Update.

1997-04-21 13:25  Ulrich Drepper  <drepper@cygnus.com>

	* manual/arith.texi: Add description for INFINITY, _Imaginary_I,
	fpclassify & friends, and complex number operations.
	Update various other math functions for ISO C 9X.
	* manual/math.texi: Update various entries for ISO C 9X.
	Add description for complex number functions.
	Add description of rand48 function family.
	* manual/string.h: Add description of a64l and l64a.

	* math/cmathcalls.h: Fix typo.

	* stdlib/a64l.c: Pretty printing.

	* stdlib/seed48_r.c: Also reset `a' and `c' to default values.
	* stdlib/srand48_r.c: Likewise.
	* stdlib/stdlib.h: Pretty printing.

	* sysdeps/i386/fpu/__math.h: Fix typo.

	* sysdeps/libm-ieee754/s_nearbyintf.c: Correctly name function.
	* sysdeps/libm-ieee754/s_nearbyintl.c: Likewise.

1997-04-19 22:16  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* sysdeps/m68k/fpu/e_pow.c: Rewrite handling of integral exponent.

1997-04-18 19:34  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* sysdeps/m68k/fpu/__math.h: Define optimized versions of
	isgreater, isgreaterequal, isless, islessequal, islessgreater, and
	isunordered.

1997-04-20 01:28  Richard Henderson  <rth@tamu.edu>

	* rellns-sh: Handle files in the same directory correctly.

1997-04-20 11:22  Ulrich Drepper  <drepper@cygnus.com>

	* csu/initfini.c: Place ALIGN instruction at correct positions.
	Patch by Richard Henderson <richard@twiddle.rth.home>.

1997-04-19 17:12  Ulrich Drepper  <drepper@cygnus.com>

	* Make-dist: Don't automatically ignore .c files if the .S or .s file
	is ignored.

	* csu/Makefile (distribute): Add defs.awk.

1997-04-19 15:39  Ulrich Drepper  <drepper@cygnus.com>

	* sysdeps/stub/shmat.c: Update to XPG4.2 interface.
	* sysdeps/stub/shmdt.c: Likewise.
	Reported by Thomas Bushnell, n/BSG.

1997-04-19 13:22  Ulrich Drepper  <drepper@cygnus.com>

	* manual/stdio.texi: Add description of printf_size and
	printf_size_info.  Partly based on the documentation by Larry McVoy.

1997-04-19 02:21  Ulrich Drepper  <drepper@cygnus.com>

	* stdio-common/printf_size.c (printf_size): Correct values for
	`units'.
	Report by Larry McVoy <lm@neteng.engr.sgi.com>.
	* stdio-common/tst-printfsz.c: New file.
	* stdio-common/Makefile (tests): Add tst-printfsz.c.
	(CFLAGS-tst-printfsz.c): Define to prevent warnings about format
	strings.

1997-04-18 15:48  Ulrich Drepper  <drepper@cygnus.com>

	* login/utmp.h: Add prototype for updwtmp.
	* login/logwtmp.c: Add new function updwtmp which allows to write
	a complete record to the wtmp file.
	Patch by Miquel van Smoorenburg <miquels@cistron.nl>.

1997-04-17 17:57  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* math/Makefile (headers): Add mathbits.h.

1997-04-16 21:20  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* sysdeps/m68k/fpu/__math.h: Add inlined sincos{,l,f}.
	* sysdeps/m68k/fpu/s_sincos.c: New file.
	* sysdeps/m68k/fpu/s_sincosf.c: New file.
	* sysdeps/m68k/fpu/s_sincosl.c: New file.

	* sysdeps/libm-ieee754/e_scalb.c: Use internal names of the
	functions.
	* sysdeps/libm-ieee754/e_scalbl.c: Likewise.

	* sysdeps/libm-ieee754/s_ctanh.c: Use isfinite instead of finite.
	* sysdeps/libm-ieee754/s_ctanhf.c: Likewise.
	* sysdeps/libm-ieee754/s_ctanhl.c: Likewise.
	* sysdeps/libm-ieee754/s_ctan.c: Likewise.
	* sysdeps/libm-ieee754/s_ctanf.c: Likewise.
	* sysdeps/libm-ieee754/s_ctanl.c: Likewise.  Fix type of `res'.

1997-04-18 11:21  Ulrich Drepper  <drepper@cygnus.com>

	* shadow/fgetspent_r.c: Set *RESULT to NULL before returning error.
	Patch by Thorsten Kukuk <kukuk@vt.uni-paderborn.de>.

1 files changed, 17 insertions, 32 deletions

diff --git a/sysdeps/m68k/fpu/e_pow.c b/sysdeps/m68k/fpu/e_pow.c
index 284f1bf294..a39b63d342 100644
--- a/sysdeps/m68k/fpu/e_pow.c
+++ b/sysdeps/m68k/fpu/e_pow.c
@@ -80,51 +80,36 @@ s(__ieee754_pow) (float_type x, float_type y)
 	z = 1 / z;
       if (m81(__signbit) (x))
 	{
-	  float_type temp = m81(__rint) (y);
-	  if (y != temp)
+	  if (y != m81(__rint) (y))
 	    {
 	      if (x == -1)
 		z = 0.0/0.0;
 	    }
 	  else
-	    {
-	      if (sizeof (float_type) == sizeof (float))
-		{
-		  long i = (long) y;
-		  if (i & 1)
-		    z = -z;
-		}
-	      else
-		{
-		  long long i = (long long) y;
-		  if ((float_type) i == y && i & 1)
-		    z = -z;
-		}
-	    }
+	    goto maybe_negate;
 	}
       return z;
     }
 
   if (x < 0.0)
     {
-      float_type temp = m81(__rint) (y);
-      if (y == temp)
+      if (y == m81(__rint) (y))
 	{
-	  long long i = (long long) y;
 	  z = m81(__ieee754_exp) (y * m81(__ieee754_log) (-x));
-	  if (sizeof (float_type) == sizeof (float))
-	    {
-	      long i = (long) y;
-	      if (i & 1)
-		z = -z;
-	    }
-	  else
-	    {
-	      /* If the conversion to long long was inexact assume that y
-		 is an even integer.  */
-	      if ((float_type) i == y && i & 1)
-		z = -z;
-	    }
+	maybe_negate:
+	  /* We always use the long double format, since y is already in
+	     this format and rounding won't change the result.  */
+	  {
+	    int32_t exponent;
+	    u_int32_t i0, i1;
+	    GET_LDOUBLE_WORDS (exponent, i0, i1, y);
+	    exponent = (exponent & 0x7fff) - 0x3fff;
+	    if (exponent <= 31
+		? i0 & (1 << (31 - exponent))
+		: (exponent <= 63
+		   && i1 & (1 << (63 - exponent))))
+	      z = -z;
+	  }
 	}
       else
 	z = 0.0/0.0;