[Bug libquadmath/94756] strtoflt128 assigns some subnormals incorrectly on MS Windows

["i.nixman at autistici dot org" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94756

--- Comment #17 from niXman <i.nixman at autistici dot org> ---
(In reply to niXman from comment #15)
> (In reply to Jakub Jelinek from comment #13)
> > Fixed now on the trunk.  I'd wait a little bit with backports, though I
> > think the gmp-param.h change doesn't need backporting.
> 
> unfortunately, the bug is still present on i686-mingw-w64.
> 
> this test has failed:
> 
> #include <stdio.h>
> #include <string.h>
> #include <assert.h>
> #include <quadmath.h> 
> 
> int main(void) {
>  __float128 f1, f2;
>  char buffer0[50], buffer1[50];
> 
>  f1 = strtoflt128("1e-4941", NULL);
>  f2 = 1e-4941Q;
> 
>  quadmath_snprintf(buffer0, 45, "%Qa", f1);
>  printf("f1: %s\n", buffer0);
> 
>  quadmath_snprintf(buffer1, 45, "%Qa", f2);
>  printf("f2: %s\n", buffer1);
> 
>  fflush(stdout);
> 
>  assert(strcmp(buffer0, buffer1) == 0);
>  
>  return 0;
> }
> 
> 
> 
> the output:
> f1: 0x0.0000000cc64f1cc4376f7da08f39p-16382
> f2: 0x0.000000014707e946d257f2f674b9p-16382

im sorry, my bad.
I looked at one test code, but compiled another %)