mips SNaN/QNaN is swapped

Alexandre Oliva aoliva@redhat.com
Tue Apr 1 20:02:00 GMT 2003

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

> On Tue, Apr 01, 2003 at 03:12:01PM -0300, Alexandre Oliva wrote:
>> Err...  Surely neither r->signalling nor r->canonical will be set for
>> a rvc_normal.  What's to be fixed?

> They are not initialized for rvc_normal.

Eeek...  I'd checked all of the decode functions, but I failed to
check other means to create real values.  /me doesn't like
uninitialized values, but, yeah, this must have been the reason why I
got a bootstrap compare failure.  Testing a revised patch now.

>> Do you mean remove the fallthrough
>> and duplicate the significand code in the rvc_normal case?

> Or move the significand code as I did for real_identical.

This will do :-)

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

More information about the Gcc-patches mailing list