On date Sunday 2014-07-06 13:46:31 +0300, Andrey Utkin encoded:
> 2014-07-05 16:19 GMT+03:00 Nicolas George <geo...@nsup.org>:
> > Le sextidi 16 messidor, an CCXXII, Andrey Utkin a écrit :
> >> +@item eif
> >
> > I find the name obscure. Where does it come from?
>
> Yes, it is somewhat obscure. It comes from "expr_int_format".
>
> >> + if (!strchr("xXdu", argv[1][0])) {
> >
> > You should probably check that argv[1][1] is 0.
>
> Can be added to check for stray trailing symbols in passed argument,
> but this doesn't affect the algorithm, it uses only first character.
>
> >> + feclearexcept(FE_ALL_EXCEPT);
> >> + intval = res;
> >> + if ((ret = fetestexcept(FE_INVALID|FE_OVERFLOW|FE_UNDERFLOW))) {
> >> + av_log(ctx, AV_LOG_ERROR, "Conversion of floating-point result to
> >> int failed. Control register: 0x%08x. Conversion result: %d\n", ret,
> >> intval);
> >> + return AVERROR(EINVAL);
> >
> > Is this portable enough? Anyway, AFAIK, we never do that kind of test in the
> > code.
>
> This must be coming from C standard. I consulted with this CERT C
> Secure Coding Standards chapter:
> https://www.securecoding.cert.org/confluence/display/seccode/FLP03-C.+Detect+and+handle+floating-point+errors
Patch applied, further tweaking can be done later.
--
FFmpeg = Freak and Frenzy Mean Powerful Enchanting God
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
