/* * ==================================================== * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved. * * Developed at SunPro, a Sun Microsystems, Inc. business. * Permission to use, copy, modify, and distribute this * software is freely granted, provided that this notice * is preserved. * ==================================================== */ /* * from: @(#)fdlibm.h 5.1 93/09/24 * $FreeBSD$ */ #ifndef _MATH_PRIVATE_H_ #define _MATH_PRIVATE_H_ #include #include /* * The original fdlibm code used statements like: * n0 = ((*(int*)&one)>>29)^1; * index of high word * * ix0 = *(n0+(int*)&x); * high word of x * * ix1 = *((1-n0)+(int*)&x); * low word of x * * to dig two 32 bit words out of the 64 bit IEEE floating point * value. That is non-ANSI, and, moreover, the gcc instruction * scheduler gets it wrong. We instead use the following macros. * Unlike the original code, we determine the endianness at compile * time, not at run time; I don't see much benefit to selecting * endianness at run time. */ /* * A union which permits us to convert between a double and two 32 bit * ints. */ #ifdef __arm__ #if defined(__VFP_FP__) || defined(__ARM_EABI__) #define IEEE_WORD_ORDER BYTE_ORDER #else #define IEEE_WORD_ORDER BIG_ENDIAN #endif #else /* __arm__ */ #define IEEE_WORD_ORDER BYTE_ORDER #endif /* A union which permits us to convert between a long double and four 32 bit ints. */ #if IEEE_WORD_ORDER == BIG_ENDIAN typedef union { long double value; struct { u_int32_t mswhi; u_int32_t mswlo; u_int32_t lswhi; u_int32_t lswlo; } parts32; struct { u_int64_t msw; u_int64_t lsw; } parts64; } ieee_quad_shape_type; #endif #if IEEE_WORD_ORDER == LITTLE_ENDIAN typedef union { long double value; struct { u_int32_t lswlo; u_int32_t lswhi; u_int32_t mswlo; u_int32_t mswhi; } parts32; struct { u_int64_t lsw; u_int64_t msw; } parts64; } ieee_quad_shape_type; #endif #if IEEE_WORD_ORDER == BIG_ENDIAN typedef union { double value; struct { u_int32_t msw; u_int32_t lsw; } parts; struct { u_int64_t w; } xparts; } ieee_double_shape_type; #endif #if IEEE_WORD_ORDER == LITTLE_ENDIAN typedef union { double value; struct { u_int32_t lsw; u_int32_t msw; } parts; struct { u_int64_t w; } xparts; } ieee_double_shape_type; #endif /* Get two 32 bit ints from a double. */ #define EXTRACT_WORDS(ix0,ix1,d) \ do { \ ieee_double_shape_type ew_u; \ ew_u.value = (d); \ (ix0) = ew_u.parts.msw; \ (ix1) = ew_u.parts.lsw; \ } while (0) /* Get a 64-bit int from a double. */ #define EXTRACT_WORD64(ix,d) \ do { \ ieee_double_shape_type ew_u; \ ew_u.value = (d); \ (ix) = ew_u.xparts.w; \ } while (0) /* Get the more significant 32 bit int from a double. */ #define GET_HIGH_WORD(i,d) \ do { \ ieee_double_shape_type gh_u; \ gh_u.value = (d); \ (i) = gh_u.parts.msw; \ } while (0) /* Get the less significant 32 bit int from a double. */ #define GET_LOW_WORD(i,d) \ do { \ ieee_double_shape_type gl_u; \ gl_u.value = (d); \ (i) = gl_u.parts.lsw; \ } while (0) /* Set a double from two 32 bit ints. */ #define INSERT_WORDS(d,ix0,ix1) \ do { \ ieee_double_shape_type iw_u; \ iw_u.parts.msw = (ix0); \ iw_u.parts.lsw = (ix1); \ (d) = iw_u.value; \ } while (0) /* Set a double from a 64-bit int. */ #define INSERT_WORD64(d,ix) \ do { \ ieee_double_shape_type iw_u; \ iw_u.xparts.w = (ix); \ (d) = iw_u.value; \ } while (0) /* Set the more significant 32 bits of a double from an int. */ #define SET_HIGH_WORD(d,v) \ do { \ ieee_double_shape_type sh_u; \ sh_u.value = (d); \ sh_u.parts.msw = (v); \ (d) = sh_u.value; \ } while (0) /* Set the less significant 32 bits of a double from an int. */ #define SET_LOW_WORD(d,v) \ do { \ ieee_double_shape_type sl_u; \ sl_u.value = (d); \ sl_u.parts.lsw = (v); \ (d) = sl_u.value; \ } while (0) /* * A union which permits us to convert between a float and a 32 bit * int. */ typedef union { float value; /* FIXME: Assumes 32 bit int. */ unsigned int word; } ieee_float_shape_type; /* Get a 32 bit int from a float. */ #define GET_FLOAT_WORD(i,d) \ do { \ ieee_float_shape_type gf_u; \ gf_u.value = (d); \ (i) = gf_u.word; \ } while (0) /* Set a float from a 32 bit int. */ #define SET_FLOAT_WORD(d,i) \ do { \ ieee_float_shape_type sf_u; \ sf_u.word = (i); \ (d) = sf_u.value; \ } while (0) /* * Get expsign and mantissa as 16 bit and 64 bit ints from an 80 bit long * double. */ #define EXTRACT_LDBL80_WORDS(ix0,ix1,d) \ do { \ union IEEEl2bits ew_u; \ ew_u.e = (d); \ (ix0) = ew_u.xbits.expsign; \ (ix1) = ew_u.xbits.man; \ } while (0) /* * Get expsign and mantissa as one 16 bit and two 64 bit ints from a 128 bit * long double. */ #define EXTRACT_LDBL128_WORDS(ix0,ix1,ix2,d) \ do { \ union IEEEl2bits ew_u; \ ew_u.e = (d); \ (ix0) = ew_u.xbits.expsign; \ (ix1) = ew_u.xbits.manh; \ (ix2) = ew_u.xbits.manl; \ } while (0) /* Get expsign as a 16 bit int from a long double. */ #define GET_LDBL_EXPSIGN(i,d) \ do { \ union IEEEl2bits ge_u; \ ge_u.e = (d); \ (i) = ge_u.xbits.expsign; \ } while (0) /* * Set an 80 bit long double from a 16 bit int expsign and a 64 bit int * mantissa. */ #define INSERT_LDBL80_WORDS(d,ix0,ix1) \ do { \ union IEEEl2bits iw_u; \ iw_u.xbits.expsign = (ix0); \ iw_u.xbits.man = (ix1); \ (d) = iw_u.e; \ } while (0) /* * Set a 128 bit long double from a 16 bit int expsign and two 64 bit ints * comprising the mantissa. */ #define INSERT_LDBL128_WORDS(d,ix0,ix1,ix2) \ do { \ union IEEEl2bits iw_u; \ iw_u.xbits.expsign = (ix0); \ iw_u.xbits.manh = (ix1); \ iw_u.xbits.manl = (ix2); \ (d) = iw_u.e; \ } while (0) /* Set expsign of a long double from a 16 bit int. */ #define SET_LDBL_EXPSIGN(d,v) \ do { \ union IEEEl2bits se_u; \ se_u.e = (d); \ se_u.xbits.expsign = (v); \ (d) = se_u.e; \ } while (0) #ifdef __i386__ /* Long double constants are broken on i386. */ #define LD80C(m, ex, v) { \ .xbits.man = __CONCAT(m, ULL), \ .xbits.expsign = (0x3fff + (ex)) | ((v) < 0 ? 0x8000 : 0), \ } #else /* The above works on non-i386 too, but we use this to check v. */ #define LD80C(m, ex, v) { .e = (v), } #endif #ifdef FLT_EVAL_METHOD /* * Attempt to get strict C99 semantics for assignment with non-C99 compilers. */ #if FLT_EVAL_METHOD == 0 || __GNUC__ == 0 #define STRICT_ASSIGN(type, lval, rval) ((lval) = (rval)) #else #define STRICT_ASSIGN(type, lval, rval) do { \ volatile type __lval; \ \ if (sizeof(type) >= sizeof(long double)) \ (lval) = (rval); \ else { \ __lval = (rval); \ (lval) = __lval; \ } \ } while (0) #endif #endif /* FLT_EVAL_METHOD */ /* Support switching the mode to FP_PE if necessary. */ #if defined(__i386__) && !defined(NO_FPSETPREC) #define ENTERI() ENTERIT(long double) #define ENTERIT(returntype) \ returntype __retval; \ fp_prec_t __oprec; \ \ if ((__oprec = fpgetprec()) != FP_PE) \ fpsetprec(FP_PE) #define RETURNI(x) do { \ __retval = (x); \ if (__oprec != FP_PE) \ fpsetprec(__oprec); \ RETURNF(__retval); \ } while (0) #define ENTERV() \ fp_prec_t __oprec; \ \ if ((__oprec = fpgetprec()) != FP_PE) \ fpsetprec(FP_PE) #define RETURNV() do { \ if (__oprec != FP_PE) \ fpsetprec(__oprec); \ return; \ } while (0) #else #define ENTERI() #define ENTERIT(x) #define RETURNI(x) RETURNF(x) #define ENTERV() #define RETURNV() return #endif /* Default return statement if hack*_t() is not used. */ #define RETURNF(v) return (v) /* * 2sum gives the same result as 2sumF without requiring |a| >= |b| or * a == 0, but is slower. */ #define _2sum(a, b) do { \ __typeof(a) __s, __w; \ \ __w = (a) + (b); \ __s = __w - (a); \ (b) = ((a) - (__w - __s)) + ((b) - __s); \ (a) = __w; \ } while (0) /* * 2sumF algorithm. * * "Normalize" the terms in the infinite-precision expression a + b for * the sum of 2 floating point values so that b is as small as possible * relative to 'a'. (The resulting 'a' is the value of the expression in * the same precision as 'a' and the resulting b is the rounding error.) * |a| must be >= |b| or 0, b's type must be no larger than 'a's type, and * exponent overflow or underflow must not occur. This uses a Theorem of * Dekker (1971). See Knuth (1981) 4.2.2 Theorem C. The name "TwoSum" * is apparently due to Skewchuk (1997). * * For this to always work, assignment of a + b to 'a' must not retain any * extra precision in a + b. This is required by C standards but broken * in many compilers. The brokenness cannot be worked around using * STRICT_ASSIGN() like we do elsewhere, since the efficiency of this * algorithm would be destroyed by non-null strict assignments. (The * compilers are correct to be broken -- the efficiency of all floating * point code calculations would be destroyed similarly if they forced the * conversions.) * * Fortunately, a case that works well can usually be arranged by building * any extra precision into the type of 'a' -- 'a' should have type float_t, * double_t or long double. b's type should be no larger than 'a's type. * Callers should use these types with scopes as large as possible, to * reduce their own extra-precision and efficiciency problems. In * particular, they shouldn't convert back and forth just to call here. */ #ifdef DEBUG #define _2sumF(a, b) do { \ __typeof(a) __w; \ volatile __typeof(a) __ia, __ib, __r, __vw; \ \ __ia = (a); \ __ib = (b); \ assert(__ia == 0 || fabsl(__ia) >= fabsl(__ib)); \ \ __w = (a) + (b); \ (b) = ((a) - __w) + (b); \ (a) = __w; \ \ /* The next 2 assertions are weak if (a) is already long double. */ \ assert((long double)__ia + __ib == (long double)(a) + (b)); \ __vw = __ia + __ib; \ __r = __ia - __vw; \ __r += __ib; \ assert(__vw == (a) && __r == (b)); \ } while (0) #else /* !DEBUG */ #define _2sumF(a, b) do { \ __typeof(a) __w; \ \ __w = (a) + (b); \ (b) = ((a) - __w) + (b); \ (a) = __w; \ } while (0) #endif /* DEBUG */ /* * Set x += c, where x is represented in extra precision as a + b. * x must be sufficiently normalized and sufficiently larger than c, * and the result is then sufficiently normalized. * * The details of ordering are that |a| must be >= |c| (so that (a, c) * can be normalized without extra work to swap 'a' with c). The details of * the normalization are that b must be small relative to the normalized 'a'. * Normalization of (a, c) makes the normalized c tiny relative to the * normalized a, so b remains small relative to 'a' in the result. However, * b need not ever be tiny relative to 'a'. For example, b might be about * 2**20 times smaller than 'a' to give about 20 extra bits of precision. * That is usually enough, and adding c (which by normalization is about * 2**53 times smaller than a) cannot change b significantly. However, * cancellation of 'a' with c in normalization of (a, c) may reduce 'a' * significantly relative to b. The caller must ensure that significant * cancellation doesn't occur, either by having c of the same sign as 'a', * or by having |c| a few percent smaller than |a|. Pre-normalization of * (a, b) may help. * * This is a variant of an algorithm of Kahan (see Knuth (1981) 4.2.2 * exercise 19). We gain considerable efficiency by requiring the terms to * be sufficiently normalized and sufficiently increasing. */ #define _3sumF(a, b, c) do { \ __typeof(a) __tmp; \ \ __tmp = (c); \ _2sumF(__tmp, (a)); \ (b) += (a); \ (a) = __tmp; \ } while (0) /* * Common routine to process the arguments to nan(), nanf(), and nanl(). */ void _scan_nan(uint32_t *__words, int __num_words, const char *__s); /* * Mix 0, 1 or 2 NaNs. First add 0 to each arg. This normally just turns * signaling NaNs into quiet NaNs by setting a quiet bit. We do this * because we want to never return a signaling NaN, and also because we * don't want the quiet bit to affect the result. Then mix the converted * args using the specified operation. * * When one arg is NaN, the result is typically that arg quieted. When both * args are NaNs, the result is typically the quietening of the arg whose * mantissa is largest after quietening. When neither arg is NaN, the * result may be NaN because it is indeterminate, or finite for subsequent * construction of a NaN as the indeterminate 0.0L/0.0L. * * Technical complications: the result in bits after rounding to the final * precision might depend on the runtime precision and/or on compiler * optimizations, especially when different register sets are used for * different precisions. Try to make the result not depend on at least the * runtime precision by always doing the main mixing step in long double * precision. Try to reduce dependencies on optimizations by adding the * the 0's in different precisions (unless everything is in long double * precision). */ #define nan_mix(x, y) (nan_mix_op((x), (y), +)) #define nan_mix_op(x, y, op) (((x) + 0.0L) op ((y) + 0)) #ifdef _COMPLEX_H /* * C99 specifies that complex numbers have the same representation as * an array of two elements, where the first element is the real part * and the second element is the imaginary part. */ typedef union { float complex f; float a[2]; } float_complex; typedef union { double complex f; double a[2]; } double_complex; typedef union { long double complex f; long double a[2]; } long_double_complex; #define REALPART(z) ((z).a[0]) #define IMAGPART(z) ((z).a[1]) /* * Inline functions that can be used to construct complex values. * * The C99 standard intends x+I*y to be used for this, but x+I*y is * currently unusable in general since gcc introduces many overflow, * underflow, sign and efficiency bugs by rewriting I*y as * (0.0+I)*(y+0.0*I) and laboriously computing the full complex product. * In particular, I*Inf is corrupted to NaN+I*Inf, and I*-0 is corrupted * to -0.0+I*0.0. * * The C11 standard introduced the macros CMPLX(), CMPLXF() and CMPLXL() * to construct complex values. Compilers that conform to the C99 * standard require the following functions to avoid the above issues. */ #ifndef CMPLXF static __inline float complex CMPLXF(float x, float y) { float_complex z; REALPART(z) = x; IMAGPART(z) = y; return (z.f); } #endif #ifndef CMPLX static __inline double complex CMPLX(double x, double y) { double_complex z; REALPART(z) = x; IMAGPART(z) = y; return (z.f); } #endif #ifndef CMPLXL static __inline long double complex CMPLXL(long double x, long double y) { long_double_complex z; REALPART(z) = x; IMAGPART(z) = y; return (z.f); } #endif #endif /* _COMPLEX_H */ /* * The rnint() family rounds to the nearest integer for a restricted range * range of args (up to about 2**MANT_DIG). We assume that the current * rounding mode is FE_TONEAREST so that this can be done efficiently. * Extra precision causes more problems in practice, and we only centralize * this here to reduce those problems, and have not solved the efficiency * problems. The exp2() family uses a more delicate version of this that * requires extracting bits from the intermediate value, so it is not * centralized here and should copy any solution of the efficiency problems. */ static inline double rnint(__double_t x) { /* * This casts to double to kill any extra precision. This depends * on the cast being applied to a double_t to avoid compiler bugs * (this is a cleaner version of STRICT_ASSIGN()). This is * inefficient if there actually is extra precision, but is hard * to improve on. We use double_t in the API to minimise conversions * for just calling here. Note that we cannot easily change the * magic number to the one that works directly with double_t, since * the rounding precision is variable at runtime on x86 so the * magic number would need to be variable. Assuming that the * rounding precision is always the default is too fragile. This * and many other complications will move when the default is * changed to FP_PE. */ return ((double)(x + 0x1.8p52) - 0x1.8p52); } static inline float rnintf(__float_t x) { /* * As for rnint(), except we could just call that to handle the * extra precision case, usually without losing efficiency. */ return ((float)(x + 0x1.8p23F) - 0x1.8p23F); } #ifdef LDBL_MANT_DIG /* * The complications for extra precision are smaller for rnintl() since it * can safely assume that the rounding precision has been increased from * its default to FP_PE on x86. We don't exploit that here to get small * optimizations from limiting the range to double. We just need it for * the magic number to work with long doubles. ld128 callers should use * rnint() instead of this if possible. ld80 callers should prefer * rnintl() since for amd64 this avoids swapping the register set, while * for i386 it makes no difference (assuming FP_PE), and for other arches * it makes little difference. */ static inline long double rnintl(long double x) { return (x + __CONCAT(0x1.8p, LDBL_MANT_DIG) / 2 - __CONCAT(0x1.8p, LDBL_MANT_DIG) / 2); } #endif /* LDBL_MANT_DIG */ /* * irint() and i64rint() give the same result as casting to their integer * return type provided their arg is a floating point integer. They can * sometimes be more efficient because no rounding is required. */ #if defined(amd64) || defined(__i386__) #define irint(x) \ (sizeof(x) == sizeof(float) && \ sizeof(__float_t) == sizeof(long double) ? irintf(x) : \ sizeof(x) == sizeof(double) && \ sizeof(__double_t) == sizeof(long double) ? irintd(x) : \ sizeof(x) == sizeof(long double) ? irintl(x) : (int)(x)) #else #define irint(x) ((int)(x)) #endif #define i64rint(x) ((int64_t)(x)) /* only needed for ld128 so not opt. */ #if defined(__i386__) static __inline int irintf(float x) { int n; __asm("fistl %0" : "=m" (n) : "t" (x)); return (n); } static __inline int irintd(double x) { int n; __asm("fistl %0" : "=m" (n) : "t" (x)); return (n); } #endif #if defined(__amd64__) || defined(__i386__) static __inline int irintl(long double x) { int n; __asm("fistl %0" : "=m" (n) : "t" (x)); return (n); } #endif /* * The following are fast floor macros for 0 <= |x| < 0x1p(N-1), where * N is the precision of the type of x. These macros are used in the * half-cycle trignometric functions (e.g., sinpi(x)). */ #define FFLOORF(x, j0, ix) do { \ (j0) = (((ix) >> 23) & 0xff) - 0x7f; \ (ix) &= ~(0x007fffff >> (j0)); \ SET_FLOAT_WORD((x), (ix)); \ } while (0) #define FFLOOR(x, j0, ix, lx) do { \ (j0) = (((ix) >> 20) & 0x7ff) - 0x3ff; \ if ((j0) < 20) { \ (ix) &= ~(0x000fffff >> (j0)); \ (lx) = 0; \ } else { \ (lx) &= ~((uint32_t)0xffffffff >> ((j0) - 20)); \ } \ INSERT_WORDS((x), (ix), (lx)); \ } while (0) #define FFLOORL80(x, j0, ix, lx) do { \ j0 = ix - 0x3fff + 1; \ if ((j0) < 32) { \ (lx) = ((lx) >> 32) << 32; \ (lx) &= ~((((lx) << 32)-1) >> (j0)); \ } else { \ uint64_t _m; \ _m = (uint64_t)-1 >> (j0); \ if ((lx) & _m) (lx) &= ~_m; \ } \ INSERT_LDBL80_WORDS((x), (ix), (lx)); \ } while (0) #define FFLOORL128(x, ai, ar) do { \ union IEEEl2bits u; \ uint64_t m; \ int e; \ u.e = (x); \ e = u.bits.exp - 16383; \ if (e < 48) { \ m = ((1llu << 49) - 1) >> (e + 1); \ u.bits.manh &= ~m; \ u.bits.manl = 0; \ } else { \ m = (uint64_t)-1 >> (e - 48); \ u.bits.manl &= ~m; \ } \ (ai) = u.e; \ (ar) = (x) - (ai); \ } while (0) #ifdef DEBUG #if defined(__amd64__) || defined(__i386__) #define breakpoint() asm("int $3") #else #include #define breakpoint() raise(SIGTRAP) #endif #endif #ifdef STRUCT_RETURN #define RETURNSP(rp) do { \ if (!(rp)->lo_set) \ RETURNF((rp)->hi); \ RETURNF((rp)->hi + (rp)->lo); \ } while (0) #define RETURNSPI(rp) do { \ if (!(rp)->lo_set) \ RETURNI((rp)->hi); \ RETURNI((rp)->hi + (rp)->lo); \ } while (0) #endif #define SUM2P(x, y) ({ \ const __typeof (x) __x = (x); \ const __typeof (y) __y = (y); \ __x + __y; \ }) /* fdlibm kernel function */ int __kernel_rem_pio2(double*,double*,int,int,int); /* double precision kernel functions */ #ifndef INLINE_REM_PIO2 int __ieee754_rem_pio2(double,double*); #endif double __kernel_sin(double,double,int); double __kernel_cos(double,double); double __kernel_tan(double,double,int); double __ldexp_exp(double,int); #ifdef _COMPLEX_H double complex __ldexp_cexp(double complex,int); #endif /* float precision kernel functions */ #ifndef INLINE_REM_PIO2F int __ieee754_rem_pio2f(float,double*); #endif #ifndef INLINE_KERNEL_SINDF float __kernel_sindf(double); #endif #ifndef INLINE_KERNEL_COSDF float __kernel_cosdf(double); #endif #ifndef INLINE_KERNEL_TANDF float __kernel_tandf(double,int); #endif float __ldexp_expf(float,int); #ifdef _COMPLEX_H float complex __ldexp_cexpf(float complex,int); #endif /* long double precision kernel functions */ long double __kernel_sinl(long double, long double, int); long double __kernel_cosl(long double, long double); long double __kernel_tanl(long double, long double, int); #endif /* !_MATH_PRIVATE_H_ */