mips SNaN/QNaN is swapped

[Alexandre Oliva]

On Apr  1, 2003, Richard Henderson <rth@redhat.com> wrote:

> You've missed the needed change to real_identical.  In this case,
> if canonical, you should ignore the significand.  This implies a
> needed change to real_hash as well.

AFAICT, this also exposes a flaw in the very notion of the `canonical'
bit.  __builtin_nan("0") and 0.0/0.0 will not pass real_identical(),
even though they produce the same bit pattern when converted to the
target format, if the target format has qnan_msb_set.  I still feel
this is heading the wrong way :-(

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                 aoliva@{redhat.com, gcc.gnu.org}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist                Professional serial bug killer