Delete unused header file math-emu/extended.h
authorRobert P. J. Day <rpjday@mindspring.com>
Tue, 8 May 2007 07:27:17 +0000 (00:27 -0700)
committerLinus Torvalds <torvalds@woody.linux-foundation.org>
Tue, 8 May 2007 18:15:05 +0000 (11:15 -0700)
Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
include/math-emu/extended.h [deleted file]

diff --git a/include/math-emu/extended.h b/include/math-emu/extended.h
deleted file mode 100644 (file)
index 84770fc..0000000
+++ /dev/null
@@ -1,396 +0,0 @@
-/* Software floating-point emulation.
-   Definitions for IEEE Extended Precision.
-   Copyright (C) 1999 Free Software Foundation, Inc.
-   This file is part of the GNU C Library.
-   Contributed by Jakub Jelinek (jj@ultra.linux.cz).
-
-   The GNU C Library is free software; you can redistribute it and/or
-   modify it under the terms of the GNU Library General Public License as
-   published by the Free Software Foundation; either version 2 of the
-   License, or (at your option) any later version.
-
-   The GNU C Library is distributed in the hope that it will be useful,
-   but WITHOUT ANY WARRANTY; without even the implied warranty of
-   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
-   Library General Public License for more details.
-
-   You should have received a copy of the GNU Library General Public
-   License along with the GNU C Library; see the file COPYING.LIB.  If
-   not, write to the Free Software Foundation, Inc.,
-   59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
-
-
-#ifndef    __MATH_EMU_EXTENDED_H__
-#define    __MATH_EMU_EXTENDED_H__
-
-#if _FP_W_TYPE_SIZE < 32
-#error "Here's a nickel, kid. Go buy yourself a real computer."
-#endif
-
-#if _FP_W_TYPE_SIZE < 64
-#define _FP_FRACTBITS_E         (4*_FP_W_TYPE_SIZE)
-#else
-#define _FP_FRACTBITS_E                (2*_FP_W_TYPE_SIZE)
-#endif
-
-#define _FP_FRACBITS_E         64
-#define _FP_FRACXBITS_E                (_FP_FRACTBITS_E - _FP_FRACBITS_E)
-#define _FP_WFRACBITS_E                (_FP_WORKBITS + _FP_FRACBITS_E)
-#define _FP_WFRACXBITS_E       (_FP_FRACTBITS_E - _FP_WFRACBITS_E)
-#define _FP_EXPBITS_E          15
-#define _FP_EXPBIAS_E          16383
-#define _FP_EXPMAX_E           32767
-
-#define _FP_QNANBIT_E          \
-       ((_FP_W_TYPE)1 << (_FP_FRACBITS_E-2) % _FP_W_TYPE_SIZE)
-#define _FP_IMPLBIT_E          \
-       ((_FP_W_TYPE)1 << (_FP_FRACBITS_E-1) % _FP_W_TYPE_SIZE)
-#define _FP_OVERFLOW_E         \
-       ((_FP_W_TYPE)1 << (_FP_WFRACBITS_E % _FP_W_TYPE_SIZE))
-
-#if _FP_W_TYPE_SIZE < 64
-
-union _FP_UNION_E
-{
-   long double flt;
-   struct 
-   {
-#if __BYTE_ORDER == __BIG_ENDIAN
-      unsigned long pad1 : _FP_W_TYPE_SIZE;
-      unsigned long pad2 : (_FP_W_TYPE_SIZE - 1 - _FP_EXPBITS_E);
-      unsigned long sign : 1;
-      unsigned long exp : _FP_EXPBITS_E;
-      unsigned long frac1 : _FP_W_TYPE_SIZE;
-      unsigned long frac0 : _FP_W_TYPE_SIZE;
-#else
-      unsigned long frac0 : _FP_W_TYPE_SIZE;
-      unsigned long frac1 : _FP_W_TYPE_SIZE;
-      unsigned exp : _FP_EXPBITS_E;
-      unsigned sign : 1;
-#endif /* not bigendian */
-   } bits __attribute__((packed));
-};
-
-
-#define FP_DECL_E(X)           _FP_DECL(4,X)
-
-#define FP_UNPACK_RAW_E(X, val)                                \
-  do {                                                 \
-    union _FP_UNION_E _flo; _flo.flt = (val);          \
-                                                       \
-    X##_f[2] = 0; X##_f[3] = 0;                                \
-    X##_f[0] = _flo.bits.frac0;                                \
-    X##_f[1] = _flo.bits.frac1;                                \
-    X##_e  = _flo.bits.exp;                            \
-    X##_s  = _flo.bits.sign;                           \
-    if (!X##_e && (X##_f[1] || X##_f[0])               \
-        && !(X##_f[1] & _FP_IMPLBIT_E))                        \
-      {                                                        \
-        X##_e++;                                       \
-        FP_SET_EXCEPTION(FP_EX_DENORM);                        \
-      }                                                        \
-  } while (0)
-
-#define FP_UNPACK_RAW_EP(X, val)                       \
-  do {                                                 \
-    union _FP_UNION_E *_flo =                          \
-    (union _FP_UNION_E *)(val);                                \
-                                                       \
-    X##_f[2] = 0; X##_f[3] = 0;                                \
-    X##_f[0] = _flo->bits.frac0;                       \
-    X##_f[1] = _flo->bits.frac1;                       \
-    X##_e  = _flo->bits.exp;                           \
-    X##_s  = _flo->bits.sign;                          \
-    if (!X##_e && (X##_f[1] || X##_f[0])               \
-        && !(X##_f[1] & _FP_IMPLBIT_E))                        \
-      {                                                        \
-        X##_e++;                                       \
-        FP_SET_EXCEPTION(FP_EX_DENORM);                        \
-      }                                                        \
-  } while (0)
-
-#define FP_PACK_RAW_E(val, X)                          \
-  do {                                                 \
-    union _FP_UNION_E _flo;                            \
-                                                       \
-    if (X##_e) X##_f[1] |= _FP_IMPLBIT_E;              \
-    else X##_f[1] &= ~(_FP_IMPLBIT_E);                 \
-    _flo.bits.frac0 = X##_f[0];                                \
-    _flo.bits.frac1 = X##_f[1];                                \
-    _flo.bits.exp   = X##_e;                           \
-    _flo.bits.sign  = X##_s;                           \
-                                                       \
-    (val) = _flo.flt;                                  \
-  } while (0)
-
-#define FP_PACK_RAW_EP(val, X)                         \
-  do {                                                 \
-    if (!FP_INHIBIT_RESULTS)                           \
-      {                                                        \
-       union _FP_UNION_E *_flo =                       \
-         (union _FP_UNION_E *)(val);                   \
-                                                       \
-       if (X##_e) X##_f[1] |= _FP_IMPLBIT_E;           \
-       else X##_f[1] &= ~(_FP_IMPLBIT_E);              \
-       _flo->bits.frac0 = X##_f[0];                    \
-       _flo->bits.frac1 = X##_f[1];                    \
-       _flo->bits.exp   = X##_e;                       \
-       _flo->bits.sign  = X##_s;                       \
-      }                                                        \
-  } while (0)
-
-#define FP_UNPACK_E(X,val)             \
-  do {                                 \
-    FP_UNPACK_RAW_E(X,val);            \
-    _FP_UNPACK_CANONICAL(E,4,X);       \
-  } while (0)
-
-#define FP_UNPACK_EP(X,val)            \
-  do {                                 \
-    FP_UNPACK_RAW_2_P(X,val);          \
-    _FP_UNPACK_CANONICAL(E,4,X);       \
-  } while (0)
-
-#define FP_PACK_E(val,X)               \
-  do {                                 \
-    _FP_PACK_CANONICAL(E,4,X);         \
-    FP_PACK_RAW_E(val,X);              \
-  } while (0)
-
-#define FP_PACK_EP(val,X)              \
-  do {                                 \
-    _FP_PACK_CANONICAL(E,4,X);         \
-    FP_PACK_RAW_EP(val,X);             \
-  } while (0)
-
-#define FP_ISSIGNAN_E(X)       _FP_ISSIGNAN(E,4,X)
-#define FP_NEG_E(R,X)          _FP_NEG(E,4,R,X)
-#define FP_ADD_E(R,X,Y)                _FP_ADD(E,4,R,X,Y)
-#define FP_SUB_E(R,X,Y)                _FP_SUB(E,4,R,X,Y)
-#define FP_MUL_E(R,X,Y)                _FP_MUL(E,4,R,X,Y)
-#define FP_DIV_E(R,X,Y)                _FP_DIV(E,4,R,X,Y)
-#define FP_SQRT_E(R,X)         _FP_SQRT(E,4,R,X)
-
-/*
- * Square root algorithms:
- * We have just one right now, maybe Newton approximation
- * should be added for those machines where division is fast.
- * This has special _E version because standard _4 square
- * root would not work (it has to start normally with the
- * second word and not the first), but as we have to do it
- * anyway, we optimize it by doing most of the calculations
- * in two UWtype registers instead of four.
- */
-#define _FP_SQRT_MEAT_E(R, S, T, X, q)                 \
-  do {                                                 \
-    q = (_FP_W_TYPE)1 << (_FP_W_TYPE_SIZE - 1);                \
-    _FP_FRAC_SRL_4(X, (_FP_WORKBITS));                 \
-    while (q)                                          \
-      {                                                        \
-       T##_f[1] = S##_f[1] + q;                        \
-       if (T##_f[1] <= X##_f[1])                       \
-         {                                             \
-           S##_f[1] = T##_f[1] + q;                    \
-           X##_f[1] -= T##_f[1];                       \
-           R##_f[1] += q;                              \
-         }                                             \
-       _FP_FRAC_SLL_2(X, 1);                           \
-       q >>= 1;                                        \
-      }                                                        \
-    q = (_FP_W_TYPE)1 << (_FP_W_TYPE_SIZE - 1);                \
-    while (q)                                          \
-      {                                                        \
-       T##_f[0] = S##_f[0] + q;                        \
-       T##_f[1] = S##_f[1];                            \
-       if (T##_f[1] < X##_f[1] ||                      \
-           (T##_f[1] == X##_f[1] &&                    \
-            T##_f[0] <= X##_f[0]))                     \
-         {                                             \
-           S##_f[0] = T##_f[0] + q;                    \
-           S##_f[1] += (T##_f[0] > S##_f[0]);          \
-           _FP_FRAC_DEC_2(X, T);                       \
-           R##_f[0] += q;                              \
-         }                                             \
-       _FP_FRAC_SLL_2(X, 1);                           \
-       q >>= 1;                                        \
-      }                                                        \
-    _FP_FRAC_SLL_4(R, (_FP_WORKBITS));                 \
-    if (X##_f[0] | X##_f[1])                           \
-      {                                                        \
-       if (S##_f[1] < X##_f[1] ||                      \
-           (S##_f[1] == X##_f[1] &&                    \
-            S##_f[0] < X##_f[0]))                      \
-         R##_f[0] |= _FP_WORK_ROUND;                   \
-       R##_f[0] |= _FP_WORK_STICKY;                    \
-      }                                                        \
-  } while (0)
-
-#define FP_CMP_E(r,X,Y,un)     _FP_CMP(E,4,r,X,Y,un)
-#define FP_CMP_EQ_E(r,X,Y)     _FP_CMP_EQ(E,4,r,X,Y)
-
-#define FP_TO_INT_E(r,X,rsz,rsg)       _FP_TO_INT(E,4,r,X,rsz,rsg)
-#define FP_TO_INT_ROUND_E(r,X,rsz,rsg) _FP_TO_INT_ROUND(E,4,r,X,rsz,rsg)
-#define FP_FROM_INT_E(X,r,rs,rt)       _FP_FROM_INT(E,4,X,r,rs,rt)
-
-#define _FP_FRAC_HIGH_E(X)     (X##_f[2])
-#define _FP_FRAC_HIGH_RAW_E(X) (X##_f[1])
-
-#else   /* not _FP_W_TYPE_SIZE < 64 */
-union _FP_UNION_E
-{
-  long double flt /* __attribute__((mode(TF))) */ ;
-  struct {
-#if __BYTE_ORDER == __BIG_ENDIAN
-    unsigned long pad : (_FP_W_TYPE_SIZE - 1 - _FP_EXPBITS_E);
-    unsigned sign  : 1;
-    unsigned exp   : _FP_EXPBITS_E;
-    unsigned long frac : _FP_W_TYPE_SIZE;
-#else
-    unsigned long frac : _FP_W_TYPE_SIZE;
-    unsigned exp   : _FP_EXPBITS_E;
-    unsigned sign  : 1;
-#endif
-  } bits;
-};
-
-#define FP_DECL_E(X)           _FP_DECL(2,X)
-
-#define FP_UNPACK_RAW_E(X, val)                                        \
-  do {                                                         \
-    union _FP_UNION_E _flo; _flo.flt = (val);                  \
-                                                               \
-    X##_f0 = _flo.bits.frac;                                   \
-    X##_f1 = 0;                                                        \
-    X##_e = _flo.bits.exp;                                     \
-    X##_s = _flo.bits.sign;                                    \
-    if (!X##_e && X##_f0 && !(X##_f0 & _FP_IMPLBIT_E))         \
-      {                                                                \
-        X##_e++;                                               \
-        FP_SET_EXCEPTION(FP_EX_DENORM);                                \
-      }                                                                \
-  } while (0)
-
-#define FP_UNPACK_RAW_EP(X, val)                               \
-  do {                                                         \
-    union _FP_UNION_E *_flo =                                  \
-      (union _FP_UNION_E *)(val);                              \
-                                                               \
-    X##_f0 = _flo->bits.frac;                                  \
-    X##_f1 = 0;                                                        \
-    X##_e = _flo->bits.exp;                                    \
-    X##_s = _flo->bits.sign;                                   \
-    if (!X##_e && X##_f0 && !(X##_f0 & _FP_IMPLBIT_E))         \
-      {                                                                \
-        X##_e++;                                               \
-        FP_SET_EXCEPTION(FP_EX_DENORM);                                \
-      }                                                                \
-  } while (0)
-
-#define FP_PACK_RAW_E(val, X)                                  \
-  do {                                                         \
-    union _FP_UNION_E _flo;                                    \
-                                                               \
-    if (X##_e) X##_f0 |= _FP_IMPLBIT_E;                                \
-    else X##_f0 &= ~(_FP_IMPLBIT_E);                           \
-    _flo.bits.frac = X##_f0;                                   \
-    _flo.bits.exp  = X##_e;                                    \
-    _flo.bits.sign = X##_s;                                    \
-                                                               \
-    (val) = _flo.flt;                                          \
-  } while (0)
-
-#define FP_PACK_RAW_EP(fs, val, X)                             \
-  do {                                                         \
-    if (!FP_INHIBIT_RESULTS)                                   \
-      {                                                                \
-       union _FP_UNION_E *_flo =                               \
-         (union _FP_UNION_E *)(val);                           \
-                                                               \
-       if (X##_e) X##_f0 |= _FP_IMPLBIT_E;                     \
-       else X##_f0 &= ~(_FP_IMPLBIT_E);                        \
-       _flo->bits.frac = X##_f0;                               \
-       _flo->bits.exp  = X##_e;                                \
-       _flo->bits.sign = X##_s;                                \
-      }                                                                \
-  } while (0)
-
-
-#define FP_UNPACK_E(X,val)             \
-  do {                                 \
-    FP_UNPACK_RAW_E(X,val);            \
-    _FP_UNPACK_CANONICAL(E,2,X);       \
-  } while (0)
-
-#define FP_UNPACK_EP(X,val)            \
-  do {                                 \
-    FP_UNPACK_RAW_EP(X,val);           \
-    _FP_UNPACK_CANONICAL(E,2,X);       \
-  } while (0)
-
-#define FP_PACK_E(val,X)               \
-  do {                                 \
-    _FP_PACK_CANONICAL(E,2,X);         \
-    FP_PACK_RAW_E(val,X);              \
-  } while (0)
-
-#define FP_PACK_EP(val,X)              \
-  do {                                 \
-    _FP_PACK_CANONICAL(E,2,X);         \
-    FP_PACK_RAW_EP(val,X);             \
-  } while (0)
-
-#define FP_ISSIGNAN_E(X)       _FP_ISSIGNAN(E,2,X)
-#define FP_NEG_E(R,X)          _FP_NEG(E,2,R,X)
-#define FP_ADD_E(R,X,Y)                _FP_ADD(E,2,R,X,Y)
-#define FP_SUB_E(R,X,Y)                _FP_SUB(E,2,R,X,Y)
-#define FP_MUL_E(R,X,Y)                _FP_MUL(E,2,R,X,Y)
-#define FP_DIV_E(R,X,Y)                _FP_DIV(E,2,R,X,Y)
-#define FP_SQRT_E(R,X)         _FP_SQRT(E,2,R,X)
-
-/*
- * Square root algorithms:
- * We have just one right now, maybe Newton approximation
- * should be added for those machines where division is fast.
- * We optimize it by doing most of the calculations
- * in one UWtype registers instead of two, although we don't
- * have to.
- */
-#define _FP_SQRT_MEAT_E(R, S, T, X, q)                 \
-  do {                                                 \
-    q = (_FP_W_TYPE)1 << (_FP_W_TYPE_SIZE - 1);                \
-    _FP_FRAC_SRL_2(X, (_FP_WORKBITS));                 \
-    while (q)                                          \
-      {                                                        \
-        T##_f0 = S##_f0 + q;                           \
-        if (T##_f0 <= X##_f0)                          \
-          {                                            \
-            S##_f0 = T##_f0 + q;                       \
-            X##_f0 -= T##_f0;                          \
-            R##_f0 += q;                               \
-          }                                            \
-        _FP_FRAC_SLL_1(X, 1);                          \
-        q >>= 1;                                       \
-      }                                                        \
-    _FP_FRAC_SLL_2(R, (_FP_WORKBITS));                 \
-    if (X##_f0)                                                \
-      {                                                        \
-       if (S##_f0 < X##_f0)                            \
-         R##_f0 |= _FP_WORK_ROUND;                     \
-       R##_f0 |= _FP_WORK_STICKY;                      \
-      }                                                        \
-  } while (0)
-#define FP_CMP_E(r,X,Y,un)     _FP_CMP(E,2,r,X,Y,un)
-#define FP_CMP_EQ_E(r,X,Y)     _FP_CMP_EQ(E,2,r,X,Y)
-
-#define FP_TO_INT_E(r,X,rsz,rsg)       _FP_TO_INT(E,2,r,X,rsz,rsg)
-#define FP_TO_INT_ROUND_E(r,X,rsz,rsg) _FP_TO_INT_ROUND(E,2,r,X,rsz,rsg)
-#define FP_FROM_INT_E(X,r,rs,rt)       _FP_FROM_INT(E,2,X,r,rs,rt)
-
-#define _FP_FRAC_HIGH_E(X)     (X##_f1)
-#define _FP_FRAC_HIGH_RAW_E(X) (X##_f0)
-
-#endif /* not _FP_W_TYPE_SIZE < 64 */
-
-#endif /* __MATH_EMU_EXTENDED_H__ */