Guardian Procedure Calls Reference Manual

FP_IEEE_EXCEPTIONS_GET_ Procedure
Summary
Syntax for C Programmers
Syntax for TAL Programmers
Returned Value
Considerations
Examples
Summary
The FP_IEEE_EXCEPTIONS_GET_ procedure reads the IEEE floating-point exception mask.
NOTE: This procedure is supported in all J-series and H-series RVUs and in G-series RVUs
beginning with G06.06. IEEE floating-point is available on all NB-Series and NS-Series systems,
and on all S-Series systems, except S70000 servers with NSR-G processors.
Syntax for C Programmers
#include <kfpieee.h>
fp_ieee_exceptions FP_IEEE_EXCEPTIONS_GET_ ( void );
Syntax for TAL Programmers
?SOURCE $SYSTEM.SYSTEM.KFPIEEE
Exceptions := FP_IEEE_EXCEPTIONS_GET_;
Returned Value
INT(32)
The exception flags. The exception flags bit values are:
Arithmetic calculations using either positive or negative infinity as an operand, zero
divided by zero, the square root of -1, the rem function with zero as a divisor (which
FP_IEEE_INVALID
causes divide by zero), comparisons with invalid numbers, or impossible
binary-decimal conversions.
Computing x/0, where x is finite and nonzero.FP_IEEE_DIVBYZERO
Result too large to represent as a normalized number.FP_IEEE_OVERFLOW
Result both inexact and too small to represent as a normalized number.FP_IEEE_UNDERFLOW
Result less accurate than it could have been with a larger exponent range or more
fraction bits. Most commonly set when rounding off a repeating fraction such as
FP_IEEE_INEXACT
1.0/3.0. Also set for underflow cases and some overflow cases, but not for division
by zero.
Considerations
• In addition to the above enumerated constants, a constant named FP_IEEE_ALL_EXCEPTS is
equivalent to a combination of all the exception bits.
• Once exception flags are set, they stay set until explicitly reset.
• More than one exception flag can result from a single floating-point operation.
646 Guardian Procedure Calls (F)