_clear87, _clearfp

Get and clear the floating-point status word.

unsignedint_clear87(void);

unsignedint_clearfp(void);

For additional compatibility information, see Compatibility in the Introduction.

Libraries

Return Value

The bits in the value returned indicate the floating-point status before the call to _clear87 or _clearfp. See FLOAT.H for a complete definition of the bits returned by _clear87. Many of the math library functions modify the 8087/80287 status word, with unpredictable results. Return values from _clear87 and _status87 become more reliable as fewer floating-point operations are performed between known states of the floating-point status word.

Remarks

The _clear87 function clears the exception flags in the floating-point status word, sets the busy bit to 0, and returns the status word. The floating-point status word is a combination of the 8087/80287 status word and other conditions detected by the 8087/80287 exception handler, such as floating-point stack overflow and underflow.

_clearfp is a platform-independent, portable version of the _clear87 routine. It is identical to _clear87 on Intel® (x86) platforms and is also supported by the MIPS® and ALPHA platforms. To ensure that your floating-point code is portable to MIPS or ALPHA, use _clearfp. If you are only targeting x86 platforms, you can use either _clear87 or _clearfp.

Example

/* CLEAR87.C: This program creates various floating-point
 * problems, then uses _clear87 to report on these problems.
 * Compile this program with Optimizations disabled (/Od).
 * Otherwise the optimizer will remove the code associated with
 * the unused floating-point values.
 */

#include <stdio.h>
#include <float.h>

void main( void )
{
   double a = 1e-40, b;
   float x, y;

   printf( "Status: %.4x - clear\n", _clear87()  );

   /* Store into y is inexact and underflows: */
   y = a;
   printf( "Status: %.4x - inexact, underflow\n", _clear87() );

   /* y is denormal: */
   b = y;
   printf( "Status: %.4x - denormal\n", _clear87() );
}

Output

Status: 0000 - clear
Status: 0003 - inexact, underflow
Status: 80000 - denormal

Floating-Point Support Routines

See Also   _control87, _status87