path: root/man3/fenv.3

'\" t
.\" Copyright (c) 2000 Andries Brouwer (aeb@cwi.nl)
.\"
.\" SPDX-License-Identifier: GPL-2.0-or-later
.\"
.\" 2000-08-14 added GNU additions from Andreas Jaeger
.\" 2000-12-05 some changes inspired by acahalan's remarks
.\"
.TH fenv 3 2023-07-20 "Linux man-pages 6.05.01"
.SH NAME
feclearexcept, fegetexceptflag, feraiseexcept, fesetexceptflag,
fetestexcept, fegetenv, fegetround, feholdexcept, fesetround,
fesetenv, feupdateenv, feenableexcept, fedisableexcept,
fegetexcept \- floating-point rounding and exception handling
.SH LIBRARY
Math library
.RI ( libm ", " \-lm )
.SH SYNOPSIS
.nf
.B #include <fenv.h>
.PP
.BI "int feclearexcept(int " excepts );
.BI "int fegetexceptflag(fexcept_t *" flagp ", int " excepts );
.BI "int feraiseexcept(int " excepts );
.BI "int fesetexceptflag(const fexcept_t *" flagp ", int " excepts );
.BI "int fetestexcept(int " excepts );
.PP
.B "int fegetround(void);"
.BI "int fesetround(int " rounding_mode );
.PP
.BI "int fegetenv(fenv_t *" envp );
.BI "int feholdexcept(fenv_t *" envp );
.BI "int fesetenv(const fenv_t *" envp );
.BI "int feupdateenv(const fenv_t *" envp );
.fi
.SH DESCRIPTION
These eleven functions were defined in C99, and describe the handling
of floating-point rounding and exceptions (overflow, zero-divide, etc.).
.SS Exceptions
The
.I divide-by-zero
exception occurs when an operation on finite numbers
produces infinity as exact answer.
.PP
The
.I overflow
exception occurs when a result has to be represented as a
floating-point number, but has (much) larger absolute value than the
largest (finite) floating-point number that is representable.
.PP
The
.I underflow
exception occurs when a result has to be represented as a
floating-point number, but has smaller absolute value than the smallest
positive normalized floating-point number (and would lose much accuracy
when represented as a denormalized number).
.PP
The
.I inexact
exception occurs when the rounded result of an operation
is not equal to the infinite precision result.
It may occur whenever
.I overflow
or
.I underflow
occurs.
.PP
The
.I invalid
exception occurs when there is no well-defined result
for an operation, as for 0/0 or infinity \- infinity or sqrt(\-1).
.SS Exception handling
Exceptions are represented in two ways: as a single bit
(exception present/absent), and these bits correspond in some
implementation-defined way with bit positions in an integer,
and also as an opaque structure that may contain more information
about the exception (perhaps the code address where it occurred).
.PP
Each of the macros
.BR FE_DIVBYZERO ,
.BR FE_INEXACT ,
.BR FE_INVALID ,
.BR FE_OVERFLOW ,
.B FE_UNDERFLOW
is defined when the implementation supports handling
of the corresponding exception, and if so then
defines the corresponding bit(s), so that one can call
exception handling functions, for example, using the integer argument
.BR FE_OVERFLOW | FE_UNDERFLOW .
Other exceptions may be supported.
The macro
.B FE_ALL_EXCEPT
is the bitwise OR of all bits corresponding to supported exceptions.
.PP
The
.BR feclearexcept ()
function clears the supported exceptions represented by the bits
in its argument.
.PP
The
.BR fegetexceptflag ()
function stores a representation of the state of the exception flags
represented by the argument
.I excepts
in the opaque object
.IR *flagp .
.PP
The
.BR feraiseexcept ()
function raises the supported exceptions represented by the bits in
.IR excepts .
.PP
The
.BR fesetexceptflag ()
function sets the complete status for the exceptions represented by
.I excepts
to the value
.IR *flagp .
This value must have been obtained by an earlier call of
.BR fegetexceptflag ()
with a last argument that contained all bits in
.IR excepts .
.PP
The
.BR fetestexcept ()
function returns a word in which the bits are set that were
set in the argument
.I excepts
and for which the corresponding exception is currently set.
.SS Rounding mode
The rounding mode determines how the result of floating-point operations
is treated when the result cannot be exactly represented in the significand.
Various rounding modes may be provided:
round to nearest (the default),
round up (toward positive infinity),
round down (toward negative infinity), and
round toward zero.
.PP
Each of the macros
.BR FE_TONEAREST ,
.BR FE_UPWARD ,
.BR FE_DOWNWARD ,
and
.B FE_TOWARDZERO
is defined when the implementation supports getting and setting
the corresponding rounding direction.
.PP
The
.BR fegetround ()
function returns the macro corresponding to the current
rounding mode.
.PP
The
.BR fesetround ()
function sets the rounding mode as specified by its argument
and returns zero when it was successful.
.PP
C99 and POSIX.1-2008 specify an identifier,
.BR FLT_ROUNDS ,
defined in
.IR <float.h> ,
which indicates the implementation-defined rounding
behavior for floating-point addition.
This identifier has one of the following values:
.TP
.B \-1
The rounding mode is not determinable.
.TP
.B 0
Rounding is toward 0.
.TP
.B 1
Rounding is toward nearest number.
.TP
.B 2
Rounding is toward positive infinity.
.TP
.B 3
Rounding is toward negative infinity.
.PP
Other values represent machine-dependent, nonstandard rounding modes.
.PP
The value of
.B FLT_ROUNDS
should reflect the current rounding mode as set by
.BR fesetround ()
(but see BUGS).
.SS Floating-point environment
The entire floating-point environment, including
control modes and status flags, can be handled
as one opaque object, of type
.IR fenv_t .
The default environment is denoted by
.B FE_DFL_ENV
(of type
.IR "const fenv_t\ *" ).
This is the environment setup at program start and it is defined by
ISO C to have round to nearest, all exceptions cleared and a nonstop
(continue on exceptions) mode.
.PP
The
.BR fegetenv ()
function saves the current floating-point environment in the object
.IR *envp .
.PP
The
.BR feholdexcept ()
function does the same, then clears all exception flags,
and sets a nonstop (continue on exceptions) mode,
if available.
It returns zero when successful.
.PP
The
.BR fesetenv ()
function restores the floating-point environment from
the object
.IR *envp .
This object must be known to be valid, for example, the result of a call to
.BR fegetenv ()
or
.BR feholdexcept ()
or equal to
.BR FE_DFL_ENV .
This call does not raise exceptions.
.PP
The
.BR feupdateenv ()
function installs the floating-point environment represented by
the object
.IR *envp ,
except that currently raised exceptions are not cleared.
After calling this function, the raised exceptions will be a bitwise OR
of those previously set with those in
.IR *envp .
As before, the object
.I *envp
must be known to be valid.
.SH RETURN VALUE
These functions return zero on success and nonzero if an error occurred.
.\" Earlier seven of these functions were listed as returning void.
.\" This was corrected in Corrigendum 1 (ISO/IEC 9899:1999/Cor.1:2001(E))
.\" of the C99 Standard.
.SH ATTRIBUTES
For an explanation of the terms used in this section, see
.BR attributes (7).
.nh
.TS
allbox;
lbx lb lb
l l l.
Interface	Attribute	Value
T{
.na
.nh
.BR feclearexcept (),
.BR fegetexceptflag (),
.BR feraiseexcept (),
.BR fesetexceptflag (),
.BR fetestexcept (),
.BR fegetround (),
.BR fesetround (),
.BR fegetenv (),
.BR feholdexcept (),
.BR fesetenv (),
.BR feupdateenv (),
.BR feenableexcept (),
.BR fedisableexcept (),
.BR fegetexcept ()
T}	Thread safety	T{
.na
.nh
MT-Safe
T}
.TE
.sp 1
.hy
.SH STANDARDS
C11, POSIX.1-2008, IEC 60559 (IEC 559:1989), ANSI/IEEE 854.
.SH HISTORY
C99, POSIX.1-2001.
glibc 2.1.
.SH NOTES
.SS glibc notes
If possible, the GNU C Library defines a macro
.B FE_NOMASK_ENV
which represents an environment where every exception raised causes a
trap to occur.
You can test for this macro using
.BR #ifdef .
It is defined only if
.B _GNU_SOURCE
is defined.
The C99 standard does not define a way to set individual bits in the
floating-point mask, for example, to trap on specific flags.
Since glibc 2.2, glibc supports the functions
.BR feenableexcept ()
and
.BR fedisableexcept ()
to set individual floating-point traps, and
.BR fegetexcept ()
to query the state.
.PP
.nf
.BR "#define _GNU_SOURCE" "         /* See feature_test_macros(7) */"
.B "#include <fenv.h>"
.PP
.BI "int feenableexcept(int " excepts );
.BI "int fedisableexcept(int " excepts );
.B "int fegetexcept(void);"
.fi
.PP
The
.BR feenableexcept ()
and
.BR fedisableexcept ()
functions enable (disable) traps for each of the exceptions represented by
.I excepts
and return the previous set of enabled exceptions when successful,
and \-1 otherwise.
The
.BR fegetexcept ()
function returns the set of all currently enabled exceptions.
.SH BUGS
C99 specifies that the value of
.B FLT_ROUNDS
should reflect changes to the current rounding mode, as set by
.BR fesetround ().
Currently,
.\" Aug 08, glibc 2.8
this does not occur:
.B FLT_ROUNDS
always has the value 1.
.\" See http://gcc.gnu.org/ml/gcc/2002-02/msg01535.html
.SH SEE ALSO
.BR math_error (7)