Thu Mar 14 06:01:07 1996 Roland McGrath <roland@charlie-brown.gnu.ai.mit.edu>

	* string/strnlen.c: New file.
	* string/Makefile (routines): Add strnlen.
	* string/string.h [__USE_GNU] (strnlen): Declare new function.
	[__OPTIMIZE__]: Define extern inline implementation of it.

	* stdlib/erand48_r.c: Use __mpn_construct_double instead of ldexp and
	addition, to avoid using anything from -lm.

Mon Mar  4 21:57:14 1996  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* sysdeps/unix/sysv/linux/m68k/Makefile: New file.

	* sysdeps/unix/sysv/linux/m68k/sysdep.h (SYS_ify): Redefine.
	(CALL_MCOUNT): New macro, empty unless [PROF].
	(ENTRY): Do CALL_MCOUNT just after the label.
	(JUMPTARGET): New macro.
	(SYSCALL_ERROR_HANDLER): Fix syntax.

Thu Mar 14 04:20:48 1996  Roland McGrath  <roland@charlie-brown.gnu.ai.mit.edu>

	* sysdeps/unix/sysv/linux/init-first.c (init): Call `__personality'
	function instead of using inline asm i386 syscall.
	* sysdeps/unix/sysv/linux/syscalls.list: Add personality syscall.

	* posix/unistd.h [__USE_BSD]: Declare usleep.

	* misc/efgcvt.c (MAXDIG): New macro computed from <floats.h> constants.
	(fcvt, ecvt): Use it for buffer size.

	* stdlib/drand48-iter.c (__drand48_iterate): Use u_int64_t instead of
	conditionalizing long vs long long.

	* stdlib/drand48-iter.c (__drand48_iterate): Don't check for null
	pointers; never return EFAULT.
	* stdlib/drand48_r.c (drand48_r): Likewise.
	* stdlib/erand48_r.c (erand48_r): Likewise.

1 files changed, 26 insertions, 16 deletions

diff --git a/stdlib/erand48_r.c b/stdlib/erand48_r.c
index 86d2f734d9..ab41474aca 100644
--- a/stdlib/erand48_r.c
+++ b/stdlib/erand48_r.c
@@ -17,9 +17,15 @@ 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.  */
 
-#include <errno.h>
-#include <math.h>
 #include <stdlib.h>
+#include "gmp.h"
+#include "gmp-mparam.h"
+#include <float.h>
+
+
+/* Function to construct a floating point number from an MP integer
+   containing the fraction bits, a base 2 exponent, and a sign flag.  */
+extern double __mpn_construct_double (mp_srcptr mpn, int exponent, int neg);
 
 int
 erand48_r (xsubi, buffer, result)
@@ -27,26 +33,30 @@ erand48_r (xsubi, buffer, result)
      struct drand48_data *buffer;
      double *result;
 {
-  int i;
-
-  /* Be generous for the arguments, detect some errors.  */
-  if (result == NULL)
-    {
-      errno = EFAULT;
-      return -1;
-    }
+  mp_limb mpn[(3 * sizeof (unsigned short int) + sizeof (mp_limb) - 1)
+	      / sizeof (mp_limb)];
 
   /* Compute next state.  */
   if (__drand48_iterate (xsubi, buffer) < 0)
     return -1;
 
-  *result = 0.0;
-  for (i = 4 / sizeof (unsigned short int); i >= 0; --i)
-    {
-      double factor = ldexp (1.0, (i - 6) * sizeof (unsigned short int));
+  /* Build a 48-bit mpn containing the 48 random bits.  */
+
+#if BITS_PER_MP_LIMB == 64
+  mpn[0] = (xsubi[0] << 32) | (xsubi[1] << 16) | xsubi[2];
+#elif BITS_PER_MP_LIMB == 32
+  mpn[0] = (xsubi[1] << 16) | xsubi[2];
+  mpn[1] = xsubi[0];
+#else
+ #error "BITS_PER_MP_LIMB value not handled"
+#endif
+
+  /* Shift them up so they are most significant bits of the fraction.  */
+  __mpn_lshift (mpn, mpn, sizeof mpn / sizeof mpn[0], DBL_MANT_DIG - 48);
 
-      *result += factor * (double) xsubi[i];
-    }
+  /* Construct a positive double using those bits for the fractional part,
+     and a zero exponent so the resulting FP number is [0.0,1.0).  */
+  *result = __mpn_construct_double (mpn, 0, 0);
 
   return 0;
 }