This is the mail archive of the libc-hacker@sourceware.org mailing list for the glibc project.

Note that libc-hacker is a closed list. You may look at the archives of this list, but subscription and posting are not open.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

[PATCH] Move powerpc-cpu math functions to trunc

From: Steven Munroe <munroesj at us dot ibm dot com>
To: Jakub Jelinek <jakub at redhat dot com>, GNU libc hacker <libc-hacker at sources dot redhat dot com>
Date: Fri, 01 Jun 2007 17:16:24 -0500
Subject: [PATCH] Move powerpc-cpu math functions to trunc
References: <4653610A.1060101@us.ibm.com> <20070522214043.GN4033@devserv.devel.redhat.com> <46560D3F.80809@us.ibm.com> <20070529213339.GJ4033@devserv.devel.redhat.com>

Jakub Jelinek wrote:
> Hi!
>
> Ulrich just asked me about your add-on, are you planning to submit it
> for inclusion into CVS (not as an add-on, but in normal sysdeps hierarchy)?
> He'd certainly prefer that over an add-on, ASAP.
>
> 	Jakub
>   

Here are the math functions and supporting Implies files, ported to
the normal sysdeps hierarchy. Apply the strings patch first.

2007-05-31  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/970/fpu/Implies: New file.
	* sysdeps/powerpc/powerpc32/power5/fpu/Implies: New file.
	* sysdeps/powerpc/powerpc32/power5+/fpu/Implies: New file.
	* sysdeps/powerpc/powerpc32/power6/fpu/Implies: New file.
	* sysdeps/powerpc/powerpc32/power6x/fpu/Implies: New file.
	* sysdeps/powerpc/powerpc64/970/fpu/Implies: New file.
	* sysdeps/powerpc/powerpc64/power5/fpu/Implies: New file.
	* sysdeps/powerpc/powerpc64/power5+/fpu/Implies: New file.
	* sysdeps/powerpc/powerpc64/power6/fpu/Implies: New file.
	* sysdeps/powerpc/powerpc64/power6x/fpu/Implies: New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc32/970/fpu/Implies: New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc32/power4/fpu/Implies:
	New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc32/power5/fpu/Implies:
	New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc32/power5+/fpu/Implies:
	New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc32/power6/fpu/Implies:
	New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc32/power6x/fpu/Implies:
	New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc64/970/fpu/Implies: New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc64/power4/fpu/Implies:
	New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc64/power5/fpu/Implies:
	New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc64/power5+/fpu/Implies:
	New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc64/power6/fpu/Implies:
	New file.
	* sysdeps/unix/sysv/linux/powerpc/powerpc64/power6x/fpu/Implies:
	New file.

2007-05-31  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/powerpc64/fpu/s_llrint.S: Move.
	* sysdeps/powerpc/powerpc32/power4/fpu/s_llrint.S: To here.
	* sysdeps/powerpc/powerpc32/powerpc64/fpu/s_llrintf.S: Move.
	* sysdeps/powerpc/powerpc32/power4/fpu/s_llrintf.S: To here.
	* sysdeps/powerpc/powerpc32/powerpc64/fpu/s_llround.S: Move.
	* sysdeps/powerpc/powerpc32/power4/fpu/s_llround.S: To here.
	* sysdeps/powerpc/powerpc32/powerpc64/fpu/s_llroundf.S: Move.
	* sysdeps/powerpc/powerpc32/power4/fpu/s_llroundf.S: To here.

2007-05-22  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/power5+/fpu/s_round.S
	(LONG_DOUBLE_COMPAT): Specify correct version, GLIBC_2_1.
	* sysdeps/powerpc/powerpc32/power5+/fpu/s_trunc.S
	(LONG_DOUBLE_COMPAT): Specify correct version, GLIBC_2_1.
	* sysdeps/powerpc/powerpc64/power5+/fpu/s_round.S
	(LONG_DOUBLE_COMPAT): Specify correct version, GLIBC_2_1.
	* sysdeps/powerpc/powerpc64/power5+/fpu/s_trunc.S
	(LONG_DOUBLE_COMPAT): Specify correct version, GLIBC_2_1.

2007-05-21  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/power4/fpu/slowexp.c: New file.
	* sysdeps/powerpc/powerpc32/power4/fpu/w_sqrt.c: New file.
	* sysdeps/powerpc/powerpc64/power4/fpu/slowexp.c: New file.
	* sysdeps/powerpc/powerpc64/power4/fpu/w_sqrt.c: New file.

2007-03-15  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/powerpc64/fpu/s_llrint.S
	[LONG_DOUBLE_COMPAT]: Add compat_symbol for llrintl@@GLIBC_2_1.

2006-02-13  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/power6/fpu/s_llrint.S: New File
	* sysdeps/powerpc/powerpc32/power6/fpu/s_llrintf.S: New File
	* sysdeps/powerpc/powerpc32/power6/fpu/s_llround.S: New File
	* sysdeps/powerpc/powerpc32/power6/fpu/s_llroundf.S: New File

2006-10-20  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/power4/fpu/slowpow.c: New file.
	* sysdeps/powerpc/powerpc64/power4/fpu/slowpow.c: New file.

2006-10-03  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/powerpc64/fpu/s_llround.S: New file.
	* sysdeps/powerpc/powerpc32/powerpc64/fpu/s_llroundf.S: New file.

	* sysdeps/powerpc/powerpc32/powerpc64/fpu/Makefile: Moved.
	* sysdeps/powerpc/powerpc32/powerpc64/fpu/mpa.c: Likewise.
	* sysdeps/powerpc/powerpc32/power4/fpu/Makefile: To here.
	* sysdeps/powerpc/powerpc32/power4/fpu/mpa.c: Likewise.

2006-09-29  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/power6x/fpu/s_lrint.S: New file.
	* sysdeps/powerpc/powerpc32/power6x/fpu/s_lround.S: New file.
	* sysdeps/powerpc/powerpc64/power6x/fpu/s_llrint.S: New file.
	* sysdeps/powerpc/powerpc64/power6x/fpu/s_llround.S: New file.

2006-09-28  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/power5+/fpu/s_llround.S: New file.
	* sysdeps/powerpc/powerpc32/power5+/fpu/s_llroundf.S: New file.
	* sysdeps/powerpc/powerpc32/power5+/fpu/s_lround.S: New file.
	* sysdeps/powerpc/powerpc32/power6x/fpu/Implies: New file.
	* sysdeps/powerpc/powerpc64/power5+/fpu/s_llround.S: New file.
	* sysdeps/powerpc/powerpc64/power6x/fpu/Implies: New file.

2006-08-31  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/powerpc64/fpu/Makefile: New file.
	* sysdeps/powerpc/powerpc32/powerpc64/fpu/mpa.c: New file.
	* sysdeps/powerpc/powerpc64/power4/fpu/Makefile: New file.
	* sysdeps/powerpc/powerpc64/power4/fpu/mpa.c: New file.

2006-06-15  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/power5+/fpu/s_ceil.S: New file.
	* sysdeps/powerpc/powerpc32/power5+/fpu/s_ceilf.S: New file.
	* sysdeps/powerpc/powerpc32/power5+/fpu/s_floor.S: New file.
	* sysdeps/powerpc/powerpc32/power5+/fpu/s_floorf.S: New file.
	* sysdeps/powerpc/powerpc32/power5+/fpu/s_round.S: New file.
	* sysdeps/powerpc/powerpc32/power5+/fpu/s_roundf.S: New file.
	* sysdeps/powerpc/powerpc32/power5+/fpu/s_trunc.S: New file.
	* sysdeps/powerpc/powerpc32/power5+/fpu/s_truncf.S: New file.
	* sysdeps/powerpc/powerpc64/power5+/fpu/s_ceil.S: New file.
	* sysdeps/powerpc/powerpc64/power5+/fpu/s_ceilf.S: New file.
	* sysdeps/powerpc/powerpc64/power5+/fpu/s_floor.S: New file.
	* sysdeps/powerpc/powerpc64/power5+/fpu/s_floorf.S: New file.
	* sysdeps/powerpc/powerpc64/power5+/fpu/s_round.S: New file.
	* sysdeps/powerpc/powerpc64/power5+/fpu/s_roundf.S: New file.
	* sysdeps/powerpc/powerpc64/power5+/fpu/s_trunc.S: New file.
	* sysdeps/powerpc/powerpc64/power5+/fpu/s_truncf.S: New file.

2006-03-20  Steven Munroe  <sjmunroe@us.ibm.com>

	* sysdeps/powerpc/powerpc32/powerpc64/fpu/s_llrint.S: New file.
	* sysdeps/powerpc/powerpc32/powerpc64/fpu/s_llrintf.S: New file.

diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/970/fpu/Implies libc25/sysdeps/powerpc/powerpc32/970/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/970/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/970/fpu/Implies	Thu May 31 08:19:49 2007
@@ -0,0 +1,1 @@
+powerpc/powerpc32/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/Makefile libc25/sysdeps/powerpc/powerpc32/power4/fpu/Makefile
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/Makefile	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power4/fpu/Makefile	Tue Oct 10 09:58:56 2006
@@ -0,0 +1,5 @@
+# Makefile fragment for POWER4/5/5+ with FPU.
+
+ifeq ($(subdir),math)
+CFLAGS-mpa.c += --param max-unroll-times=4 -funroll-loops -fpeel-loops -ftree-loop-linear 
+endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/mpa.c libc25/sysdeps/powerpc/powerpc32/power4/fpu/mpa.c
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/mpa.c	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power4/fpu/mpa.c	Tue Oct 10 09:58:56 2006
@@ -0,0 +1,549 @@
+
+/*
+ * IBM Accurate Mathematical Library
+ * written by International Business Machines Corp.
+ * Copyright (C) 2001, 2006 Free Software Foundation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU Lesser General Public License as published by
+ * the Free Software Foundation; either version 2.1 of the License, or
+ * (at your option) any later version.
+ *
+ * This program 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 Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+/************************************************************************/
+/*  MODULE_NAME: mpa.c                                                  */
+/*                                                                      */
+/*  FUNCTIONS:                                                          */
+/*               mcr                                                    */
+/*               acr                                                    */
+/*               cr                                                     */
+/*               cpy                                                    */
+/*               cpymn                                                  */
+/*               norm                                                   */
+/*               denorm                                                 */
+/*               mp_dbl                                                 */
+/*               dbl_mp                                                 */
+/*               add_magnitudes                                         */
+/*               sub_magnitudes                                         */
+/*               add                                                    */
+/*               sub                                                    */
+/*               mul                                                    */
+/*               inv                                                    */
+/*               dvd                                                    */
+/*                                                                      */
+/* Arithmetic functions for multiple precision numbers.                 */
+/* Relative errors are bounded                                          */
+/************************************************************************/
+
+
+#include "endian.h"
+#include "mpa.h"
+#include "mpa2.h"
+#include <sys/param.h>	/* For MIN() */
+/* mcr() compares the sizes of the mantissas of two multiple precision  */
+/* numbers. Mantissas are compared regardless of the signs of the       */
+/* numbers, even if x->d[0] or y->d[0] are zero. Exponents are also     */
+/* disregarded.                                                         */
+static int mcr(const mp_no *x, const mp_no *y, int p) {
+  long i;
+  long p2 = p;
+  for (i=1; i<=p2; i++) {
+    if      (X[i] == Y[i])  continue;
+    else if (X[i] >  Y[i])  return  1;
+    else                    return -1; }
+  return 0;
+}
+
+
+
+/* acr() compares the absolute values of two multiple precision numbers */
+int __acr(const mp_no *x, const mp_no *y, int p) {
+  long i;
+
+  if      (X[0] == ZERO) {
+    if    (Y[0] == ZERO) i= 0;
+    else                 i=-1;
+  }
+  else if (Y[0] == ZERO) i= 1;
+  else {
+    if      (EX >  EY)   i= 1;
+    else if (EX <  EY)   i=-1;
+    else                 i= mcr(x,y,p);
+  }
+
+  return i;
+}
+
+
+/* cr90 compares the values of two multiple precision numbers           */
+int  __cr(const mp_no *x, const mp_no *y, int p) {
+  int i;
+
+  if      (X[0] > Y[0])  i= 1;
+  else if (X[0] < Y[0])  i=-1;
+  else if (X[0] < ZERO ) i= __acr(y,x,p);
+  else                   i= __acr(x,y,p);
+
+  return i;
+}
+
+
+/* Copy a multiple precision number. Set *y=*x. x=y is permissible.      */
+void __cpy(const mp_no *x, mp_no *y, int p) {
+  long i;
+
+  EY = EX;
+  for (i=0; i <= p; i++)    Y[i] = X[i];
+
+  return;
+}
+
+
+/* Copy a multiple precision number x of precision m into a */
+/* multiple precision number y of precision n. In case n>m, */
+/* the digits of y beyond the m'th are set to zero. In case */
+/* n<m, the digits of x beyond the n'th are ignored.        */
+/* x=y is permissible.                                      */
+
+void __cpymn(const mp_no *x, int m, mp_no *y, int n) {
+
+  long i,k;
+  long n2 = n;
+  long m2 = m;
+
+  EY = EX;     k=MIN(m2,n2);
+  for (i=0; i <= k; i++)    Y[i] = X[i];
+  for (   ; i <= n2; i++)    Y[i] = ZERO;
+
+  return;
+}
+
+/* Convert a multiple precision number *x into a double precision */
+/* number *y, normalized case  (|x| >= 2**(-1022))) */
+static void norm(const mp_no *x, double *y, int p)
+{
+  #define R  radixi.d
+  long i;
+#if 0
+  int k;
+#endif
+  double a,c,u,v,z[5];
+  if (p<5) {
+    if      (p==1) c = X[1];
+    else if (p==2) c = X[1] + R* X[2];
+    else if (p==3) c = X[1] + R*(X[2]  +   R* X[3]);
+    else if (p==4) c =(X[1] + R* X[2]) + R*R*(X[3] + R*X[4]);
+  }
+  else {
+    for (a=ONE, z[1]=X[1]; z[1] < TWO23; )
+        {a *= TWO;   z[1] *= TWO; }
+
+    for (i=2; i<5; i++) {
+      z[i] = X[i]*a;
+      u = (z[i] + CUTTER)-CUTTER;
+      if  (u > z[i])  u -= RADIX;
+      z[i] -= u;
+      z[i-1] += u*RADIXI;
+    }
+
+    u = (z[3] + TWO71) - TWO71;
+    if (u > z[3])   u -= TWO19;
+    v = z[3]-u;
+
+    if (v == TWO18) {
+      if (z[4] == ZERO) {
+        for (i=5; i <= p; i++) {
+          if (X[i] == ZERO)   continue;
+          else                {z[3] += ONE;   break; }
+        }
+      }
+      else              z[3] += ONE;
+    }
+
+    c = (z[1] + R *(z[2] + R * z[3]))/a;
+  }
+
+  c *= X[0];
+
+  for (i=1; i<EX; i++)   c *= RADIX;
+  for (i=1; i>EX; i--)   c *= RADIXI;
+
+  *y = c;
+  return;
+#undef R
+}
+
+/* Convert a multiple precision number *x into a double precision */
+/* number *y, denormalized case  (|x| < 2**(-1022))) */
+static void denorm(const mp_no *x, double *y, int p)
+{
+  long i,k;
+  long p2 = p;
+  double c,u,z[5];
+#if 0
+  double a,v;
+#endif
+
+#define R  radixi.d
+  if (EX<-44 || (EX==-44 && X[1]<TWO5))
+     { *y=ZERO; return; }
+
+  if      (p2==1) {
+    if      (EX==-42) {z[1]=X[1]+TWO10;  z[2]=ZERO;  z[3]=ZERO;  k=3;}
+    else if (EX==-43) {z[1]=     TWO10;  z[2]=X[1];  z[3]=ZERO;  k=2;}
+    else              {z[1]=     TWO10;  z[2]=ZERO;  z[3]=X[1];  k=1;}
+  }
+  else if (p2==2) {
+    if      (EX==-42) {z[1]=X[1]+TWO10;  z[2]=X[2];  z[3]=ZERO;  k=3;}
+    else if (EX==-43) {z[1]=     TWO10;  z[2]=X[1];  z[3]=X[2];  k=2;}
+    else              {z[1]=     TWO10;  z[2]=ZERO;  z[3]=X[1];  k=1;}
+  }
+  else {
+    if      (EX==-42) {z[1]=X[1]+TWO10;  z[2]=X[2];  k=3;}
+    else if (EX==-43) {z[1]=     TWO10;  z[2]=X[1];  k=2;}
+    else              {z[1]=     TWO10;  z[2]=ZERO;  k=1;}
+    z[3] = X[k];
+  }
+
+  u = (z[3] + TWO57) - TWO57;
+  if  (u > z[3])   u -= TWO5;
+
+  if (u==z[3]) {
+    for (i=k+1; i <= p2; i++) {
+      if (X[i] == ZERO)   continue;
+      else {z[3] += ONE;   break; }
+    }
+  }
+
+  c = X[0]*((z[1] + R*(z[2] + R*z[3])) - TWO10);
+
+  *y = c*TWOM1032;
+  return;
+
+#undef R
+}
+
+/* Convert a multiple precision number *x into a double precision number *y. */
+/* The result is correctly rounded to the nearest/even. *x is left unchanged */
+
+void __mp_dbl(const mp_no *x, double *y, int p) {
+#if 0
+  int i,k;
+  double a,c,u,v,z[5];
+#endif
+
+  if (X[0] == ZERO)  {*y = ZERO;  return; }
+
+  if      (EX> -42)                 norm(x,y,p);
+  else if (EX==-42 && X[1]>=TWO10)  norm(x,y,p);
+  else                              denorm(x,y,p);
+}
+
+
+/* dbl_mp() converts a double precision number x into a multiple precision  */
+/* number *y. If the precision p is too small the result is truncated. x is */
+/* left unchanged.                                                          */
+
+void __dbl_mp(double x, mp_no *y, int p) {
+
+  long i,n;
+  long p2 = p;
+  double u;
+
+  /* Sign */
+  if      (x == ZERO)  {Y[0] = ZERO;  return; }
+  else if (x >  ZERO)   Y[0] = ONE;
+  else                 {Y[0] = MONE;  x=-x;   }
+
+  /* Exponent */
+  for (EY=ONE; x >= RADIX; EY += ONE)   x *= RADIXI;
+  for (      ; x <  ONE;   EY -= ONE)   x *= RADIX;
+
+  /* Digits */
+  n=MIN(p2,4);
+  for (i=1; i<=n; i++) {
+    u = (x + TWO52) - TWO52;
+    if (u>x)   u -= ONE;
+    Y[i] = u;     x -= u;    x *= RADIX; }
+  for (   ; i<=p2; i++)     Y[i] = ZERO;
+  return;
+}
+
+
+/*  add_magnitudes() adds the magnitudes of *x & *y assuming that           */
+/*  abs(*x) >= abs(*y) > 0.                                                 */
+/* The sign of the sum *z is undefined. x&y may overlap but not x&z or y&z. */
+/* No guard digit is used. The result equals the exact sum, truncated.      */
+/* *x & *y are left unchanged.                                              */
+
+static void add_magnitudes(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  long i,j,k;
+  long p2 = p;
+
+  EZ = EX;
+
+  i=p2;    j=p2+ EY - EX;    k=p2+1;
+
+  if (j<1)
+     {__cpy(x,z,p);  return; }
+  else   Z[k] = ZERO;
+
+  for (; j>0; i--,j--) {
+    Z[k] += X[i] + Y[j];
+    if (Z[k] >= RADIX) {
+      Z[k]  -= RADIX;
+      Z[--k] = ONE; }
+    else
+      Z[--k] = ZERO;
+  }
+
+  for (; i>0; i--) {
+    Z[k] += X[i];
+    if (Z[k] >= RADIX) {
+      Z[k]  -= RADIX;
+      Z[--k] = ONE; }
+    else
+      Z[--k] = ZERO;
+  }
+
+  if (Z[1] == ZERO) {
+    for (i=1; i<=p2; i++)    Z[i] = Z[i+1]; }
+  else   EZ += ONE;
+}
+
+
+/*  sub_magnitudes() subtracts the magnitudes of *x & *y assuming that      */
+/*  abs(*x) > abs(*y) > 0.                                                  */
+/* The sign of the difference *z is undefined. x&y may overlap but not x&z  */
+/* or y&z. One guard digit is used. The error is less than one ulp.         */
+/* *x & *y are left unchanged.                                              */
+
+static void sub_magnitudes(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  long i,j,k;
+  long p2 = p;
+
+  EZ = EX;
+
+  if (EX == EY) {
+    i=j=k=p2;
+    Z[k] = Z[k+1] = ZERO; }
+  else {
+    j= EX - EY;
+    if (j > p2)  {__cpy(x,z,p);  return; }
+    else {
+      i=p2;   j=p2+1-j;   k=p2;
+      if (Y[j] > ZERO) {
+        Z[k+1] = RADIX - Y[j--];
+        Z[k]   = MONE; }
+      else {
+        Z[k+1] = ZERO;
+        Z[k]   = ZERO;   j--;}
+    }
+  }
+
+  for (; j>0; i--,j--) {
+    Z[k] += (X[i] - Y[j]);
+    if (Z[k] < ZERO) {
+      Z[k]  += RADIX;
+      Z[--k] = MONE; }
+    else
+      Z[--k] = ZERO;
+  }
+
+  for (; i>0; i--) {
+    Z[k] += X[i];
+    if (Z[k] < ZERO) {
+      Z[k]  += RADIX;
+      Z[--k] = MONE; }
+    else
+      Z[--k] = ZERO;
+  }
+
+  for (i=1; Z[i] == ZERO; i++) ;
+  EZ = EZ - i + 1;
+  for (k=1; i <= p2+1; )
+    Z[k++] = Z[i++];
+  for (; k <= p2; )
+    Z[k++] = ZERO;
+
+  return;
+}
+
+
+/* Add two multiple precision numbers. Set *z = *x + *y. x&y may overlap  */
+/* but not x&z or y&z. One guard digit is used. The error is less than    */
+/* one ulp. *x & *y are left unchanged.                                   */
+
+void __add(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  int n;
+
+  if      (X[0] == ZERO)     {__cpy(y,z,p);  return; }
+  else if (Y[0] == ZERO)     {__cpy(x,z,p);  return; }
+
+  if (X[0] == Y[0])   {
+    if (__acr(x,y,p) > 0)      {add_magnitudes(x,y,z,p);  Z[0] = X[0]; }
+    else                     {add_magnitudes(y,x,z,p);  Z[0] = Y[0]; }
+  }
+  else                       {
+    if ((n=__acr(x,y,p)) == 1) {sub_magnitudes(x,y,z,p);  Z[0] = X[0]; }
+    else if (n == -1)        {sub_magnitudes(y,x,z,p);  Z[0] = Y[0]; }
+    else                      Z[0] = ZERO;
+  }
+  return;
+}
+
+
+/* Subtract two multiple precision numbers. *z is set to *x - *y. x&y may */
+/* overlap but not x&z or y&z. One guard digit is used. The error is      */
+/* less than one ulp. *x & *y are left unchanged.                         */
+
+void __sub(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  int n;
+
+  if      (X[0] == ZERO)     {__cpy(y,z,p);  Z[0] = -Z[0];  return; }
+  else if (Y[0] == ZERO)     {__cpy(x,z,p);                 return; }
+
+  if (X[0] != Y[0])    {
+    if (__acr(x,y,p) > 0)      {add_magnitudes(x,y,z,p);  Z[0] =  X[0]; }
+    else                     {add_magnitudes(y,x,z,p);  Z[0] = -Y[0]; }
+  }
+  else                       {
+    if ((n=__acr(x,y,p)) == 1) {sub_magnitudes(x,y,z,p);  Z[0] =  X[0]; }
+    else if (n == -1)        {sub_magnitudes(y,x,z,p);  Z[0] = -Y[0]; }
+    else                      Z[0] = ZERO;
+  }
+  return;
+}
+
+
+/* Multiply two multiple precision numbers. *z is set to *x * *y. x&y      */
+/* may overlap but not x&z or y&z. In case p=1,2,3 the exact result is     */
+/* truncated to p digits. In case p>3 the error is bounded by 1.001 ulp.   */
+/* *x & *y are left unchanged.                                             */
+
+void __mul(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  long i, i1, i2, j, k, k2;
+  long p2 = p;
+  double u, zk, zk2;
+
+                      /* Is z=0? */
+  if (X[0]*Y[0]==ZERO)
+     { Z[0]=ZERO;  return; }
+
+                       /* Multiply, add and carry */
+  k2 = (p2<3) ? p2+p2 : p2+3;
+  zk = Z[k2]=ZERO;
+  for (k=k2; k>1; ) {
+    if (k > p2)  {i1=k-p2; i2=p2+1; }
+    else        {i1=1;   i2=k;   }
+#if 1
+    /* rearange this inner loop to allow the fmadd instructions to be
+       independent and execute in parallel on processors that have
+       dual symetrical FP pipelines.  */
+    if (i1 < (i2-1))
+    {
+	/* make sure we have at least 2 iterations */
+	if (((i2 - i1) & 1L) == 1L)
+	{
+                /* Handle the odd iterations case.  */
+		zk2 = x->d[i2-1]*y->d[i1];
+	}
+	else
+		zk2 = zero.d;
+	/* Do two multiply/adds per loop iteration, using independent
+           accumulators; zk and zk2.  */
+	for (i=i1,j=i2-1; i<i2-1; i+=2,j-=2) 
+	{
+		zk += x->d[i]*y->d[j];
+		zk2 += x->d[i+1]*y->d[j-1];
+	}
+	zk += zk2; /* final sum.  */
+    }
+    else
+    {
+        /* Special case when iterations is 1.  */
+	zk += x->d[i1]*y->d[i1];
+    }
+#else
+    /* The orginal code.  */
+    for (i=i1,j=i2-1; i<i2; i++,j--)  zk += X[i]*Y[j];
+#endif
+
+    u = (zk + CUTTER)-CUTTER;
+    if  (u > zk)  u -= RADIX;
+    Z[k]  = zk - u;
+    zk = u*RADIXI;
+    --k;
+  }
+  Z[k] = zk;
+
+                 /* Is there a carry beyond the most significant digit? */
+  if (Z[1] == ZERO) {
+    for (i=1; i<=p2; i++)  Z[i]=Z[i+1];
+    EZ = EX + EY - 1; }
+  else
+    EZ = EX + EY;
+
+  Z[0] = X[0] * Y[0];
+  return;
+}
+
+
+/* Invert a multiple precision number. Set *y = 1 / *x.                     */
+/* Relative error bound = 1.001*r**(1-p) for p=2, 1.063*r**(1-p) for p=3,   */
+/* 2.001*r**(1-p) for p>3.                                                  */
+/* *x=0 is not permissible. *x is left unchanged.                           */
+
+void __inv(const mp_no *x, mp_no *y, int p) {
+  long i;
+#if 0
+  int l;
+#endif
+  double t;
+  mp_no z,w;
+  static const int np1[] = {0,0,0,0,1,2,2,2,2,3,3,3,3,3,3,3,3,3,
+                            4,4,4,4,4,4,4,4,4,4,4,4,4,4,4};
+  const mp_no mptwo = {1,{1.0,2.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,
+                         0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,
+                         0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,
+                         0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0}};
+
+  __cpy(x,&z,p);  z.e=0;  __mp_dbl(&z,&t,p);
+  t=ONE/t;   __dbl_mp(t,y,p);    EY -= EX;
+
+  for (i=0; i<np1[p]; i++) {
+    __cpy(y,&w,p);
+    __mul(x,&w,y,p);
+    __sub(&mptwo,y,&z,p);
+    __mul(&w,&z,y,p);
+  }
+  return;
+}
+
+
+/* Divide one multiple precision number by another.Set *z = *x / *y. *x & *y */
+/* are left unchanged. x&y may overlap but not x&z or y&z.                   */
+/* Relative error bound = 2.001*r**(1-p) for p=2, 2.063*r**(1-p) for p=3     */
+/* and 3.001*r**(1-p) for p>3. *y=0 is not permissible.                      */
+
+void __dvd(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  mp_no w;
+
+  if (X[0] == ZERO)    Z[0] = ZERO;
+  else                {__inv(y,&w,p);   __mul(x,&w,z,p);}
+  return;
+}
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/s_llrint.S libc25/sysdeps/powerpc/powerpc32/power4/fpu/s_llrint.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/s_llrint.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power4/fpu/s_llrint.S	Thu May 10 11:28:20 2007
@@ -0,0 +1,47 @@
+/* Round double to long int.  PowerPC32 on PowerPC64 version.
+   Copyright (C) 2004, 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA
+   02110-1301 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+/* long long int[r3, r4] __llrint (double x[fp1])  */
+ENTRY (__llrint)	
+	CALL_MCOUNT
+	stwu	r1,-16(r1)
+	cfi_adjust_cfa_offset (16)
+	fctid	fp13,fp1
+	stfd	fp13,8(r1)
+	nop	/* Insure the following load is in a different dispatch group */
+	nop	/* to avoid pipe stall on POWER4&5.  */
+	nop
+	lwz	r3,8(r1)
+	lwz	r4,12(r1)
+	addi	r1,r1,16	
+	blr
+	END (__llrint)
+
+weak_alias (__llrint, llrint)
+
+#ifdef NO_LONG_DOUBLE
+strong_alias (__llrint, __llrintl)
+weak_alias (__llrint, llrintl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __llrint, llrintl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/s_llrintf.S libc25/sysdeps/powerpc/powerpc32/power4/fpu/s_llrintf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/s_llrintf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power4/fpu/s_llrintf.S	Thu Apr 06 10:50:35 2006
@@ -0,0 +1,39 @@
+/* Round float to long int.  PowerPC32 on PowerPC64 version.
+   Copyright (C) 2004, 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA
+   02110-1301 USA.  */
+
+#include <sysdep.h>
+
+/* long long int[r3, r4] __llrintf (float x[fp1])  */
+ENTRY (__llrintf)	
+	CALL_MCOUNT
+	stwu	r1,-16(r1)
+	cfi_adjust_cfa_offset (16)
+	fctid	fp13,fp1
+	stfd	fp13,8(r1)
+	nop	/* Insure the following load is in a different dispatch group */
+	nop	/* to avoid pipe stall on POWER4&5.  */
+	nop
+	lwz	r3,8(r1)
+	lwz	r4,12(r1)
+	addi	r1,r1,16	
+	blr
+	END (__llrintf)
+
+weak_alias (__llrintf, llrintf)
+
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/s_llround.S libc25/sysdeps/powerpc/powerpc32/power4/fpu/s_llround.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/s_llround.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power4/fpu/s_llround.S	Tue Oct 10 09:39:01 2006
@@ -0,0 +1,97 @@
+/* llround function.  PowerPC32 on PowerPC64 version.
+   Copyright (C) 2004, 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 1 Franklin Street, Fifth Floor, Boston MA
+   02110-1301 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+	.section	.rodata.cst8,"aM",@progbits,8
+	.align	2
+.LC0:	/* 0.0 */
+	.long 0x00000000
+.LC1:	/* 0.5 */
+	.long 0x3f000000
+
+	.section	".text"
+	
+/* long [r3] lround (float x [fp1])
+   IEEE 1003.1 lround function.  IEEE specifies "round to the nearest 
+   integer value, rounding halfway cases away from zero, regardless of
+   the current rounding mode."  However PowerPC Architecture defines
+   "round to Nearest" as "Choose the best approximation. In case of a 
+   tie, choose the one that is even (least significant bit o).". 
+   So we can't use the PowerPC "round to Nearest" mode. Instead we set
+   "round toward Zero" mode and round by adding +-0.5 before rounding
+   to the integer value.  */
+
+ENTRY (__llround)
+	stwu    r1,-16(r1)
+	cfi_adjust_cfa_offset (16)
+#ifdef SHARED
+	mflr	r11
+	cfi_register(lr,r11)
+# ifdef HAVE_ASM_PPC_REL16
+	bcl	20,31,1f
+1:	mflr	r9
+	addis	r9,r9,.LC0-1b@ha
+	addi	r9,r9,.LC0-1b@l
+# else
+	bl	_GLOBAL_OFFSET_TABLE_@local-4
+	mflr	r10
+	lwz	r9,.LC0@got(10)
+# endif
+	mtlr	r11
+	cfi_same_value (lr)
+	lfs	fp12,0(r9)
+	lfs	fp10,.LC1-.LC0(r9)
+#else
+	lis	r9,.LC0@ha
+	lis	r10,.LC1@ha
+	lfs	fp12,.LC0@l(r9)
+	lfs	fp10,.LC1@l(r10)
+#endif
+	fcmpu	cr6,fp1,fp12	/* if (x > 0.0)  */
+	ble-	cr6,.L4
+	fadd	fp1,fp1,fp10	/* x+= 0.5;  */
+.L9:
+	fctidz	fp2,fp1		/* Convert To Integer DW round toward 0.  */
+	stfd	fp2,8(r1)
+	nop	/* Ensure the following load is in a different dispatch  */
+	nop	/* group to avoid pipe stall on POWER4&5.  */
+	nop
+	lwz	r4,12(r1)
+	lwz	r3,8(r1)
+	addi	r1,r1,16
+	blr
+.L4:
+	fsub	fp1,fp1,fp10	/* x-= 0.5;  */
+	b	.L9
+	END (__llround)
+
+weak_alias (__llround, llround)
+
+strong_alias (__llround, __llroundf)
+weak_alias (__llround, llroundf)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__llround, llroundl)
+strong_alias (__llround, __llroundl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __llround, llroundl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/s_llroundf.S libc25/sysdeps/powerpc/powerpc32/power4/fpu/s_llroundf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/s_llroundf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power4/fpu/s_llroundf.S	Tue Oct 10 09:39:01 2006
@@ -0,0 +1,1 @@
+/* __llroundf is in s_llround.S */
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/slowexp.c libc25/sysdeps/powerpc/powerpc32/power4/fpu/slowexp.c
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/slowexp.c	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power4/fpu/slowexp.c	Mon May 21 11:46:25 2007
@@ -0,0 +1,66 @@
+/*
+ * IBM Accurate Mathematical Library
+ * written by International Business Machines Corp.
+ * Copyright (C) 2001, 2007 Free Software Foundation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU Lesser General Public License as published by
+ * the Free Software Foundation; either version 2.1 of the License, or
+ * (at your option) any later version.
+ *
+ * This program 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 Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+/**************************************************************************/
+/*  MODULE_NAME:slowexp.c                                                 */
+/*                                                                        */
+/*  FUNCTION:slowexp                                                      */
+/*                                                                        */
+/*  FILES NEEDED:mpa.h                                                    */
+/*               mpa.c mpexp.c                                            */
+/*                                                                        */
+/*Converting from double precision to Multi-precision and calculating     */
+/* e^x                                                                    */
+/**************************************************************************/
+#include "math_private.h"
+
+#ifdef NO_LONG_DOUBLE
+#include "mpa.h"
+void __mpexp(mp_no *x, mp_no *y, int p);
+#endif
+
+/*Converting from double precision to Multi-precision and calculating  e^x */
+double __slowexp(double x) {
+#ifdef NO_LONG_DOUBLE
+  double w,z,res,eps=3.0e-26;
+  int p;
+  mp_no mpx, mpy, mpz,mpw,mpeps,mpcor;
+
+  p=6;
+  __dbl_mp(x,&mpx,p); /* Convert a double precision number  x               */
+                    /* into a multiple precision number mpx with prec. p. */
+  __mpexp(&mpx, &mpy, p); /* Multi-Precision exponential function */
+  __dbl_mp(eps,&mpeps,p);
+  __mul(&mpeps,&mpy,&mpcor,p);
+  __add(&mpy,&mpcor,&mpw,p);
+  __sub(&mpy,&mpcor,&mpz,p);
+  __mp_dbl(&mpw, &w, p);
+  __mp_dbl(&mpz, &z, p);
+  if (w == z) return w;
+  else  {                   /* if calculating is not exactly   */
+    p = 32;
+    __dbl_mp(x,&mpx,p);
+    __mpexp(&mpx, &mpy, p);
+    __mp_dbl(&mpy, &res, p);
+    return res;
+  }
+#else
+  return (double) __ieee754_expl((long double)x);
+#endif
+}
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/slowpow.c libc25/sysdeps/powerpc/powerpc32/power4/fpu/slowpow.c
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/slowpow.c	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power4/fpu/slowpow.c	Fri Oct 20 15:30:05 2006
@@ -0,0 +1,94 @@
+/*
+ * IBM Accurate Mathematical Library
+ * written by International Business Machines Corp.
+ * Copyright (C) 2001, 2006 Free Software Foundation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU Lesser General Public License as published by
+ * the Free Software Foundation; either version 2.1 of the License, or
+ * (at your option) any later version.
+ *
+ * This program 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 Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+/*************************************************************************/
+/* MODULE_NAME:slowpow.c                                                 */
+/*                                                                       */
+/* FUNCTION:slowpow                                                      */
+/*                                                                       */
+/*FILES NEEDED:mpa.h                                                     */
+/*             mpa.c mpexp.c mplog.c halfulp.c                           */
+/*                                                                       */
+/* Given two IEEE double machine numbers y,x , routine  computes the     */
+/* correctly  rounded (to nearest) value of x^y. Result calculated  by   */
+/* multiplication (in halfulp.c) or if result isn't accurate enough      */
+/* then routine converts x and y into multi-precision doubles and        */
+/* recompute.                                                            */
+/*************************************************************************/
+
+#include "mpa.h"
+#include "math_private.h"
+
+void __mpexp (mp_no * x, mp_no * y, int p);
+void __mplog (mp_no * x, mp_no * y, int p);
+double ulog (double);
+double __halfulp (double x, double y);
+
+double
+__slowpow (double x, double y, double z)
+{
+  double res, res1;
+  long double ldw, ldz, ldpp;
+  static const long double ldeps = 0x4.0p-96;
+
+  res = __halfulp (x, y);	/* halfulp() returns -10 or x^y             */
+  if (res >= 0)
+    return res;			/* if result was really computed by halfulp */
+  /*  else, if result was not really computed by halfulp */
+
+  /* Compute pow as long double, 106 bits */
+  ldz = __ieee754_logl ((long double) x);
+  ldw = (long double) y *ldz;
+  ldpp = __ieee754_expl (ldw);
+  res = (double) (ldpp + ldeps);
+  res1 = (double) (ldpp - ldeps);
+
+  if (res != res1)		/* if result still not accurate enough */
+    {				/* use mpa for higher persision.  */
+      mp_no mpx, mpy, mpz, mpw, mpp, mpr, mpr1;
+      static const mp_no eps = { -3, {1.0, 4.0} };
+      int p;
+
+      p = 10;			/*  p=precision 240 bits  */
+      __dbl_mp (x, &mpx, p);
+      __dbl_mp (y, &mpy, p);
+      __dbl_mp (z, &mpz, p);
+      __mplog (&mpx, &mpz, p);		/* log(x) = z   */
+      __mul (&mpy, &mpz, &mpw, p);	/*  y * z =w    */
+      __mpexp (&mpw, &mpp, p);		/*  e^w =pp     */
+      __add (&mpp, &eps, &mpr, p);	/*  pp+eps =r   */
+      __mp_dbl (&mpr, &res, p);
+      __sub (&mpp, &eps, &mpr1, p);	/*  pp -eps =r1 */
+      __mp_dbl (&mpr1, &res1, p);	/*  converting into double precision */
+      if (res == res1)
+	return res;
+
+      /* if we get here result wasn't calculated exactly, continue for
+         more exact calculation using 768 bits.  */
+      p = 32;
+      __dbl_mp (x, &mpx, p);
+      __dbl_mp (y, &mpy, p);
+      __dbl_mp (z, &mpz, p);
+      __mplog (&mpx, &mpz, p);		/* log(c)=z  */
+      __mul (&mpy, &mpz, &mpw, p);	/* y*z =w    */
+      __mpexp (&mpw, &mpp, p);		/* e^w=pp    */
+      __mp_dbl (&mpp, &res, p);		/* converting into double precision */
+    }
+  return res;
+}
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/w_sqrt.c libc25/sysdeps/powerpc/powerpc32/power4/fpu/w_sqrt.c
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/w_sqrt.c	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power4/fpu/w_sqrt.c	Fri Jun 01 10:16:26 2007
@@ -0,0 +1,62 @@
+/* Double-precision floating point square root wrapper.
+   Copyright (C) 2004, 2007 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <math_ldbl_opt.h>
+#include "math.h"
+#include "math_private.h"
+#include <fenv_libc.h>
+
+#ifdef __STDC__
+double
+__sqrt (double x)		/* wrapper sqrt */
+#else
+double
+__sqrt (x)			/* wrapper sqrt */
+     double x;
+#endif
+{
+  double z;
+/* Power4 (ISA V2.0) and above implement sqrt in hardware.  */
+   __asm __volatile (
+	"	fsqrt	%0,%1\n"
+		: "=f" (z)
+		: "f" (x));
+#ifdef _IEEE_LIBM
+  return z;
+#else
+  if (__builtin_expect (_LIB_VERSION == _IEEE_, 0))
+    return z;
+    
+  if (__builtin_expect (x != x, 0))
+    return z;
+
+  if  (__builtin_expect (x < 0.0, 0))
+    return __kernel_standard (x, x, 26);	/* sqrt(negative) */
+  else
+    return z;
+#endif
+}
+
+weak_alias (__sqrt, sqrt)
+#ifdef NO_LONG_DOUBLE
+  strong_alias (__sqrt, __sqrtl) weak_alias (__sqrt, sqrtl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_0)
+compat_symbol (libm, __sqrt, sqrtl, GLIBC_2_0);
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/w_sqrtf.c libc25/sysdeps/powerpc/powerpc32/power4/fpu/w_sqrtf.c
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power4/fpu/w_sqrtf.c	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power4/fpu/w_sqrtf.c	Fri Jun 01 10:23:29 2007
@@ -0,0 +1,60 @@
+/* Single-precision floating point square root wrapper.
+   Copyright (C) 2004, 2007 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include "math.h"
+#include "math_private.h"
+#include <fenv_libc.h>
+
+#include <sysdep.h>
+#include <ldsodefs.h>
+
+#ifdef __STDC__
+float
+__sqrtf (float x)		/* wrapper sqrtf */
+#else
+float
+__sqrtf (x)			/* wrapper sqrtf */
+     float x;
+#endif
+{
+#ifdef _IEEE_LIBM
+  return __ieee754_sqrtf (x);
+#else
+  float z;
+/* Power4 (ISA V2.0) and above implement sqrtf in hardware.  */
+   __asm __volatile (
+	"	fsqrts	%0,%1\n"
+		: "=f" (z)
+		: "f" (x));
+
+  if (__builtin_expect (_LIB_VERSION == _IEEE_, 0))
+    return z;
+    
+  if (__builtin_expect (x != x, 0))
+    return z;
+    
+  if  (__builtin_expect (x < 0.0, 0))
+    /* sqrtf(negative) */
+    return (float) __kernel_standard ((double) x, (double) x, 126);
+  else
+    return z;
+#endif
+}
+
+weak_alias (__sqrtf, sqrtf)
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/Implies libc25/sysdeps/powerpc/powerpc32/power5+/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/Implies	Thu May 31 08:20:34 2007
@@ -0,0 +1,1 @@
+powerpc/powerpc32/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_ceil.S libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_ceil.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_ceil.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_ceil.S	Thu Jun 15 13:29:18 2006
@@ -0,0 +1,37 @@
+/* ceil function.  PowerPC32/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+	.machine	"power5"
+EALIGN (__ceil, 4, 0)
+	frip	fp1, fp1
+	blr
+	END (__ceil)
+
+weak_alias (__ceil, ceil)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__ceil, ceill)
+strong_alias (__ceil, __ceill)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_0)
+compat_symbol (libm, __ceil, ceill, GLIBC_2_0)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_ceilf.S libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_ceilf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_ceilf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_ceilf.S	Thu Jun 15 13:29:08 2006
@@ -0,0 +1,30 @@
+/* ceilf function.  PowerPC32/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+	.machine	"power5"
+EALIGN (__ceilf, 4, 0)
+	frip	fp1, fp1	/* The rounding instructions are double.  */
+	frsp	fp1, fp1	/* But we need to set ooverflow for float.  */
+	blr
+	END (__ceilf)
+
+weak_alias (__ceilf, ceilf)
+
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_floor.S libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_floor.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_floor.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_floor.S	Thu Jun 15 13:28:56 2006
@@ -0,0 +1,37 @@
+/* floor function.  PowerPC32/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+	.machine	"power5"
+EALIGN (__floor, 4, 0)
+	frim	fp1, fp1
+	blr
+	END (__floor)
+
+weak_alias (__floor, floor)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__floor, floorl)
+strong_alias (__floor, __floorl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_0)
+compat_symbol (libm, __floor, floorl, GLIBC_2_0)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_floorf.S libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_floorf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_floorf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_floorf.S	Thu Jun 15 13:28:47 2006
@@ -0,0 +1,30 @@
+/* floorf function.  PowerPC32/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+	.machine	"power5"
+EALIGN (__floorf, 4, 0)
+	frim	fp1, fp1	/* The rounding instructions are double.  */
+	frsp	fp1, fp1	/* But we need to set ooverflow for float.  */
+	blr
+	END (__floorf)
+
+weak_alias (__floorf, floorf)
+
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_llround.S libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_llround.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_llround.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_llround.S	Mon Oct 09 14:00:22 2006
@@ -0,0 +1,60 @@
+/* lround function.  POWER5+, PowerPC32 version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 1 Franklin Street, Fifth Floor, Boston MA
+   02110-1301 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+	
+/* long [r3] llround (float x [fp1])
+   IEEE 1003.1 lround function.  IEEE specifies "round to the nearest 
+   integer value, rounding halfway cases away from zero, regardless of
+   the current rounding mode."  However PowerPC Architecture defines
+   "round to Nearest" as "Choose the best approximation. In case of a 
+   tie, choose the one that is even (least significant bit o).". 
+   So we pre-round using the V2.02 Floating Round to Integer Nearest
+   instruction before we use the Floating Convert to Integer Word with
+   round to zero instruction.  */
+
+	.machine	"power5"
+ENTRY (__llround)
+	stwu    r1,-16(r1)
+	cfi_adjust_cfa_offset (16)
+	frin	fp2,fp1
+	fctidz	fp3,fp2		/* Convert To Integer Word lround toward 0.  */
+	stfd	fp3,8(r1)
+	nop	/* Ensure the following load is in a different dispatch  */
+	nop	/* group to avoid pipe stall on POWER4&5.  */
+	nop
+	lwz	r4,12(r1)
+	lwz	r3,8(r1)
+	addi	r1,r1,16
+	blr
+	END (__llround)
+
+weak_alias (__llround, llround)
+
+strong_alias (__llround, __llroundf)
+weak_alias (__llround, llroundf)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__llround, llroundl)
+strong_alias (__llround, __llroundl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __llround, llroundl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_llroundf.S libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_llroundf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_llroundf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_llroundf.S	Mon Oct 09 14:00:22 2006
@@ -0,0 +1,2 @@
+/* __llroundf is in s_llround.S  */
+/* __llroundf is in s_llround.S  */
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_lround.S libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_lround.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_lround.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_lround.S	Mon Oct 09 14:00:22 2006
@@ -0,0 +1,58 @@
+/* lround function.  POWER5+, PowerPC32 version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 1 Franklin Street, Fifth Floor, Boston MA
+   02110-1301 USA.  */
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+	
+/* long [r3] lround (float x [fp1])
+   IEEE 1003.1 lround function.  IEEE specifies "round to the nearest 
+   integer value, rounding halfway cases away from zero, regardless of
+   the current rounding mode."  However PowerPC Architecture defines
+   "round to Nearest" as "Choose the best approximation. In case of a 
+   tie, choose the one that is even (least significant bit o).". 
+   So we pre-round using the V2.02 Floating Round to Integer Nearest
+   instruction before we use the Floating Convert to Integer Word with
+   round to zero instruction.  */
+
+	.machine	"power5"
+ENTRY (__lround)
+	stwu    r1,-16(r1)
+	cfi_adjust_cfa_offset (16)
+	frin	fp2,fp1
+	fctiwz	fp3,fp2		/* Convert To Integer Word lround toward 0.  */
+	stfd	fp3,8(r1)
+	nop	/* Ensure the following load is in a different dispatch  */
+	nop	/* group to avoid pipe stall on POWER4&5.  */
+	nop
+	lwz	r3,12(r1)
+	addi	r1,r1,16
+	blr
+	END (__lround)
+
+weak_alias (__lround, lround)
+
+strong_alias (__lround, __lroundf)
+weak_alias (__lround, lroundf)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__lround, lroundl)
+strong_alias (__lround, __lroundl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __lround, lroundl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_round.S libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_round.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_round.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_round.S	Tue May 22 17:57:07 2007
@@ -0,0 +1,37 @@
+/* round function.  PowerPC32/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+	.machine	"power5"
+EALIGN (__round, 4, 0)
+	frin	fp1, fp1
+	blr
+	END (__round)
+
+weak_alias (__round, round)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__round, roundl)
+strong_alias (__round, __roundl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __round, roundl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_roundf.S libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_roundf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_roundf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_roundf.S	Thu Jun 15 13:29:47 2006
@@ -0,0 +1,30 @@
+/* roundf function.  PowerPC32/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+	.machine	"power5"
+EALIGN (__roundf, 4, 0)
+	frin	fp1, fp1	/* The rounding instructions are double.  */
+	frsp	fp1, fp1	/* But we need to set ooverflow for float.  */
+	blr
+	END (__roundf)
+
+weak_alias (__roundf, roundf)
+
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_trunc.S libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_trunc.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_trunc.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_trunc.S	Tue May 22 17:58:40 2007
@@ -0,0 +1,37 @@
+/* trunc function.  PowerPC32/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+	.machine	"power5"
+EALIGN (__trunc, 4, 0)
+	friz	fp1, fp1
+	blr
+	END (__trunc)
+
+weak_alias (__trunc, trunc)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__trunc, truncl)
+strong_alias (__trunc, __truncl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __trunc, truncl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_truncf.S libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_truncf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5+/fpu/s_truncf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5+/fpu/s_truncf.S	Thu Jun 15 13:30:33 2006
@@ -0,0 +1,30 @@
+/* truncf function.  PowerPC32/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+	.machine	"power5"
+EALIGN (__truncf, 4, 0)
+	friz	fp1, fp1	/* The rounding instructions are double.  */
+	frsp	fp1, fp1	/* But we need to set ooverflow for float.  */
+	blr
+	END (__truncf)
+
+weak_alias (__truncf, truncf)
+
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5/fpu/Implies libc25/sysdeps/powerpc/powerpc32/power5/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power5/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power5/fpu/Implies	Thu May 31 08:20:16 2007
@@ -0,0 +1,1 @@
+powerpc/powerpc32/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6/fpu/Implies libc25/sysdeps/powerpc/powerpc32/power6/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power6/fpu/Implies	Thu May 31 08:20:49 2007
@@ -0,0 +1,2 @@
+powerpc/powerpc32/power5+/fpu
+powerpc/powerpc32/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6/fpu/s_llrint.S libc25/sysdeps/powerpc/powerpc32/power6/fpu/s_llrint.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6/fpu/s_llrint.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power6/fpu/s_llrint.S	Tue May 22 17:43:36 2007
@@ -0,0 +1,46 @@
+/* Round double to long int.  PowerPC32 on PowerPC64 version.
+   Copyright (C) 2004, 2006, 2007 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA
+   02110-1301 USA.  */
+
+#include <sysdep.h>
+
+/* long long int[r3, r4] __llrint (double x[fp1])  */
+ENTRY (__llrint)	
+	CALL_MCOUNT
+	stwu	r1,-16(r1)
+	cfi_adjust_cfa_offset (16)
+	fctid	fp13,fp1
+	stfd	fp13,8(r1)
+/* Insure the following load is in a different dispatch group by
+   inserting "group ending nop".  */
+	ori	r1,r1,0
+	lwz	r3,8(r1)
+	lwz	r4,12(r1)
+	addi	r1,r1,16	
+	blr
+	END (__llrint)
+
+weak_alias (__llrint, llrint)
+
+#ifdef NO_LONG_DOUBLE
+strong_alias (__llrint, __llrintl)
+weak_alias (__llrint, llrintl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __llrint, llrintl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6/fpu/s_llrintf.S libc25/sysdeps/powerpc/powerpc32/power6/fpu/s_llrintf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6/fpu/s_llrintf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power6/fpu/s_llrintf.S	Thu May 10 11:28:20 2007
@@ -0,0 +1,39 @@
+/* Round float to long int.  PowerPC32 on PowerPC64 version.
+   Copyright (C) 2004, 2006, 2007 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA
+   02110-1301 USA.  */
+
+#include <sysdep.h>
+
+/* long long int[r3, r4] __llrintf (float x[fp1])  */
+ENTRY (__llrintf)	
+	CALL_MCOUNT
+	stwu	r1,-16(r1)
+	cfi_adjust_cfa_offset (16)
+	fctid	fp13,fp1
+	stfd	fp13,8(r1)
+/* Insure the following load is in a different dispatch group by
+   inserting "group ending nop".  */
+	ori	r1,r1,0
+	lwz	r3,8(r1)
+	lwz	r4,12(r1)
+	addi	r1,r1,16	
+	blr
+	END (__llrintf)
+
+weak_alias (__llrintf, llrintf)
+
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6/fpu/s_llround.S libc25/sysdeps/powerpc/powerpc32/power6/fpu/s_llround.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6/fpu/s_llround.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power6/fpu/s_llround.S	Thu May 10 11:28:20 2007
@@ -0,0 +1,60 @@
+/* lround function.  POWER5+, PowerPC32 version.
+   Copyright (C) 2006, 2007 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 1 Franklin Street, Fifth Floor, Boston MA
+   02110-1301 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+	
+/* long [r3] llround (float x [fp1])
+   IEEE 1003.1 lround function.  IEEE specifies "round to the nearest 
+   integer value, rounding halfway cases away from zero, regardless of
+   the current rounding mode."  However PowerPC Architecture defines
+   "round to Nearest" as "Choose the best approximation. In case of a 
+   tie, choose the one that is even (least significant bit o).". 
+   So we pre-round using the V2.02 Floating Round to Integer Nearest
+   instruction before we use the Floating Convert to Integer Word with
+   round to zero instruction.  */
+
+	.machine	"power5"
+ENTRY (__llround)
+	stwu    r1,-16(r1)
+	cfi_adjust_cfa_offset (16)
+	frin	fp2,fp1
+	fctidz	fp3,fp2		/* Convert To Integer Word lround toward 0.  */
+	stfd	fp3,8(r1)
+/* Insure the following load is in a different dispatch group by
+   inserting "group ending nop".  */
+	ori	r1,r1,0
+	lwz	r4,12(r1)
+	lwz	r3,8(r1)
+	addi	r1,r1,16
+	blr
+	END (__llround)
+
+weak_alias (__llround, llround)
+
+strong_alias (__llround, __llroundf)
+weak_alias (__llround, llroundf)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__llround, llroundl)
+strong_alias (__llround, __llroundl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __llround, llroundl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6/fpu/s_llroundf.S libc25/sysdeps/powerpc/powerpc32/power6/fpu/s_llroundf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6/fpu/s_llroundf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power6/fpu/s_llroundf.S	Thu May 10 11:28:20 2007
@@ -0,0 +1,2 @@
+/* __llroundf is in s_llround.S  */
+/* __llroundf is in s_llround.S  */
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6x/fpu/Implies libc25/sysdeps/powerpc/powerpc32/power6x/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6x/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power6x/fpu/Implies	Thu May 31 08:21:03 2007
@@ -0,0 +1,3 @@
+powerpc/powerpc32/power6/fpu
+powerpc/powerpc32/power5+/fpu
+powerpc/powerpc32/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6x/fpu/s_lrint.S libc25/sysdeps/powerpc/powerpc32/power6x/fpu/s_lrint.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6x/fpu/s_lrint.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power6x/fpu/s_lrint.S	Mon Oct 09 14:00:56 2006
@@ -0,0 +1,42 @@
+/* Round double to long int.  POWER6x PowerPC32 version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+	.machine	"power6"
+/* long int[r3] __lrint (double x[fp1])  */
+ENTRY (__lrint)
+	fctiw	fp13,fp1
+	mftgpr  r3,fp13
+	blr
+	END (__lrint)
+
+weak_alias (__lrint, lrint)
+
+strong_alias (__lrint, __lrintf)
+weak_alias (__lrint, lrintf)
+
+#ifdef NO_LONG_DOUBLE
+strong_alias (__lrint, __lrintl)
+weak_alias (__lrint, lrintl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __lrint, lrintl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6x/fpu/s_lround.S libc25/sysdeps/powerpc/powerpc32/power6x/fpu/s_lround.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc32/power6x/fpu/s_lround.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc32/power6x/fpu/s_lround.S	Mon Oct 09 14:00:56 2006
@@ -0,0 +1,52 @@
+/* lround function.  POWER6x, PowerPC32 version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 1 Franklin Street, Fifth Floor, Boston MA
+   02110-1301 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+	
+/* long [r3] lround (float x [fp1])
+   IEEE 1003.1 lround function.  IEEE specifies "round to the nearest 
+   integer value, rounding halfway cases away from zero, regardless of
+   the current rounding mode."  However PowerPC Architecture defines
+   "round to Nearest" as "Choose the best approximation. In case of a 
+   tie, choose the one that is even (least significant bit o).". 
+   So we pre-round using the V2.02 Floating Round to Integer Nearest
+   instruction before we use the Floating Convert to Integer Word with
+   round to zero instruction.  */
+
+	.machine	"power6"
+ENTRY (__lround)
+	frin	fp2,fp1	/* Pre-round +-0.5.  */
+	fctiwz	fp3,fp2	/* Convert To Integer Word lround toward 0.  */
+	mftgpr	r3,fp3	/* Transfer fpr3 to r3.  */
+	blr
+	END (__lround)
+
+weak_alias (__lround, lround)
+
+strong_alias (__lround, __lroundf)
+weak_alias (__lround, lroundf)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__lround, lroundl)
+strong_alias (__lround, __lroundl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __lround, lroundl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/970/fpu/Implies libc25/sysdeps/powerpc/powerpc64/970/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/970/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/970/fpu/Implies	Sun Aug 20 09:06:20 2006
@@ -0,0 +1,1 @@
+powerpc/powerpc64/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/Makefile libc25/sysdeps/powerpc/powerpc64/power4/fpu/Makefile
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/Makefile	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power4/fpu/Makefile	Thu Aug 10 12:49:45 2006
@@ -0,0 +1,5 @@
+# Makefile fragment for POWER4/5/5+ platforms with FPU.
+
+ifeq ($(subdir),math)
+CFLAGS-mpa.c += --param max-unroll-times=4 -funroll-loops -fpeel-loops -ftree-loop-linear 
+endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/mpa.c libc25/sysdeps/powerpc/powerpc64/power4/fpu/mpa.c
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/mpa.c	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power4/fpu/mpa.c	Sun Aug 20 08:59:40 2006
@@ -0,0 +1,549 @@
+
+/*
+ * IBM Accurate Mathematical Library
+ * written by International Business Machines Corp.
+ * Copyright (C) 2001, 2006 Free Software Foundation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU Lesser General Public License as published by
+ * the Free Software Foundation; either version 2.1 of the License, or
+ * (at your option) any later version.
+ *
+ * This program 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 Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+/************************************************************************/
+/*  MODULE_NAME: mpa.c                                                  */
+/*                                                                      */
+/*  FUNCTIONS:                                                          */
+/*               mcr                                                    */
+/*               acr                                                    */
+/*               cr                                                     */
+/*               cpy                                                    */
+/*               cpymn                                                  */
+/*               norm                                                   */
+/*               denorm                                                 */
+/*               mp_dbl                                                 */
+/*               dbl_mp                                                 */
+/*               add_magnitudes                                         */
+/*               sub_magnitudes                                         */
+/*               add                                                    */
+/*               sub                                                    */
+/*               mul                                                    */
+/*               inv                                                    */
+/*               dvd                                                    */
+/*                                                                      */
+/* Arithmetic functions for multiple precision numbers.                 */
+/* Relative errors are bounded                                          */
+/************************************************************************/
+
+
+#include "endian.h"
+#include "mpa.h"
+#include "mpa2.h"
+#include <sys/param.h>	/* For MIN() */
+/* mcr() compares the sizes of the mantissas of two multiple precision  */
+/* numbers. Mantissas are compared regardless of the signs of the       */
+/* numbers, even if x->d[0] or y->d[0] are zero. Exponents are also     */
+/* disregarded.                                                         */
+static int mcr(const mp_no *x, const mp_no *y, int p) {
+  long i;
+  long p2 = p;
+  for (i=1; i<=p2; i++) {
+    if      (X[i] == Y[i])  continue;
+    else if (X[i] >  Y[i])  return  1;
+    else                    return -1; }
+  return 0;
+}
+
+
+
+/* acr() compares the absolute values of two multiple precision numbers */
+int __acr(const mp_no *x, const mp_no *y, int p) {
+  long i;
+
+  if      (X[0] == ZERO) {
+    if    (Y[0] == ZERO) i= 0;
+    else                 i=-1;
+  }
+  else if (Y[0] == ZERO) i= 1;
+  else {
+    if      (EX >  EY)   i= 1;
+    else if (EX <  EY)   i=-1;
+    else                 i= mcr(x,y,p);
+  }
+
+  return i;
+}
+
+
+/* cr90 compares the values of two multiple precision numbers           */
+int  __cr(const mp_no *x, const mp_no *y, int p) {
+  int i;
+
+  if      (X[0] > Y[0])  i= 1;
+  else if (X[0] < Y[0])  i=-1;
+  else if (X[0] < ZERO ) i= __acr(y,x,p);
+  else                   i= __acr(x,y,p);
+
+  return i;
+}
+
+
+/* Copy a multiple precision number. Set *y=*x. x=y is permissible.      */
+void __cpy(const mp_no *x, mp_no *y, int p) {
+  long i;
+
+  EY = EX;
+  for (i=0; i <= p; i++)    Y[i] = X[i];
+
+  return;
+}
+
+
+/* Copy a multiple precision number x of precision m into a */
+/* multiple precision number y of precision n. In case n>m, */
+/* the digits of y beyond the m'th are set to zero. In case */
+/* n<m, the digits of x beyond the n'th are ignored.        */
+/* x=y is permissible.                                      */
+
+void __cpymn(const mp_no *x, int m, mp_no *y, int n) {
+
+  long i,k;
+  long n2 = n;
+  long m2 = m;
+
+  EY = EX;     k=MIN(m2,n2);
+  for (i=0; i <= k; i++)    Y[i] = X[i];
+  for (   ; i <= n2; i++)    Y[i] = ZERO;
+
+  return;
+}
+
+/* Convert a multiple precision number *x into a double precision */
+/* number *y, normalized case  (|x| >= 2**(-1022))) */
+static void norm(const mp_no *x, double *y, int p)
+{
+  #define R  radixi.d
+  long i;
+#if 0
+  int k;
+#endif
+  double a,c,u,v,z[5];
+  if (p<5) {
+    if      (p==1) c = X[1];
+    else if (p==2) c = X[1] + R* X[2];
+    else if (p==3) c = X[1] + R*(X[2]  +   R* X[3]);
+    else if (p==4) c =(X[1] + R* X[2]) + R*R*(X[3] + R*X[4]);
+  }
+  else {
+    for (a=ONE, z[1]=X[1]; z[1] < TWO23; )
+        {a *= TWO;   z[1] *= TWO; }
+
+    for (i=2; i<5; i++) {
+      z[i] = X[i]*a;
+      u = (z[i] + CUTTER)-CUTTER;
+      if  (u > z[i])  u -= RADIX;
+      z[i] -= u;
+      z[i-1] += u*RADIXI;
+    }
+
+    u = (z[3] + TWO71) - TWO71;
+    if (u > z[3])   u -= TWO19;
+    v = z[3]-u;
+
+    if (v == TWO18) {
+      if (z[4] == ZERO) {
+        for (i=5; i <= p; i++) {
+          if (X[i] == ZERO)   continue;
+          else                {z[3] += ONE;   break; }
+        }
+      }
+      else              z[3] += ONE;
+    }
+
+    c = (z[1] + R *(z[2] + R * z[3]))/a;
+  }
+
+  c *= X[0];
+
+  for (i=1; i<EX; i++)   c *= RADIX;
+  for (i=1; i>EX; i--)   c *= RADIXI;
+
+  *y = c;
+  return;
+#undef R
+}
+
+/* Convert a multiple precision number *x into a double precision */
+/* number *y, denormalized case  (|x| < 2**(-1022))) */
+static void denorm(const mp_no *x, double *y, int p)
+{
+  long i,k;
+  long p2 = p;
+  double c,u,z[5];
+#if 0
+  double a,v;
+#endif
+
+#define R  radixi.d
+  if (EX<-44 || (EX==-44 && X[1]<TWO5))
+     { *y=ZERO; return; }
+
+  if      (p2==1) {
+    if      (EX==-42) {z[1]=X[1]+TWO10;  z[2]=ZERO;  z[3]=ZERO;  k=3;}
+    else if (EX==-43) {z[1]=     TWO10;  z[2]=X[1];  z[3]=ZERO;  k=2;}
+    else              {z[1]=     TWO10;  z[2]=ZERO;  z[3]=X[1];  k=1;}
+  }
+  else if (p2==2) {
+    if      (EX==-42) {z[1]=X[1]+TWO10;  z[2]=X[2];  z[3]=ZERO;  k=3;}
+    else if (EX==-43) {z[1]=     TWO10;  z[2]=X[1];  z[3]=X[2];  k=2;}
+    else              {z[1]=     TWO10;  z[2]=ZERO;  z[3]=X[1];  k=1;}
+  }
+  else {
+    if      (EX==-42) {z[1]=X[1]+TWO10;  z[2]=X[2];  k=3;}
+    else if (EX==-43) {z[1]=     TWO10;  z[2]=X[1];  k=2;}
+    else              {z[1]=     TWO10;  z[2]=ZERO;  k=1;}
+    z[3] = X[k];
+  }
+
+  u = (z[3] + TWO57) - TWO57;
+  if  (u > z[3])   u -= TWO5;
+
+  if (u==z[3]) {
+    for (i=k+1; i <= p2; i++) {
+      if (X[i] == ZERO)   continue;
+      else {z[3] += ONE;   break; }
+    }
+  }
+
+  c = X[0]*((z[1] + R*(z[2] + R*z[3])) - TWO10);
+
+  *y = c*TWOM1032;
+  return;
+
+#undef R
+}
+
+/* Convert a multiple precision number *x into a double precision number *y. */
+/* The result is correctly rounded to the nearest/even. *x is left unchanged */
+
+void __mp_dbl(const mp_no *x, double *y, int p) {
+#if 0
+  int i,k;
+  double a,c,u,v,z[5];
+#endif
+
+  if (X[0] == ZERO)  {*y = ZERO;  return; }
+
+  if      (EX> -42)                 norm(x,y,p);
+  else if (EX==-42 && X[1]>=TWO10)  norm(x,y,p);
+  else                              denorm(x,y,p);
+}
+
+
+/* dbl_mp() converts a double precision number x into a multiple precision  */
+/* number *y. If the precision p is too small the result is truncated. x is */
+/* left unchanged.                                                          */
+
+void __dbl_mp(double x, mp_no *y, int p) {
+
+  long i,n;
+  long p2 = p;
+  double u;
+
+  /* Sign */
+  if      (x == ZERO)  {Y[0] = ZERO;  return; }
+  else if (x >  ZERO)   Y[0] = ONE;
+  else                 {Y[0] = MONE;  x=-x;   }
+
+  /* Exponent */
+  for (EY=ONE; x >= RADIX; EY += ONE)   x *= RADIXI;
+  for (      ; x <  ONE;   EY -= ONE)   x *= RADIX;
+
+  /* Digits */
+  n=MIN(p2,4);
+  for (i=1; i<=n; i++) {
+    u = (x + TWO52) - TWO52;
+    if (u>x)   u -= ONE;
+    Y[i] = u;     x -= u;    x *= RADIX; }
+  for (   ; i<=p2; i++)     Y[i] = ZERO;
+  return;
+}
+
+
+/*  add_magnitudes() adds the magnitudes of *x & *y assuming that           */
+/*  abs(*x) >= abs(*y) > 0.                                                 */
+/* The sign of the sum *z is undefined. x&y may overlap but not x&z or y&z. */
+/* No guard digit is used. The result equals the exact sum, truncated.      */
+/* *x & *y are left unchanged.                                              */
+
+static void add_magnitudes(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  long i,j,k;
+  long p2 = p;
+
+  EZ = EX;
+
+  i=p2;    j=p2+ EY - EX;    k=p2+1;
+
+  if (j<1)
+     {__cpy(x,z,p);  return; }
+  else   Z[k] = ZERO;
+
+  for (; j>0; i--,j--) {
+    Z[k] += X[i] + Y[j];
+    if (Z[k] >= RADIX) {
+      Z[k]  -= RADIX;
+      Z[--k] = ONE; }
+    else
+      Z[--k] = ZERO;
+  }
+
+  for (; i>0; i--) {
+    Z[k] += X[i];
+    if (Z[k] >= RADIX) {
+      Z[k]  -= RADIX;
+      Z[--k] = ONE; }
+    else
+      Z[--k] = ZERO;
+  }
+
+  if (Z[1] == ZERO) {
+    for (i=1; i<=p2; i++)    Z[i] = Z[i+1]; }
+  else   EZ += ONE;
+}
+
+
+/*  sub_magnitudes() subtracts the magnitudes of *x & *y assuming that      */
+/*  abs(*x) > abs(*y) > 0.                                                  */
+/* The sign of the difference *z is undefined. x&y may overlap but not x&z  */
+/* or y&z. One guard digit is used. The error is less than one ulp.         */
+/* *x & *y are left unchanged.                                              */
+
+static void sub_magnitudes(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  long i,j,k;
+  long p2 = p;
+
+  EZ = EX;
+
+  if (EX == EY) {
+    i=j=k=p2;
+    Z[k] = Z[k+1] = ZERO; }
+  else {
+    j= EX - EY;
+    if (j > p2)  {__cpy(x,z,p);  return; }
+    else {
+      i=p2;   j=p2+1-j;   k=p2;
+      if (Y[j] > ZERO) {
+        Z[k+1] = RADIX - Y[j--];
+        Z[k]   = MONE; }
+      else {
+        Z[k+1] = ZERO;
+        Z[k]   = ZERO;   j--;}
+    }
+  }
+
+  for (; j>0; i--,j--) {
+    Z[k] += (X[i] - Y[j]);
+    if (Z[k] < ZERO) {
+      Z[k]  += RADIX;
+      Z[--k] = MONE; }
+    else
+      Z[--k] = ZERO;
+  }
+
+  for (; i>0; i--) {
+    Z[k] += X[i];
+    if (Z[k] < ZERO) {
+      Z[k]  += RADIX;
+      Z[--k] = MONE; }
+    else
+      Z[--k] = ZERO;
+  }
+
+  for (i=1; Z[i] == ZERO; i++) ;
+  EZ = EZ - i + 1;
+  for (k=1; i <= p2+1; )
+    Z[k++] = Z[i++];
+  for (; k <= p2; )
+    Z[k++] = ZERO;
+
+  return;
+}
+
+
+/* Add two multiple precision numbers. Set *z = *x + *y. x&y may overlap  */
+/* but not x&z or y&z. One guard digit is used. The error is less than    */
+/* one ulp. *x & *y are left unchanged.                                   */
+
+void __add(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  int n;
+
+  if      (X[0] == ZERO)     {__cpy(y,z,p);  return; }
+  else if (Y[0] == ZERO)     {__cpy(x,z,p);  return; }
+
+  if (X[0] == Y[0])   {
+    if (__acr(x,y,p) > 0)      {add_magnitudes(x,y,z,p);  Z[0] = X[0]; }
+    else                     {add_magnitudes(y,x,z,p);  Z[0] = Y[0]; }
+  }
+  else                       {
+    if ((n=__acr(x,y,p)) == 1) {sub_magnitudes(x,y,z,p);  Z[0] = X[0]; }
+    else if (n == -1)        {sub_magnitudes(y,x,z,p);  Z[0] = Y[0]; }
+    else                      Z[0] = ZERO;
+  }
+  return;
+}
+
+
+/* Subtract two multiple precision numbers. *z is set to *x - *y. x&y may */
+/* overlap but not x&z or y&z. One guard digit is used. The error is      */
+/* less than one ulp. *x & *y are left unchanged.                         */
+
+void __sub(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  int n;
+
+  if      (X[0] == ZERO)     {__cpy(y,z,p);  Z[0] = -Z[0];  return; }
+  else if (Y[0] == ZERO)     {__cpy(x,z,p);                 return; }
+
+  if (X[0] != Y[0])    {
+    if (__acr(x,y,p) > 0)      {add_magnitudes(x,y,z,p);  Z[0] =  X[0]; }
+    else                     {add_magnitudes(y,x,z,p);  Z[0] = -Y[0]; }
+  }
+  else                       {
+    if ((n=__acr(x,y,p)) == 1) {sub_magnitudes(x,y,z,p);  Z[0] =  X[0]; }
+    else if (n == -1)        {sub_magnitudes(y,x,z,p);  Z[0] = -Y[0]; }
+    else                      Z[0] = ZERO;
+  }
+  return;
+}
+
+
+/* Multiply two multiple precision numbers. *z is set to *x * *y. x&y      */
+/* may overlap but not x&z or y&z. In case p=1,2,3 the exact result is     */
+/* truncated to p digits. In case p>3 the error is bounded by 1.001 ulp.   */
+/* *x & *y are left unchanged.                                             */
+
+void __mul(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  long i, i1, i2, j, k, k2;
+  long p2 = p;
+  double u, zk, zk2;
+
+                      /* Is z=0? */
+  if (X[0]*Y[0]==ZERO)
+     { Z[0]=ZERO;  return; }
+
+                       /* Multiply, add and carry */
+  k2 = (p2<3) ? p2+p2 : p2+3;
+  zk = Z[k2]=ZERO;
+  for (k=k2; k>1; ) {
+    if (k > p2)  {i1=k-p2; i2=p2+1; }
+    else        {i1=1;   i2=k;   }
+#if 1
+    /* rearange this inner loop to allow the fmadd instructions to be
+       independent and execute in parallel on processors that have
+       dual symetrical FP pipelines.  */
+    if (i1 < (i2-1))
+    {
+	/* make sure we have at least 2 iterations */
+	if (((i2 - i1) & 1L) == 1L)
+	{
+                /* Handle the odd iterations case.  */
+		zk2 = x->d[i2-1]*y->d[i1];
+	}
+	else
+		zk2 = zero.d;
+	/* Do two multiply/adds per loop iteration, using independent
+           accumulators; zk and zk2.  */
+	for (i=i1,j=i2-1; i<i2-1; i+=2,j-=2) 
+	{
+		zk += x->d[i]*y->d[j];
+		zk2 += x->d[i+1]*y->d[j-1];
+	}
+	zk += zk2; /* final sum.  */
+    }
+    else
+    {
+        /* Special case when iterations is 1.  */
+	zk += x->d[i1]*y->d[i1];
+    }
+#else
+    /* The orginal code.  */
+    for (i=i1,j=i2-1; i<i2; i++,j--)  zk += X[i]*Y[j];
+#endif
+
+    u = (zk + CUTTER)-CUTTER;
+    if  (u > zk)  u -= RADIX;
+    Z[k]  = zk - u;
+    zk = u*RADIXI;
+    --k;
+  }
+  Z[k] = zk;
+
+                 /* Is there a carry beyond the most significant digit? */
+  if (Z[1] == ZERO) {
+    for (i=1; i<=p2; i++)  Z[i]=Z[i+1];
+    EZ = EX + EY - 1; }
+  else
+    EZ = EX + EY;
+
+  Z[0] = X[0] * Y[0];
+  return;
+}
+
+
+/* Invert a multiple precision number. Set *y = 1 / *x.                     */
+/* Relative error bound = 1.001*r**(1-p) for p=2, 1.063*r**(1-p) for p=3,   */
+/* 2.001*r**(1-p) for p>3.                                                  */
+/* *x=0 is not permissible. *x is left unchanged.                           */
+
+void __inv(const mp_no *x, mp_no *y, int p) {
+  long i;
+#if 0
+  int l;
+#endif
+  double t;
+  mp_no z,w;
+  static const int np1[] = {0,0,0,0,1,2,2,2,2,3,3,3,3,3,3,3,3,3,
+                            4,4,4,4,4,4,4,4,4,4,4,4,4,4,4};
+  const mp_no mptwo = {1,{1.0,2.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,
+                         0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,
+                         0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,
+                         0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0}};
+
+  __cpy(x,&z,p);  z.e=0;  __mp_dbl(&z,&t,p);
+  t=ONE/t;   __dbl_mp(t,y,p);    EY -= EX;
+
+  for (i=0; i<np1[p]; i++) {
+    __cpy(y,&w,p);
+    __mul(x,&w,y,p);
+    __sub(&mptwo,y,&z,p);
+    __mul(&w,&z,y,p);
+  }
+  return;
+}
+
+
+/* Divide one multiple precision number by another.Set *z = *x / *y. *x & *y */
+/* are left unchanged. x&y may overlap but not x&z or y&z.                   */
+/* Relative error bound = 2.001*r**(1-p) for p=2, 2.063*r**(1-p) for p=3     */
+/* and 3.001*r**(1-p) for p>3. *y=0 is not permissible.                      */
+
+void __dvd(const mp_no *x, const mp_no *y, mp_no *z, int p) {
+
+  mp_no w;
+
+  if (X[0] == ZERO)    Z[0] = ZERO;
+  else                {__inv(y,&w,p);   __mul(x,&w,z,p);}
+  return;
+}
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/slowexp.c libc25/sysdeps/powerpc/powerpc64/power4/fpu/slowexp.c
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/slowexp.c	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power4/fpu/slowexp.c	Mon May 21 11:44:09 2007
@@ -0,0 +1,66 @@
+/*
+ * IBM Accurate Mathematical Library
+ * written by International Business Machines Corp.
+ * Copyright (C) 2001, 2007 Free Software Foundation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU Lesser General Public License as published by
+ * the Free Software Foundation; either version 2.1 of the License, or
+ * (at your option) any later version.
+ *
+ * This program 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 Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+/**************************************************************************/
+/*  MODULE_NAME:slowexp.c                                                 */
+/*                                                                        */
+/*  FUNCTION:slowexp                                                      */
+/*                                                                        */
+/*  FILES NEEDED:mpa.h                                                    */
+/*               mpa.c mpexp.c                                            */
+/*                                                                        */
+/*Converting from double precision to Multi-precision and calculating     */
+/* e^x                                                                    */
+/**************************************************************************/
+#include "math_private.h"
+
+#ifdef NO_LONG_DOUBLE
+#include "mpa.h"
+void __mpexp(mp_no *x, mp_no *y, int p);
+#endif
+
+/*Converting from double precision to Multi-precision and calculating  e^x */
+double __slowexp(double x) {
+#ifdef NO_LONG_DOUBLE
+  double w,z,res,eps=3.0e-26;
+  int p;
+  mp_no mpx, mpy, mpz,mpw,mpeps,mpcor;
+
+  p=6;
+  __dbl_mp(x,&mpx,p); /* Convert a double precision number  x               */
+                    /* into a multiple precision number mpx with prec. p. */
+  __mpexp(&mpx, &mpy, p); /* Multi-Precision exponential function */
+  __dbl_mp(eps,&mpeps,p);
+  __mul(&mpeps,&mpy,&mpcor,p);
+  __add(&mpy,&mpcor,&mpw,p);
+  __sub(&mpy,&mpcor,&mpz,p);
+  __mp_dbl(&mpw, &w, p);
+  __mp_dbl(&mpz, &z, p);
+  if (w == z) return w;
+  else  {                   /* if calculating is not exactly   */
+    p = 32;
+    __dbl_mp(x,&mpx,p);
+    __mpexp(&mpx, &mpy, p);
+    __mp_dbl(&mpy, &res, p);
+    return res;
+  }
+#else
+  return (double) __ieee754_expl((long double)x);
+#endif
+}
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/slowpow.c libc25/sysdeps/powerpc/powerpc64/power4/fpu/slowpow.c
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/slowpow.c	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power4/fpu/slowpow.c	Fri Oct 20 15:24:56 2006
@@ -0,0 +1,94 @@
+/*
+ * IBM Accurate Mathematical Library
+ * written by International Business Machines Corp.
+ * Copyright (C) 2001, 2006 Free Software Foundation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU Lesser General Public License as published by
+ * the Free Software Foundation; either version 2.1 of the License, or
+ * (at your option) any later version.
+ *
+ * This program 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 Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ */
+/*************************************************************************/
+/* MODULE_NAME:slowpow.c                                                 */
+/*                                                                       */
+/* FUNCTION:slowpow                                                      */
+/*                                                                       */
+/*FILES NEEDED:mpa.h                                                     */
+/*             mpa.c mpexp.c mplog.c halfulp.c                           */
+/*                                                                       */
+/* Given two IEEE double machine numbers y,x , routine  computes the     */
+/* correctly  rounded (to nearest) value of x^y. Result calculated  by   */
+/* multiplication (in halfulp.c) or if result isn't accurate enough      */
+/* then routine converts x and y into multi-precision doubles and        */
+/* recompute.                                                            */
+/*************************************************************************/
+
+#include "mpa.h"
+#include "math_private.h"
+
+void __mpexp (mp_no * x, mp_no * y, int p);
+void __mplog (mp_no * x, mp_no * y, int p);
+double ulog (double);
+double __halfulp (double x, double y);
+
+double
+__slowpow (double x, double y, double z)
+{
+  double res, res1;
+  long double ldw, ldz, ldpp;
+  static const long double ldeps = 0x4.0p-96;
+
+  res = __halfulp (x, y);	/* halfulp() returns -10 or x^y             */
+  if (res >= 0)
+    return res;			/* if result was really computed by halfulp */
+  /*  else, if result was not really computed by halfulp */
+
+  /* Compute pow as long double, 106 bits */
+  ldz = __ieee754_logl ((long double) x);
+  ldw = (long double) y *ldz;
+  ldpp = __ieee754_expl (ldw);
+  res = (double) (ldpp + ldeps);
+  res1 = (double) (ldpp - ldeps);
+
+  if (res != res1)		/* if result still not accurate enough */
+    {				/* use mpa for higher persision.  */
+      mp_no mpx, mpy, mpz, mpw, mpp, mpr, mpr1;
+      static const mp_no eps = { -3, {1.0, 4.0} };
+      int p;
+
+      p = 10;			/*  p=precision 240 bits  */
+      __dbl_mp (x, &mpx, p);
+      __dbl_mp (y, &mpy, p);
+      __dbl_mp (z, &mpz, p);
+      __mplog (&mpx, &mpz, p);		/* log(x) = z   */
+      __mul (&mpy, &mpz, &mpw, p);	/*  y * z =w    */
+      __mpexp (&mpw, &mpp, p);		/*  e^w =pp     */
+      __add (&mpp, &eps, &mpr, p);	/*  pp+eps =r   */
+      __mp_dbl (&mpr, &res, p);
+      __sub (&mpp, &eps, &mpr1, p);	/*  pp -eps =r1 */
+      __mp_dbl (&mpr1, &res1, p);	/*  converting into double precision */
+      if (res == res1)
+	return res;
+
+      /* if we get here result wasn't calculated exactly, continue for
+         more exact calculation using 768 bits.  */
+      p = 32;
+      __dbl_mp (x, &mpx, p);
+      __dbl_mp (y, &mpy, p);
+      __dbl_mp (z, &mpz, p);
+      __mplog (&mpx, &mpz, p);		/* log(c)=z  */
+      __mul (&mpy, &mpz, &mpw, p);	/* y*z =w    */
+      __mpexp (&mpw, &mpp, p);		/* e^w=pp    */
+      __mp_dbl (&mpp, &res, p);		/* converting into double precision */
+    }
+  return res;
+}
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/w_sqrt.c libc25/sysdeps/powerpc/powerpc64/power4/fpu/w_sqrt.c
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/w_sqrt.c	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power4/fpu/w_sqrt.c	Fri Jun 01 10:18:46 2007
@@ -0,0 +1,62 @@
+/* Double-precision floating point square root wrapper.
+   Copyright (C) 2004, 2007 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <math_ldbl_opt.h>
+#include "math.h"
+#include "math_private.h"
+#include <fenv_libc.h>
+
+#ifdef __STDC__
+double
+__sqrt (double x)		/* wrapper sqrt */
+#else
+double
+__sqrt (x)			/* wrapper sqrt */
+     double x;
+#endif
+{
+  double z;
+/* Power4 (ISA V2.0) and above implement sqrt in hardware.  */
+   __asm __volatile (
+	"	fsqrt	%0,%1\n"
+		: "=f" (z)
+		: "f" (x));
+#ifdef _IEEE_LIBM
+  return z;
+#else
+  if (__builtin_expect (_LIB_VERSION == _IEEE_, 0))
+    return z;
+    
+  if (__builtin_expect (x != x, 0))
+    return z;
+    
+  if  (__builtin_expect (x < 0.0, 0))
+    return __kernel_standard (x, x, 26);	/* sqrt(negative) */
+  else
+    return z;
+#endif
+}
+
+weak_alias (__sqrt, sqrt)
+#ifdef NO_LONG_DOUBLE
+  strong_alias (__sqrt, __sqrtl) weak_alias (__sqrt, sqrtl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_0)
+compat_symbol (libm, __sqrt, sqrtl, GLIBC_2_0);
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/w_sqrtf.c libc25/sysdeps/powerpc/powerpc64/power4/fpu/w_sqrtf.c
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power4/fpu/w_sqrtf.c	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power4/fpu/w_sqrtf.c	Fri Jun 01 10:23:54 2007
@@ -0,0 +1,60 @@
+/* Single-precision floating point square root wrapper.
+   Copyright (C) 2004, 2007 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include "math.h"
+#include "math_private.h"
+#include <fenv_libc.h>
+
+#include <sysdep.h>
+#include <ldsodefs.h>
+
+#ifdef __STDC__
+float
+__sqrtf (float x)		/* wrapper sqrtf */
+#else
+float
+__sqrtf (x)			/* wrapper sqrtf */
+     float x;
+#endif
+{
+#ifdef _IEEE_LIBM
+  return __ieee754_sqrtf (x);
+#else
+  float z;
+/* Power4 (ISA V2.0) and above implement sqrtf in hardware.  */
+   __asm __volatile (
+	"	fsqrts	%0,%1\n"
+		: "=f" (z)
+		: "f" (x));
+
+  if (__builtin_expect (_LIB_VERSION == _IEEE_, 0))
+    return z;
+    
+  if (__builtin_expect (x != x, 0))
+    return z;
+    
+  if  (__builtin_expect (x < 0.0, 0))
+    /* sqrtf(negative) */
+    return (float) __kernel_standard ((double) x, (double) x, 126);
+  else
+    return z;
+#endif
+}
+
+weak_alias (__sqrtf, sqrtf)
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/Implies libc25/sysdeps/powerpc/powerpc64/power5+/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power5+/fpu/Implies	Sun Aug 20 09:05:08 2006
@@ -0,0 +1,1 @@
+powerpc/powerpc64/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_ceil.S libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_ceil.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_ceil.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_ceil.S	Thu Jun 15 11:45:17 2006
@@ -0,0 +1,38 @@
+/* ceil function.  PowerPC64/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+	.machine	"power5"
+EALIGN (__ceil, 4, 0)
+	CALL_MCOUNT 0
+	frip	fp1, fp1
+	blr
+	END (__ceil)
+
+weak_alias (__ceil, ceil)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__ceil, ceill)
+strong_alias (__ceil, __ceill)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_0)
+compat_symbol (libm, __ceil, ceill, GLIBC_2_0)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_ceilf.S libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_ceilf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_ceilf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_ceilf.S	Thu Jun 15 11:45:38 2006
@@ -0,0 +1,31 @@
+/* ceilf function.  PowerPC64/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+	.machine	"power5"
+EALIGN (__ceilf, 4, 0)
+	CALL_MCOUNT 0
+	frip	fp1, fp1	/* The rounding instructions are double.  */
+	frsp	fp1, fp1	/* But we need to set ooverflow for float.  */
+	blr
+	END (__ceilf)
+
+weak_alias (__ceilf, ceilf)
+
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_floor.S libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_floor.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_floor.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_floor.S	Thu Jun 15 11:45:07 2006
@@ -0,0 +1,38 @@
+/* floor function.  PowerPC64/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+	.machine	"power5"
+EALIGN (__floor, 4, 0)
+	CALL_MCOUNT 0
+	frim	fp1, fp1
+	blr
+	END (__floor)
+
+weak_alias (__floor, floor)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__floor, floorl)
+strong_alias (__floor, __floorl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_0)
+compat_symbol (libm, __floor, floorl, GLIBC_2_0)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_floorf.S libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_floorf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_floorf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_floorf.S	Thu Jun 15 11:44:52 2006
@@ -0,0 +1,31 @@
+/* floorf function.  PowerPC64/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+	.machine	"power5"
+EALIGN (__floorf, 4, 0)
+	CALL_MCOUNT 0
+	frim	fp1, fp1	/* The rounding instructions are double.  */
+	frsp	fp1, fp1	/* But we need to set ooverflow for float.  */
+	blr
+	END (__floorf)
+
+weak_alias (__floorf, floorf)
+
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_llround.S libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_llround.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_llround.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_llround.S	Mon Oct 09 14:00:22 2006
@@ -0,0 +1,59 @@
+/* llround function.  POWER5+, PowerPC64 version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+	
+/* long long [r3] llround (float x [fp1])
+   IEEE 1003.1 llround function.  IEEE specifies "round to the nearest 
+   integer value, rounding halfway cases away from zero, regardless of
+   the current rounding mode."  However PowerPC Architecture defines
+   "round to Nearest" as "Choose the best approximation. In case of a 
+   tie, choose the one that is even (least significant bit o).". 
+   So we pre-round using the V2.02 Floating Round to Integer Nearest
+   instruction before we use Floating Convert to Integer Word with
+   round to zero instruction.  */
+
+	.machine	"power5"
+EALIGN (__llround, 4, 0)
+	CALL_MCOUNT 0
+	frin	fp2, fp1	/* Round to nearest +-0.5.  */	
+	fctidz	fp3, fp2	/* Convert To Integer DW round toward 0.  */
+	stfd	fp3, -16(r1)
+	nop	/* Insure the following load is in a different dispatch group */
+	nop	/* to avoid pipe stall on POWER4&5.  */
+	nop
+	ld	r3, -16(r1)
+	blr
+	END (__llround)
+
+strong_alias (__llround, __lround)
+weak_alias (__llround, llround)
+weak_alias (__lround, lround)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__llround, llroundl)
+strong_alias (__llround, __llroundl)
+weak_alias (__lround, lroundl)
+strong_alias (__lround, __lroundl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __llround, llroundl, GLIBC_2_1)
+compat_symbol (libm, __lround, lroundl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_round.S libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_round.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_round.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_round.S	Tue May 22 17:56:44 2007
@@ -0,0 +1,38 @@
+/* round function.  PowerPC64/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+	.machine	"power5"
+EALIGN (__round, 4, 0)
+	CALL_MCOUNT 0
+	frin	fp1, fp1
+	blr
+	END (__round)
+
+weak_alias (__round, round)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__round, roundl)
+strong_alias (__round, __roundl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __round, roundl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_roundf.S libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_roundf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_roundf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_roundf.S	Thu Jun 15 11:43:56 2006
@@ -0,0 +1,31 @@
+/* roundf function.  PowerPC64/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+	.machine	"power5"
+EALIGN (__roundf, 4, 0)
+	CALL_MCOUNT 0
+	frin	fp1, fp1	/* The rounding instructions are double.  */
+	frsp	fp1, fp1	/* But we need to set ooverflow for float.  */
+	blr
+	END (__roundf)
+
+weak_alias (__roundf, roundf)
+
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_trunc.S libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_trunc.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_trunc.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_trunc.S	Tue May 22 17:57:49 2007
@@ -0,0 +1,38 @@
+/* trunc function.  PowerPC64/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+	.machine	"power5"
+EALIGN (__trunc, 4, 0)
+	CALL_MCOUNT 0
+	friz	fp1, fp1
+	blr
+	END (__trunc)
+
+weak_alias (__trunc, trunc)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__trunc, truncl)
+strong_alias (__trunc, __truncl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __trunc, truncl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_truncf.S libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_truncf.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5+/fpu/s_truncf.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power5+/fpu/s_truncf.S	Thu Jun 15 11:36:15 2006
@@ -0,0 +1,31 @@
+/* truncf function.  PowerPC64/power5+ version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+
+	.machine	"power5"
+EALIGN (__truncf, 4, 0)
+	CALL_MCOUNT 0
+	friz	fp1, fp1	/* The rounding instructions are double.  */
+	frsp	fp1, fp1	/* But we need to set ooverflow for float.  */
+	blr
+	END (__truncf)
+
+weak_alias (__truncf, truncf)
+
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5/fpu/Implies libc25/sysdeps/powerpc/powerpc64/power5/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power5/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power5/fpu/Implies	Thu Nov 09 14:12:01 2006
@@ -0,0 +1,1 @@
+powerpc/powerpc64/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power6/fpu/Implies libc25/sysdeps/powerpc/powerpc64/power6/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power6/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power6/fpu/Implies	Sun Aug 20 09:06:40 2006
@@ -0,0 +1,2 @@
+powerpc/powerpc64/power5+/fpu
+powerpc/powerpc64/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power6x/fpu/Implies libc25/sysdeps/powerpc/powerpc64/power6x/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power6x/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power6x/fpu/Implies	Mon Oct 09 14:00:22 2006
@@ -0,0 +1,3 @@
+powerpc/powerpc64/power6/fpu
+powerpc/powerpc64/power5+/fpu
+powerpc/powerpc64/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power6x/fpu/s_llrint.S libc25/sysdeps/powerpc/powerpc64/power6x/fpu/s_llrint.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power6x/fpu/s_llrint.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power6x/fpu/s_llrint.S	Mon Oct 09 14:00:56 2006
@@ -0,0 +1,45 @@
+/* Round double to long int.  POWER6x PowerPC64 version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+
+	.machine	"power6"
+/* long long int[r3] __llrint (double x[fp1])  */
+ENTRY (__llrint)	
+	CALL_MCOUNT 0
+	fctid	fp13,fp1
+	mftgpr  r3,fp13
+	blr
+	END (__llrint)
+
+strong_alias (__llrint, __lrint)
+weak_alias (__llrint, llrint)
+weak_alias (__lrint, lrint)
+
+#ifdef NO_LONG_DOUBLE
+strong_alias (__llrint, __llrintl)
+weak_alias (__llrint, llrintl)
+strong_alias (__lrint, __lrintl)
+weak_alias (__lrint, lrintl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __llrint, llrintl, GLIBC_2_1)
+compat_symbol (libm, __lrint, lrintl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power6x/fpu/s_llround.S libc25/sysdeps/powerpc/powerpc64/power6x/fpu/s_llround.S
--- libc25-cvstip-20070510/sysdeps/powerpc/powerpc64/power6x/fpu/s_llround.S	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/powerpc/powerpc64/power6x/fpu/s_llround.S	Mon Oct 09 14:00:56 2006
@@ -0,0 +1,55 @@
+/* llround function.  POWER6x PowerPC64 version.
+   Copyright (C) 2006 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 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
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <sysdep.h>
+#include <math_ldbl_opt.h>
+	
+/* long long [r3] llround (float x [fp1])
+   IEEE 1003.1 llround function.  IEEE specifies "round to the nearest 
+   integer value, rounding halfway cases away from zero, regardless of
+   the current rounding mode."  However PowerPC Architecture defines
+   "round to Nearest" as "Choose the best approximation. In case of a 
+   tie, choose the one that is even (least significant bit o).". 
+   So we pre-round using the V2.02 Floating Round to Integer Nearest
+   instruction before we use Floating Convert to Integer Word with
+   round to zero instruction.  */
+
+	.machine	"power6"
+ENTRY (__llround)
+	CALL_MCOUNT 0
+	frin	fp2,fp1	/* Round to nearest +-0.5.  */	
+	fctidz	fp3,fp2	/* Convert To Integer DW round toward 0.  */
+	mftgpr	r3,fp3  /* Transfer integer to R3.  */
+	blr
+	END (__llround)
+
+strong_alias (__llround, __lround)
+weak_alias (__llround, llround)
+weak_alias (__lround, lround)
+
+#ifdef NO_LONG_DOUBLE
+weak_alias (__llround, llroundl)
+strong_alias (__llround, __llroundl)
+weak_alias (__lround, lroundl)
+strong_alias (__lround, __lroundl)
+#endif
+#if LONG_DOUBLE_COMPAT(libm, GLIBC_2_1)
+compat_symbol (libm, __llround, llroundl, GLIBC_2_1)
+compat_symbol (libm, __lround, lroundl, GLIBC_2_1)
+#endif
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/970/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/970/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/970/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/970/fpu/Implies	Thu Apr 06 10:50:35 2006
@@ -0,0 +1,3 @@
+# Make sure this comes before the powerpc/powerpc32/fpu that's
+# listed in unix/sysv/linux/powerpc/powerpc32/fpu/Implies.
+powerpc/powerpc32/970/fpu
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/power4/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/power4/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/power4/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/power4/fpu/Implies	Thu Apr 06 10:50:35 2006
@@ -0,0 +1,3 @@
+# Make sure this comes before the powerpc/powerpc32/fpu that's
+# listed in unix/sysv/linux/powerpc/powerpc32/fpu/Implies.
+powerpc/powerpc32/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/power5+/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/power5+/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/power5+/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/power5+/fpu/Implies	Thu Apr 06 10:50:35 2006
@@ -0,0 +1,3 @@
+# Make sure this comes before the powerpc/powerpc32/fpu that's
+# listed in unix/sysv/linux/powerpc/powerpc32/fpu/Implies.
+powerpc/powerpc32/power5+/fpu
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/power5/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/power5/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/power5/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/power5/fpu/Implies	Thu Apr 06 10:50:35 2006
@@ -0,0 +1,3 @@
+# Make sure this comes before the powerpc/powerpc32/fpu that's
+# listed in unix/sysv/linux/powerpc/powerpc32/fpu/Implies.
+powerpc/powerpc32/power5/fpu
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/power6/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/power6/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/power6/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/power6/fpu/Implies	Mon Jun 19 15:25:23 2006
@@ -0,0 +1,3 @@
+# Make sure this comes before the powerpc/powerpc32/fpu that's
+# listed in unix/sysv/linux/powerpc/powerpc32/fpu/Implies.
+powerpc/powerpc32/power6/fpu
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/power6x/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/power6x/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc32/power6x/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc32/power6x/fpu/Implies	Mon Oct 09 14:00:22 2006
@@ -0,0 +1,4 @@
+# Make sure this comes before the powerpc/powerpc32/fpu that's
+# listed in unix/sysv/linux/powerpc/powerpc32/fpu/Implies.
+powerpc/powerpc32/power6x/fpu
+powerpc/powerpc32/power6/fpu
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/970/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/970/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/970/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/970/fpu/Implies	Sun Aug 20 09:13:50 2006
@@ -0,0 +1,1 @@
+powerpc/powerpc64/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/power4/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/power4/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/power4/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/power4/fpu/Implies	Sun Aug 20 09:14:01 2006
@@ -0,0 +1,1 @@
+powerpc/powerpc64/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/power5+/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/power5+/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/power5+/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/power5+/fpu/Implies	Thu Jun 15 08:44:40 2006
@@ -0,0 +1,3 @@
+# Make sure this comes before the powerpc/powerpc64/fpu that's
+# listed in unix/sysv/linux/powerpc/powerpc64/fpu/Implies.
+powerpc/powerpc64/power5+/fpu
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/power5/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/power5/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/power5/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/power5/fpu/Implies	Sun Aug 20 09:14:05 2006
@@ -0,0 +1,1 @@
+powerpc/powerpc64/power4/fpu
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/power6/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/power6/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/power6/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/power6/fpu/Implies	Mon Jun 19 14:51:14 2006
@@ -0,0 +1,3 @@
+# Make sure this comes before the powerpc/powerpc64/fpu that's
+# listed in unix/sysv/linux/powerpc/powerpc64/fpu/Implies.
+powerpc/powerpc64/power6/fpu
diff -urN libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/power6x/fpu/Implies libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/power6x/fpu/Implies
--- libc25-cvstip-20070510/sysdeps/unix/sysv/linux/powerpc/powerpc64/power6x/fpu/Implies	Wed Dec 31 18:00:00 1969
+++ libc25/sysdeps/unix/sysv/linux/powerpc/powerpc64/power6x/fpu/Implies	Mon Oct 09 14:00:22 2006
@@ -0,0 +1,4 @@
+# Make sure this comes before the powerpc/powerpc64/fpu that's
+# listed in unix/sysv/linux/powerpc/powerpc64/fpu/Implies.
+powerpc/powerpc64/power6x/fpu
+powerpc/powerpc64/power6/fpu

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]