diff options
| author | Danny Smith <dannysmith@users.sourceforge.net> | 2004-07-02 10:01:51 +0000 |
|---|---|---|
| committer | Danny Smith <dannysmith@users.sourceforge.net> | 2004-07-02 10:01:51 +0000 |
| commit | 64b9893f0d1e9c25d75bda78a49dc6e189d35e06 (patch) | |
| tree | 0f0aa6f732725b88219628a08a3fe2f0363ad74a /winsup/mingw/include | |
| parent | 2cb47d686eff408aab237c0a54f9094a7fc1bd76 (diff) | |
| download | newlib-64b9893f0d1e9c25d75bda78a49dc6e189d35e06.zip newlib-64b9893f0d1e9c25d75bda78a49dc6e189d35e06.tar.gz newlib-64b9893f0d1e9c25d75bda78a49dc6e189d35e06.tar.bz2 | |
2 * mingwex/Makefile.in (MATH_DISTFILES): Remove pow.c,
(MATH_OBJS): Remove pow,o.
* include/math.h (pow): Declare with _CRTIMP.
Add comment on how to avoid excess precision problems.
Diffstat (limited to 'winsup/mingw/include')
| -rw-r--r-- | winsup/mingw/include/math.h | 44 |
1 files changed, 39 insertions, 5 deletions
diff --git a/winsup/mingw/include/math.h b/winsup/mingw/include/math.h index fd6ce95..68ba5c7 100644 --- a/winsup/mingw/include/math.h +++ b/winsup/mingw/include/math.h @@ -129,7 +129,6 @@ struct _exception double retval; }; - _CRTIMP double __cdecl sin (double); _CRTIMP double __cdecl cos (double); _CRTIMP double __cdecl tan (double); @@ -143,10 +142,7 @@ _CRTIMP double __cdecl atan2 (double, double); _CRTIMP double __cdecl exp (double); _CRTIMP double __cdecl log (double); _CRTIMP double __cdecl log10 (double); -#ifdef __NO_ISOCEXT -_CRTIMP -#endif - double __cdecl pow (double, double); +_CRTIMP double __cdecl pow (double, double); _CRTIMP double __cdecl sqrt (double); _CRTIMP double __cdecl ceil (double); _CRTIMP double __cdecl floor (double); @@ -156,6 +152,44 @@ _CRTIMP double __cdecl frexp (double, int*); _CRTIMP double __cdecl modf (double, double*); _CRTIMP double __cdecl fmod (double, double); +/* Excess precision when using a 64-bit mantissa for FPU math ops can + cause unexpected results with some of the MSVCRT math functions. For + example, unless the function return value is stored (truncating to + 53-bit mantissa), calls to pow with both x and y as integral values + sometimes produce a non-integral result. + One workaround is to reset the FPU env to 53-bit mantissa + by a call to fesetenv (FE_PC53_ENV). Amother is to force storage + of the return value of individual math functions using wrappers. + NB, using these wrappers will disable builtin math functions and + hence disable the folding of function results at compile time when + arguments are constant. */ + +#if 0 +#define __DEFINE_FLOAT_STORE_MATHFN_D1(fn1) \ +static __inline__ double \ +__float_store_ ## fn1 (double x) \ +{ \ + __volatile__ double res = (fn1) (x); \ + return res; \ +} + +#define __DEFINE_FLOAT_STORE_MATHFN_D2(fn2) \ +static __inline__ double \ +__float_store_ ## fn2 (double x, double y) \ +{ \ + __volatile__ double res = (fn2) (x, y); \ + return res; \ +} +#endif + +/* For example, here is how to force the result of the pow function + to be stored: */ +#if 0 +#undef pow +/* Define the ___float_store_pow function and use it instead of pow(). */ +__DEFINE_FLOAT_STORE_MATHFN_D2 (pow) +#define pow __float_store_pow +#endif #ifndef __STRICT_ANSI__ |