This work is based in part on code from NetBSD libm, libc and kernel. The library is partly public domain and partly BSD-style licensed.
284 lines
8.2 KiB
Groff
284 lines
8.2 KiB
Groff
.\" $NetBSD: fenv.3,v 1.2 2010/08/04 18:58:28 wiz Exp $
|
|
.\" Copyright (c) 2004 David Schultz <das@FreeBSD.org>
|
|
.\" All rights reserved.
|
|
.\"
|
|
.\" Redistribution and use in source and binary forms, with or without
|
|
.\" modification, are permitted provided that the following conditions
|
|
.\" are met:
|
|
.\" 1. Redistributions of source code must retain the above copyright
|
|
.\" notice, this list of conditions and the following disclaimer.
|
|
.\" 2. Redistributions in binary form must reproduce the above copyright
|
|
.\" notice, this list of conditions and the following disclaimer in the
|
|
.\" documentation and/or other materials provided with the distribution.
|
|
.\"
|
|
.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
|
.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
|
.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|
.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|
.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|
.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
.\" SUCH DAMAGE.
|
|
.\"
|
|
.Dd March 16, 2005
|
|
.Dt FENV 3
|
|
.Os
|
|
.Sh NAME
|
|
.Nm feclearexcept ,
|
|
.Nm fegetexceptflag ,
|
|
.Nm feraiseexcept ,
|
|
.Nm fesetexceptflag ,
|
|
.Nm fetestexcept ,
|
|
.Nm fegetround ,
|
|
.Nm fesetround ,
|
|
.Nm fegetenv ,
|
|
.Nm feholdexcept ,
|
|
.Nm fesetenv ,
|
|
.Nm feupdateenv ,
|
|
.Nm feenableexcept ,
|
|
.Nm fedisableexcept ,
|
|
.Nm fegetexcept
|
|
.Nd floating-point environment control
|
|
.Sh LIBRARY
|
|
.Lb libm
|
|
.Sh SYNOPSIS
|
|
.In fenv.h
|
|
.Fd "#pragma STDC FENV_ACCESS ON"
|
|
.Ft int
|
|
.Fn feclearexcept "int excepts"
|
|
.Ft int
|
|
.Fn fegetexceptflag "fexcept_t *flagp" "int excepts"
|
|
.Ft int
|
|
.Fn feraiseexcept "int excepts"
|
|
.Ft int
|
|
.Fn fesetexceptflag "const fexcept_t *flagp" "int excepts"
|
|
.Ft int
|
|
.Fn fetestexcept "int excepts"
|
|
.Ft int
|
|
.Fn fegetround void
|
|
.Ft int
|
|
.Fn fesetround "int round"
|
|
.Ft int
|
|
.Fn fegetenv "fenv_t *envp"
|
|
.Ft int
|
|
.Fn feholdexcept "fenv_t *envp"
|
|
.Ft int
|
|
.Fn fesetenv "const fenv_t *envp"
|
|
.Ft int
|
|
.Fn feupdateenv "const fenv_t *envp"
|
|
.Ft int
|
|
.Fn feenableexcept "int excepts"
|
|
.Ft int
|
|
.Fn fedisableexcept "int excepts"
|
|
.Ft int
|
|
.Fn fegetexcept void
|
|
.Sh DESCRIPTION
|
|
The
|
|
.In fenv.h
|
|
routines manipulate the floating-point environment,
|
|
which includes the exception flags and rounding modes defined in
|
|
.St -ieee754 .
|
|
.Ss Exceptions
|
|
Exception flags are set as side-effects of floating-point arithmetic
|
|
operations and math library routines, and they remain set until
|
|
explicitly cleared.
|
|
The following macros expand to bit flags of type
|
|
.Vt int
|
|
representing the five standard floating-point exceptions.
|
|
.Bl -tag -width ".Dv FE_DIVBYZERO"
|
|
.It Dv FE_DIVBYZERO
|
|
A divide-by-zero exception occurs when the program attempts to
|
|
divide a finite non-zero number by zero.
|
|
.It Dv FE_INEXACT
|
|
An inexact exception is raised whenever there is a loss of precision
|
|
due to rounding.
|
|
.It Dv FE_INVALID
|
|
Invalid operation exceptions occur when a program attempts to
|
|
perform calculations for which there is no reasonable representable
|
|
answer.
|
|
For instance, subtraction of infinities, division of zero by zero,
|
|
ordered comparison involving \*(Nas, and taking the square root of a
|
|
negative number are all invalid operations.
|
|
.It Dv FE_OVERFLOW
|
|
An overflow exception occurs when the magnitude of the result of a
|
|
computation is too large to fit in the destination type.
|
|
.It Dv FE_UNDERFLOW
|
|
Underflow occurs when the result of a computation is too close to zero
|
|
to be represented as a non-zero value in the destination type.
|
|
.El
|
|
.Pp
|
|
Additionally, the
|
|
.Dv FE_ALL_EXCEPT
|
|
macro expands to the bitwise OR of the above flags and any
|
|
architecture-specific flags.
|
|
Combinations of these flags are passed to the
|
|
.Fn feclearexcept ,
|
|
.Fn fegetexceptflag ,
|
|
.Fn feraiseexcept ,
|
|
.Fn fesetexceptflag ,
|
|
and
|
|
.Fn fetestexcept
|
|
functions to clear, save, raise, restore, and examine the
|
|
processor's floating-point exception flags, respectively.
|
|
.Pp
|
|
Exceptions may be
|
|
.Em unmasked
|
|
with
|
|
.Fn feenableexcept
|
|
and masked with
|
|
.Fn fedisableexcept .
|
|
Unmasked exceptions cause a trap when they are produced, and
|
|
all exceptions are masked by default.
|
|
The current mask can be tested with
|
|
.Fn fegetexcept .
|
|
.Ss Rounding Modes
|
|
.St -ieee754
|
|
specifies four rounding modes.
|
|
These modes control the direction in which results are rounded
|
|
from their exact values in order to fit them into binary
|
|
floating-point variables.
|
|
The four modes correspond with the following symbolic constants.
|
|
.Bl -tag -width ".Dv FE_TOWARDZERO"
|
|
.It Dv FE_TONEAREST
|
|
Results are rounded to the closest representable value.
|
|
If the exact result is exactly half way between two representable
|
|
values, the value whose last binary digit is even (zero) is chosen.
|
|
This is the default mode.
|
|
.It Dv FE_DOWNWARD
|
|
Results are rounded towards negative \*[If].
|
|
.It Dv FE_UPWARD
|
|
Results are rounded towards positive \*[If].
|
|
.It Dv FE_TOWARDZERO
|
|
Results are rounded towards zero.
|
|
.El
|
|
.Pp
|
|
The
|
|
.Fn fegetround
|
|
and
|
|
.Fn fesetround
|
|
functions query and set the rounding mode.
|
|
.Ss Environment Control
|
|
The
|
|
.Fn fegetenv
|
|
and
|
|
.Fn fesetenv
|
|
functions save and restore the floating-point environment,
|
|
which includes exception flags, the current exception mask,
|
|
the rounding mode, and possibly other implementation-specific
|
|
state.
|
|
The
|
|
.Fn feholdexcept
|
|
function behaves like
|
|
.Fn fegetenv ,
|
|
but with the additional effect of clearing the exception flags and
|
|
installing a
|
|
.Em non-stop
|
|
mode.
|
|
In non-stop mode, floating-point operations will set exception flags
|
|
as usual, but no
|
|
.Dv SIGFPE
|
|
signals will be generated as a result.
|
|
Non-stop mode is the default, but it may be altered by
|
|
non-standard mechanisms.
|
|
.\" XXX Mention fe[gs]etmask() here after the interface is finalized
|
|
.\" XXX and ready to be officially documented.
|
|
The
|
|
.Fn feupdateenv
|
|
function restores a saved environment similarly to
|
|
.Fn fesetenv ,
|
|
but it also re-raises any floating-point exceptions from the old
|
|
environment.
|
|
.Pp
|
|
The macro
|
|
.Dv FE_DFL_ENV
|
|
expands to a pointer to the default environment.
|
|
.Sh EXAMPLES
|
|
The following routine computes the square root function.
|
|
It explicitly raises an invalid exception on appropriate inputs using
|
|
.Fn feraiseexcept .
|
|
It also defers inexact exceptions while it computes intermediate
|
|
values, and then it allows an inexact exception to be raised only if
|
|
the final answer is inexact.
|
|
.Bd -literal -offset indent
|
|
#pragma STDC FENV_ACCESS ON
|
|
double sqrt(double n) {
|
|
double x = 1.0;
|
|
fenv_t env;
|
|
|
|
if (isnan(n) || n \*[Lt] 0.0) {
|
|
feraiseexcept(FE_INVALID);
|
|
return (NAN);
|
|
}
|
|
if (isinf(n) || n == 0.0)
|
|
return (n);
|
|
feholdexcept(\*[Am]env);
|
|
while (fabs((x * x) - n) \*[Gt] DBL_EPSILON * 2 * x)
|
|
x = (x / 2) + (n / (2 * x));
|
|
if (x * x == n)
|
|
feclearexcept(FE_INEXACT);
|
|
feupdateenv(\*[Am]env);
|
|
return (x);
|
|
}
|
|
.Ed
|
|
.Sh SEE ALSO
|
|
.Xr c99 1 ,
|
|
.Xr feclearexcept 3 ,
|
|
.Xr fedisableexcept 3 ,
|
|
.Xr feenableexcept 3 ,
|
|
.Xr fegetenv 3 ,
|
|
.Xr fegetexcept 3 ,
|
|
.Xr fegetexceptflag 3 ,
|
|
.Xr fegetround 3 ,
|
|
.Xr feholdexcept 3 ,
|
|
.Xr feraiseexcept 3 ,
|
|
.Xr fesetenv 3 ,
|
|
.Xr fesetexceptflag 3 ,
|
|
.Xr fesetround 3 ,
|
|
.Xr fetestexcept 3 ,
|
|
.Xr feupdateenv 3
|
|
.\"Xr fpgetprec 3 ,
|
|
.\"Xr fpsetprec 3
|
|
.Sh STANDARDS
|
|
Except as noted below,
|
|
.In fenv.h
|
|
conforms to
|
|
.St -isoC-99 .
|
|
The
|
|
.Fn feenableexcept ,
|
|
.Fn fedisableexcept ,
|
|
and
|
|
.Fn fegetexcept
|
|
routines are extensions.
|
|
.Sh HISTORY
|
|
The
|
|
.In fenv.h
|
|
header first appeared in
|
|
.Fx 5.3
|
|
and
|
|
.Nx 6.0 .
|
|
It supersedes the non-standard routines defined in
|
|
.In ieeefp.h
|
|
and documented in
|
|
.Xr fpgetround 3 .
|
|
.Sh CAVEATS
|
|
The FENV_ACCESS pragma can be enabled with
|
|
.Dl "#pragma STDC FENV_ACCESS ON"
|
|
and disabled with the
|
|
.Dl "#pragma STDC FENV_ACCESS OFF"
|
|
directive.
|
|
This lexically-scoped annotation tells the compiler that the program
|
|
may access the floating-point environment, so optimizations that would
|
|
violate strict IEEE-754 semantics are disabled.
|
|
If execution reaches a block of code for which
|
|
.Dv FENV_ACCESS
|
|
is off, the floating-point environment will become undefined.
|
|
.Sh BUGS
|
|
The
|
|
.Dv FENV_ACCESS
|
|
pragma is unimplemented in the system compiler.
|
|
However, non-constant expressions generally produce the correct
|
|
side-effects at low optimization levels.
|