f4a5643005
Symptoms of the problem included assembler warnings and nondeterministic runtime behavior when a fe*() call that affects the fpsr is closely followed by a float point op. The bug (at least, I think it's a bug) is that gcc does not insert a break between a volatile asm and a dependent instruction if the volatile asm came from an inlined function. Volatile asms seem to be fine in other circumstances, even without -mvolatile-asm-stop, so perhaps the compiler adds the stop bits before inlining takes place. The problem does not occur at -O0 because inlining is disabled, and it doesn't happen at -O2 because -fschedule-insns2 knows better.
232 lines
5.1 KiB
C
232 lines
5.1 KiB
C
/*-
|
|
* 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.
|
|
*
|
|
* $FreeBSD$
|
|
*/
|
|
|
|
#ifndef _FENV_H_
|
|
#define _FENV_H_
|
|
|
|
#include <sys/_types.h>
|
|
|
|
typedef __uint64_t fenv_t;
|
|
typedef __uint16_t fexcept_t;
|
|
|
|
/* Exception flags */
|
|
#define FE_INVALID 0x01
|
|
#define FE_DENORMAL 0x02
|
|
#define FE_DIVBYZERO 0x04
|
|
#define FE_OVERFLOW 0x08
|
|
#define FE_UNDERFLOW 0x10
|
|
#define FE_INEXACT 0x20
|
|
#define FE_ALL_EXCEPT (FE_DIVBYZERO | FE_DENORMAL | FE_INEXACT | \
|
|
FE_INVALID | FE_OVERFLOW | FE_UNDERFLOW)
|
|
|
|
/* Rounding modes */
|
|
#define FE_TONEAREST 0x0000
|
|
#define FE_DOWNWARD 0x0400
|
|
#define FE_UPWARD 0x0800
|
|
#define FE_TOWARDZERO 0x0c00
|
|
#define _ROUND_MASK (FE_TONEAREST | FE_DOWNWARD | \
|
|
FE_UPWARD | FE_TOWARDZERO)
|
|
|
|
__BEGIN_DECLS
|
|
|
|
/* Default floating-point environment */
|
|
extern const fenv_t __fe_dfl_env;
|
|
#define FE_DFL_ENV (&__fe_dfl_env)
|
|
|
|
#define _FPUSW_SHIFT 13
|
|
|
|
#define __stfpsr(__r) __asm __volatile("mov %0=ar.fpsr" : "=r" (*(__r)))
|
|
#define __ldfpsr(__r) __asm __volatile("mov ar.fpsr=%0;;" : : "r" (__r))
|
|
|
|
static __inline int
|
|
feclearexcept(int __excepts)
|
|
{
|
|
fenv_t __fpsr;
|
|
|
|
__stfpsr(&__fpsr);
|
|
__fpsr &= ~((fenv_t)__excepts << _FPUSW_SHIFT);
|
|
__ldfpsr(__fpsr);
|
|
return (0);
|
|
}
|
|
|
|
static __inline int
|
|
fegetexceptflag(fexcept_t *__flagp, int __excepts)
|
|
{
|
|
fenv_t __fpsr;
|
|
|
|
__stfpsr(&__fpsr);
|
|
*__flagp = (fexcept_t)(__fpsr >> _FPUSW_SHIFT) & __excepts;
|
|
return (0);
|
|
}
|
|
|
|
static __inline int
|
|
fesetexceptflag(const fexcept_t *__flagp, int __excepts)
|
|
{
|
|
fenv_t __fpsr;
|
|
|
|
__stfpsr(&__fpsr);
|
|
__fpsr &= ~((fenv_t)__excepts << _FPUSW_SHIFT);
|
|
__fpsr |= (fenv_t)(__excepts & *__flagp) << _FPUSW_SHIFT;
|
|
__ldfpsr(__fpsr);
|
|
return (0);
|
|
}
|
|
|
|
/*
|
|
* It is worthwhile to use the inline version of this function iff it
|
|
* is called with arguments that are compile-time constants (due to
|
|
* dead code elimination). Unfortunately, gcc isn't smart enough to
|
|
* figure this out automatically, and there's no way to tell it.
|
|
* We assume that constant arguments will be the common case.
|
|
*/
|
|
static __inline int
|
|
feraiseexcept(int __excepts)
|
|
{
|
|
volatile double d;
|
|
|
|
/*
|
|
* With a compiler that supports the FENV_ACCESS pragma
|
|
* properly, simple expressions like '0.0 / 0.0' should
|
|
* be sufficient to generate traps. Unfortunately, we
|
|
* need to bring a volatile variable into the equation
|
|
* to prevent incorrect optimizations.
|
|
*/
|
|
if (__excepts & FE_INVALID) {
|
|
d = 0.0;
|
|
d = 0.0 / d;
|
|
}
|
|
if (__excepts & FE_DIVBYZERO) {
|
|
d = 0.0;
|
|
d = 1.0 / d;
|
|
}
|
|
if (__excepts & FE_OVERFLOW) {
|
|
d = 0x1.ffp1023;
|
|
d *= 2.0;
|
|
}
|
|
if (__excepts & FE_UNDERFLOW) {
|
|
d = 0x1p-1022;
|
|
d /= 0x1p1023;
|
|
}
|
|
if (__excepts & FE_INEXACT) {
|
|
d = 0x1p-1022;
|
|
d += 1.0;
|
|
}
|
|
return (0);
|
|
}
|
|
|
|
static __inline int
|
|
fetestexcept(int __excepts)
|
|
{
|
|
fenv_t __fpsr;
|
|
|
|
__stfpsr(&__fpsr);
|
|
return ((__fpsr >> _FPUSW_SHIFT) & __excepts);
|
|
}
|
|
|
|
|
|
static __inline int
|
|
fegetround(void)
|
|
{
|
|
fenv_t __fpsr;
|
|
|
|
__stfpsr(&__fpsr);
|
|
return (__fpsr & _ROUND_MASK);
|
|
}
|
|
|
|
static __inline int
|
|
fesetround(int __round)
|
|
{
|
|
fenv_t __fpsr;
|
|
|
|
if (__round & ~_ROUND_MASK)
|
|
return (-1);
|
|
__stfpsr(&__fpsr);
|
|
__fpsr &= ~_ROUND_MASK;
|
|
__fpsr |= __round;
|
|
__ldfpsr(__fpsr);
|
|
return (0);
|
|
}
|
|
|
|
static __inline int
|
|
fegetenv(fenv_t *__envp)
|
|
{
|
|
|
|
__stfpsr(__envp);
|
|
return (0);
|
|
}
|
|
|
|
static __inline int
|
|
feholdexcept(fenv_t *__envp)
|
|
{
|
|
fenv_t __fpsr;
|
|
|
|
__stfpsr(&__fpsr);
|
|
*__envp = __fpsr;
|
|
__fpsr &= ~((fenv_t)FE_ALL_EXCEPT << _FPUSW_SHIFT);
|
|
__fpsr |= FE_ALL_EXCEPT;
|
|
__ldfpsr(__fpsr);
|
|
return (0);
|
|
}
|
|
|
|
static __inline int
|
|
fesetenv(const fenv_t *__envp)
|
|
{
|
|
|
|
__ldfpsr(*__envp);
|
|
return (0);
|
|
}
|
|
|
|
int feupdateenv(const fenv_t *__envp);
|
|
|
|
#if __BSD_VISIBLE
|
|
|
|
static __inline int
|
|
fesetmask(int __mask)
|
|
{
|
|
fenv_t __newfpsr, __oldfpsr;
|
|
|
|
__stfpsr(&__oldfpsr);
|
|
__newfpsr = (__oldfpsr | FE_ALL_EXCEPT) & ~__mask;
|
|
__ldfpsr(__newfpsr);
|
|
return (~__oldfpsr & FE_ALL_EXCEPT);
|
|
}
|
|
|
|
static __inline int
|
|
fegetmask(void)
|
|
{
|
|
fenv_t __fpsr;
|
|
|
|
__stfpsr(&__fpsr);
|
|
return (~__fpsr & FE_ALL_EXCEPT);
|
|
}
|
|
|
|
#endif /* __BSD_VISIBLE */
|
|
|
|
__END_DECLS
|
|
|
|
#endif /* !_FENV_H_ */
|