6, 2019 at 3:55 PM Carl Eugen Hoyos <ceffmpeg at gmail.com>
wrote:
> >>
> >> 2019-03-06 4:18 GMT+01:00, Jun Zhao <mypopydev at gmail.com>:
>
> >> > +    // accumulation of 8-bits uint_8 (uint8_t *src) into 32-bits
> >> > (uint32_t *ii)
> >> > +    // data type, it will have a risk of an integral value becoming
> >> > larger than
> >> > +    // the 32-bits integer capacity and resulting in an integer
> >> > overflow.
> >> > +    if ((w * h * UINT8_MAX) > UINT32_MAX)
> >>
> >> I don't think UINT8_MAX increases readability and I suspect
> >> this should contain "UINT32_MAX / (w*h)" or similar on
> >> one side.
> >>
> > You means like: UINT32_MAX/w < (UINT8_MAX * h) ?
>
> Actually: (UINT32_MAX / w < 255 * h)
> (But that may only be me)
>
Update V3 patch for this check in https://patchwork.ffmpeg.org/patch/12230/,
Tks