Re: toolchain/45310: some mixed int/float computations are wrong on mips64

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Tue, Aug 30, 2011 at 06:50:04PM +0000, David Holland wrote:
>  Here's a second test that could allow to narrow down the issue:
>  #include <stdio.h>
>  #include <stdlib.h>
>  #include <sys/types.h>
>  
>  main()
>  {
>          float a;
>          int64_t b;
>  
>          a = 33.000000;
>          b = a;
>  
>          printf("a %f b %d\n",a, (int)b);
>  }
>  
>  cuba:/home/bouyer>./test3
>  a 33.000000 b 0
>  
>  looks like the problem could be in __fixsfdi() (or one of the functions used 
> by
>  __fixsfdi()).

I __fixsfdi() uses __fixunssfdi() which uses double to uint casts, which is
the problem as shown by this new test:
#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>

main()
{
        double a;
        uint b;

        a = 33.000000;
        b = a;

        printf("a %f b %d\n",a, b);
}

cuba:/home/bouyer> gcc -o test4 test4.c
cuba:/home/bouyer> ./test4
a 33.000000 b 0

tracking this down, I ended up in fixunsgen_ieee754.c:__fixunsgen32.
It's called with
exp=5, sign=0, mant_dig=53, fracbits=20, frac={0, 0, 32768}

as exp - mant_dig is larger than 32, __fixunsgen32 returns 0.
mant_dig is a constant from mips/float.h; I've no idea if it's right or
wrong here ...

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--