Fix math/tst-CMPLX.c

[Joseph S. Myers]

On Mon, 6 Feb 2012, Marek Polacek wrote:

> On 02/03/2012 01:19 AM, Joseph S. Myers wrote:
> > (though it doesn't test the more interesting aspects of the macros - that 
> > they get cases involving signed zeros, infinities and NaNs right, which 
> > was why the macros were added in C11 - if you wanted to test those, it 
> > would probably be appropriate to start with the test I wrote for GCC, 
> > gcc.dg/torture/builtin-complex-1.c, and adapt it to use the macros instead 
> > of using __builtin_complex directly).
> 
> Something like this?
> 
> (Or, don't create a new test and add more checking into current
> tst-CMPLX.)

Yes, this looks good assuming it's been tested with 4.7.  I think adding a 
new test is better than putting more in the existing one.

-- 
Joseph S. Myers
joseph@codesourcery.com