Cleanup FMA4 patch

Move the FMA4 code into its own section.  Avoid some of the duplication
of data resulting from the double use of source files.

47 files changed, 494 insertions, 198 deletions

diff --git a/ChangeLog b/ChangeLog
index 992ad4cb2a..2383ec8229 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,53 @@
+2011-10-25  Ulrich Drepper  <drepper@gmail.com>
+
+	* sysdeps/ieee754/dbl-64/branred.c: Move FMA4 code into separate
+	.text section.  Avoid duplicate constants.
+	* sysdeps/ieee754/dbl-64/doasin.c: Likewise.
+	* sysdeps/ieee754/dbl-64/dosincos.c: Likewise.
+	* sysdeps/ieee754/dbl-64/e_asin.c: Likewise.
+	* sysdeps/ieee754/dbl-64/e_atan2.c: Likewise.
+	* sysdeps/ieee754/dbl-64/e_exp.c: Likewise.
+	* sysdeps/ieee754/dbl-64/e_log.c: Likewise.
+	* sysdeps/ieee754/dbl-64/e_pow.c: Likewise.
+	* sysdeps/ieee754/dbl-64/halfulp.c: Likewise.
+	* sysdeps/ieee754/dbl-64/mpa.c: Likewise.
+	* sysdeps/ieee754/dbl-64/mpa.h: Likewise.
+	* sysdeps/ieee754/dbl-64/mpatan.c: Likewise.
+	* sysdeps/ieee754/dbl-64/mpatan.h: Likewise.
+	* sysdeps/ieee754/dbl-64/mpatan2.c: Likewise.
+	* sysdeps/ieee754/dbl-64/mpexp.c: Likewise.
+	* sysdeps/ieee754/dbl-64/mpexp.h: Likewise.
+	* sysdeps/ieee754/dbl-64/mpsqrt.c: Likewise.
+	* sysdeps/ieee754/dbl-64/mpsqrt.h: Likewise.
+	* sysdeps/ieee754/dbl-64/mptan.c: Likewise.
+	* sysdeps/ieee754/dbl-64/s_sin.c: Likewise.
+	* sysdeps/ieee754/dbl-64/s_tan.c: Likewise.
+	* sysdeps/ieee754/dbl-64/sincos32.c: Likewise.
+	* sysdeps/ieee754/dbl-64/slowexp.c: Likewise.
+	* sysdeps/ieee754/dbl-64/slowpow.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/brandred-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/doasin-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/dosincos-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/e_asin-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/e_atan2-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/e_exp-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/e_log-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/e_pow-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/halfulp-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/mpa-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/mpatan-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/mpatan2-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/mpexp-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/mplog-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/mpsqrt-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/mptan-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/s_atan-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/s_sin-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/s_tan-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/sincos32-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/slowexp-fma4.c: Likewise.
+	* sysdeps/x86_64/fpu/multiarch/slowpow-fma4.c: Likewise.
+
 2011-10-24  Ulrich Drepper  <drepper@gmail.com>
 
 	* sysdeps/x86_64/dla.h: Move to ...
diff --git a/sysdeps/ieee754/dbl-64/branred.c b/sysdeps/ieee754/dbl-64/branred.c
index 76015f0c5c..c8483034af 100644
--- a/sysdeps/ieee754/dbl-64/branred.c
+++ b/sysdeps/ieee754/dbl-64/branred.c
@@ -1,7 +1,7 @@
 /*
  * IBM Accurate Mathematical Library
  * Written by International Business Machines Corp.
- * Copyright (C) 2001 Free Software Foundation, Inc.
+ * Copyright (C) 2001, 2011 Free Software Foundation, Inc.
  *
  * 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
@@ -38,6 +38,10 @@
 #include "branred.h"
 #include "math_private.h"
 
+#ifndef SECTION
+# define SECTION
+#endif
+
 
 /*******************************************************************/
 /* Routine  branred() performs range  reduction of a double number */
@@ -45,7 +49,9 @@
 /* x=n*pi/2+(a+aa), abs(a+aa)<pi/4, n=0,+-1,+-2,....               */
 /* Routine return integer (n mod 4)                                */
 /*******************************************************************/
-int __branred(double x, double *a, double *aa)
+int
+SECTION
+__branred(double x, double *a, double *aa)
 {
   int i,k;
 #if 0
diff --git a/sysdeps/ieee754/dbl-64/doasin.c b/sysdeps/ieee754/dbl-64/doasin.c
index 64abc3cbb1..14958b5ca2 100644
--- a/sysdeps/ieee754/dbl-64/doasin.c
+++ b/sysdeps/ieee754/dbl-64/doasin.c
@@ -34,11 +34,17 @@
 #include <dla.h>
 #include "math_private.h"
 
+#ifndef SECTION
+# define SECTION
+#endif
+
 /********************************************************************/
 /* Compute arcsin(x,dx,v) of double-length number (x+dx) the result */
 /* stored in v where v= v[0]+v[1] =arcsin(x+dx)                     */
 /********************************************************************/
-void __doasin(double x, double dx, double v[]) {
+void
+SECTION
+__doasin(double x, double dx, double v[]) {
 
 #include "doasin.h"
 
diff --git a/sysdeps/ieee754/dbl-64/dosincos.c b/sysdeps/ieee754/dbl-64/dosincos.c
index 712d585b9e..e8890ff8de 100644
--- a/sysdeps/ieee754/dbl-64/dosincos.c
+++ b/sysdeps/ieee754/dbl-64/dosincos.c
@@ -39,6 +39,10 @@
 #include "dosincos.h"
 #include "math_private.h"
 
+#ifndef SECTION
+# define SECTION
+#endif
+
 extern const union
 {
   int4 i[880];
@@ -52,7 +56,9 @@ extern const union
 /*(x+dx) between 0 and PI/4                                            */
 /***********************************************************************/
 
-void __dubsin(double x, double dx, double v[]) {
+void
+SECTION
+__dubsin(double x, double dx, double v[]) {
   double r,s,c,cc,d,dd,d2,dd2,e,ee,
     sn,ssn,cs,ccs,ds,dss,dc,dcc;
 #ifndef DLA_FMS
@@ -106,7 +112,9 @@ void __dubsin(double x, double dx, double v[]) {
 /*(x+dx) between 0 and PI/4                                           */
 /**********************************************************************/
 
-void __dubcos(double x, double dx, double v[]) {
+void
+SECTION
+__dubcos(double x, double dx, double v[]) {
   double r,s,c,cc,d,dd,d2,dd2,e,ee,
     sn,ssn,cs,ccs,ds,dss,dc,dcc;
 #ifndef DLA_FMS
@@ -172,7 +180,9 @@ void __dubcos(double x, double dx, double v[]) {
 /* Routine receive Double-Length number (x+dx) and computes cos(x+dx) */
 /* as Double-Length number and store it in array v                    */
 /**********************************************************************/
-void __docos(double x, double dx, double v[]) {
+void
+SECTION
+__docos(double x, double dx, double v[]) {
   double y,yy,p,w[2];
   if (x>0) {y=x; yy=dx;}
      else {y=-x; yy=-dx;}
diff --git a/sysdeps/ieee754/dbl-64/e_asin.c b/sysdeps/ieee754/dbl-64/e_asin.c
index cd4cc2e2c2..65319c0b58 100644
--- a/sysdeps/ieee754/dbl-64/e_asin.c
+++ b/sysdeps/ieee754/dbl-64/e_asin.c
@@ -42,6 +42,10 @@
 #include "uasncs.h"
 #include "math_private.h"
 
+#ifndef SECTION
+# define SECTION
+#endif
+
 void __doasin(double x, double dx, double w[]);
 void __dubsin(double x, double dx, double v[]);
 void __dubcos(double x, double dx, double v[]);
@@ -53,7 +57,9 @@ double __cos32(double x, double res, double res1);
 /* An ultimate asin routine. Given an IEEE double machine number x         */
 /* it computes the correctly rounded (to nearest) value of arcsin(x)       */
 /***************************************************************************/
-double __ieee754_asin(double x){
+double
+SECTION
+__ieee754_asin(double x){
   double x1,x2,xx,s1,s2,res1,p,t,res,r,cor,cc,y,c,z,w[2];
   mynumber u,v;
   int4 k,m,n;
@@ -334,7 +340,9 @@ strong_alias (__ieee754_asin, __asin_finite)
 /*                                                                 */
 /*******************************************************************/
 
-double __ieee754_acos(double x)
+double
+SECTION
+__ieee754_acos(double x)
 {
   double x1,x2,xx,s1,s2,res1,p,t,res,r,cor,cc,y,c,z,w[2],eps;
 #if 0
diff --git a/sysdeps/ieee754/dbl-64/e_atan2.c b/sysdeps/ieee754/dbl-64/e_atan2.c
index 9caacccf4c..64dae3e8d5 100644
--- a/sysdeps/ieee754/dbl-64/e_atan2.c
+++ b/sysdeps/ieee754/dbl-64/e_atan2.c
@@ -44,6 +44,10 @@
 #include "atnat2.h"
 #include "math_private.h"
 
+#ifndef SECTION
+# define SECTION
+#endif
+
 /************************************************************************/
 /* An ultimate atan2 routine. Given two IEEE double machine numbers y,x */
 /* it computes the correctly rounded (to nearest) value of atan2(y,x).  */
@@ -59,7 +63,9 @@ static double signArctan2(double y,double z)
 static double normalized(double ,double,double ,double);
 void __mpatan2(mp_no *,mp_no *,mp_no *,int);
 
-double __ieee754_atan2(double y,double x) {
+double
+SECTION
+__ieee754_atan2(double y,double x) {
 
   int i,de,ux,dx,uy,dy;
 #if 0
@@ -384,7 +390,9 @@ strong_alias (__ieee754_atan2, __atan2_finite)
 #endif
 
   /* Treat the Denormalized case */
-static double  normalized(double ax,double ay,double y, double z)
+static double
+SECTION
+normalized(double ax,double ay,double y, double z)
     { int p;
       mp_no mpx,mpy,mpz,mperr,mpz2,mpt1;
   p=6;
@@ -394,7 +402,9 @@ static double  normalized(double ax,double ay,double y, double z)
   return signArctan2(y,z);
 }
   /* Stage 3: Perform a multi-Precision computation */
-static double  atan2Mp(double x,double y,const int pr[])
+static double
+SECTION
+atan2Mp(double x,double y,const int pr[])
 {
   double z1,z2;
   int i,p;
diff --git a/sysdeps/ieee754/dbl-64/e_exp.c b/sysdeps/ieee754/dbl-64/e_exp.c
index 48bbb05ed8..e7a839d42e 100644
--- a/sysdeps/ieee754/dbl-64/e_exp.c
+++ b/sysdeps/ieee754/dbl-64/e_exp.c
@@ -40,13 +40,19 @@
 #include "uexp.tbl"
 #include "math_private.h"
 
+#ifndef SECTION
+# define SECTION
+#endif
+
 double __slowexp(double);
 
 /***************************************************************************/
 /* An ultimate exp routine. Given an IEEE double machine number x          */
 /* it computes the correctly rounded (to nearest) value of e^x             */
 /***************************************************************************/
-double __ieee754_exp(double x) {
+double
+SECTION
+__ieee754_exp(double x) {
   double bexp, t, eps, del, base, y, al, bet, res, rem, cor;
   mynumber junk1, junk2, binexp  = {{0,0}};
 #if 0
@@ -156,7 +162,9 @@ strong_alias (__ieee754_exp, __exp_finite)
 /*else return   e^(x + xx)   (always positive )                         */
 /************************************************************************/
 
-double __exp1(double x, double xx, double error) {
+double
+SECTION
+__exp1(double x, double xx, double error) {
   double bexp, t, eps, del, base, y, al, bet, res, rem, cor;
   mynumber junk1, junk2, binexp  = {{0,0}};
 #if 0
diff --git a/sysdeps/ieee754/dbl-64/e_log.c b/sysdeps/ieee754/dbl-64/e_log.c
index 7a0a26f251..e45520eba8 100644
--- a/sysdeps/ieee754/dbl-64/e_log.c
+++ b/sysdeps/ieee754/dbl-64/e_log.c
@@ -41,13 +41,19 @@
 #include "MathLib.h"
 #include "math_private.h"
 
+#ifndef SECTION
+# define SECTION
+#endif
+
 void __mplog(mp_no *, mp_no *, int);
 
 /*********************************************************************/
 /* An ultimate log routine. Given an IEEE double machine number x     */
 /* it computes the correctly rounded (to nearest) value of log(x).   */
 /*********************************************************************/
-double __ieee754_log(double x) {
+double
+SECTION
+__ieee754_log(double x) {
 #define M 4
   static const int pr[M]={8,10,18,32};
   int i,j,n,ux,dx,p;
diff --git a/sysdeps/ieee754/dbl-64/e_pow.c b/sysdeps/ieee754/dbl-64/e_pow.c
index 94b1ab8961..350e93986d 100644
--- a/sysdeps/ieee754/dbl-64/e_pow.c
+++ b/