powerpc: Fix fsqrt build in libm [BZ#16576]

Some powerpc64 processors (e5500 core for instance) does not provide the
fsqrt instruction, however current check to use in math_private.h is
__WORDSIZE and _ARCH_PWR4 (ISA 2.02).  This is patch change it to use
the compiler flag _ARCH_PPCSQ (which is the same condition GCC uses to
decide whether to generate fsqrt instruction).

It fixes BZ#16576.

(cherry picked from commit 08cee2a464f614a6d4275b5af6c52481f1aa16e6)

diff --git a/ChangeLog b/ChangeLog
index 30da116ced8820f04b8e32a25af08568900b9829..92b8a2e35cc3b44c99fd59e71ad6816e3a5377e8 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,20 @@
+2015-01-28  Adhemerval Zanellla  <azanella@linux.vnet.ibm.com>
+
+       [BZ #16576]
+       * sysdeps/powerpc/fpu/math_private.h [__CPU_HAS_FSQRT]: Remove define
+       and use _ARCH_PPCSQ instead.
+       (__ieee754_sqrt): Likewise.
+       (__ieee754_sqrtf): Likewise.
+       * sysdeps/powerpc/fpu/e_sqrt.c (__slow_ieee754_sqrt): Build only if
+       _ARCH_PPCSQ is defined.
+       (__ieee754_sqrt): Use _ARCH_PPCSQ to select wheter to use hardware
+       fsqrt instruction.
+       * sysdeps/powerpc/fpu/e_sqrtf.c (__ieee754_sqrtf): Build only if
+       _ARCH_PPCSQ is defined.
+       (__ieee754_sqrtf): Use _ARCH_PPCSQ to select wheter to use hardware
+       fsqrts instruction.
+       * sysdeps/powerpc/powerpc64/fpu/e_sqrt.c: Remove file.
+
 2016-08-15  Andreas Schwab  <schwab@suse.de>
 
        [BZ #20435]

diff --git a/NEWS b/NEWS
index a1bb8341ea795dd9f7deccb777a7237a844889aa..f62b876d81fadaa6ea3c39721b8cfb866854c972 100644 (file)
--- a/NEWS
+++ b/NEWS
@@ -9,11 +9,11 @@ Version 2.19.1
 
 * The following bugs are resolved with this release:
 
-  15946, 16009, 16545, 16574, 16623, 16657, 16695, 16743, 16758, 16759,
-  16760, 16878, 16882, 16885, 16916, 16932, 16943, 16958, 17048, 17062,
-  17069, 17079, 17137, 17153, 17213, 17263, 17269, 17325, 17523, 17555,
-  17905, 18007, 18032, 18080, 18240, 18287, 18508, 18665, 18905, 18928,
-  19018, 19779, 19791, 19879, 20010, 20112.
+  15946, 16009, 16545, 16574, 16576, 16623, 16657, 16695, 16743, 16758,
+  16759, 16760, 16878, 16882, 16885, 16916, 16932, 16943, 16958, 17048,
+  17062, 17069, 17079, 17137, 17153, 17213, 17263, 17269, 17325, 17523,
+  17555, 17905, 18007, 18032, 18080, 18240, 18287, 18508, 18665, 18905,
+  18928, 19018, 19779, 19791, 19879, 20010, 20112.
 
 * A buffer overflow in gethostbyname_r and related functions performing DNS
   requests has been fixed.  If the NSS functions were called with a

diff --git a/sysdeps/powerpc/fpu/e_sqrt.c b/sysdeps/powerpc/fpu/e_sqrt.c
index 0368e57e92ec553c9047734fa5eb5c709ec0be0b..24dfe686255fbbd0a23f1b981f1dbb2b82fe12f1 100644 (file)
--- a/sysdeps/powerpc/fpu/e_sqrt.c
+++ b/sysdeps/powerpc/fpu/e_sqrt.c
@@ -24,6 +24,7 @@
 #include <sysdep.h>
 #include <ldsodefs.h>
 
+#ifndef _ARCH_PPCSQ
 static const double almost_half = 0.5000000000000001;  /* 0.5 + 2^-53 */
 static const ieee_float_shape_type a_nan = {.word = 0x7fc00000 };
 static const ieee_float_shape_type a_inf = {.word = 0x7f800000 };
@@ -152,6 +153,7 @@ __slow_ieee754_sqrt (double x)
     }
   return f_wash (x);
 }
+#endif /* _ARCH_PPCSQ  */
 
 #undef __ieee754_sqrt
 double
@@ -159,16 +161,11 @@ __ieee754_sqrt (double x)
 {
   double z;
 
-  /* If the CPU is 64-bit we can use the optional FP instructions.  */
-  if (__CPU_HAS_FSQRT)
-    {
-      /* Volatile is required to prevent the compiler from moving the
-        fsqrt instruction above the branch.  */
-      __asm __volatile ("      fsqrt   %0,%1\n"
-                               :"=f" (z):"f" (x));
-    }
-  else
-    z = __slow_ieee754_sqrt (x);
+#ifdef _ARCH_PPCSQ
+  asm ("fsqrt %0,%1\n" :"=f" (z):"f" (x));
+#else
+  z = __slow_ieee754_sqrt (x);
+#endif
 
   return z;
 }

diff --git a/sysdeps/powerpc/fpu/e_sqrtf.c b/sysdeps/powerpc/fpu/e_sqrtf.c
index fcc74aeb1998821e6e552c91f9c26c1f7331c945..7d2bfb7b04d86d0013bd52fc4238131dcbbdbbf2 100644 (file)
--- a/sysdeps/powerpc/fpu/e_sqrtf.c
+++ b/sysdeps/powerpc/fpu/e_sqrtf.c
@@ -24,6 +24,7 @@
 #include <sysdep.h>
 #include <ldsodefs.h>
 
+#ifndef _ARCH_PPCSQ
 static const float almost_half = 0.50000006;   /* 0.5 + 2^-24 */
 static const ieee_float_shape_type a_nan = {.word = 0x7fc00000 };
 static const ieee_float_shape_type a_inf = {.word = 0x7f800000 };
@@ -128,6 +129,7 @@ __slow_ieee754_sqrtf (float x)
     }
   return f_washf (x);
 }
+#endif /* _ARCH_PPCSQ  */
 
 #undef __ieee754_sqrtf
 float
@@ -135,16 +137,11 @@ __ieee754_sqrtf (float x)
 {
   double z;
 
-  /* If the CPU is 64-bit we can use the optional FP instructions.  */
-  if (__CPU_HAS_FSQRT)
-    {
-      /* Volatile is required to prevent the compiler from moving the
-        fsqrt instruction above the branch.  */
-      __asm __volatile ("      fsqrts  %0,%1\n"
-                               :"=f" (z):"f" (x));
-    }
-  else
-    z = __slow_ieee754_sqrtf (x);
+#ifdef _ARCH_PPCSQ
+  asm ("fsqrts %0,%1\n" :"=f" (z):"f" (x));
+#else
+  z = __slow_ieee754_sqrtf (x);
+#endif
 
   return z;
 }

diff --git a/sysdeps/powerpc/fpu/math_private.h b/sysdeps/powerpc/fpu/math_private.h
index dde153d37ae7c49c8fabc77747e6c5a1b09b8471..904871506d92d8c69a16d8159da864d3376ddcfa 100644 (file)
--- a/sysdeps/powerpc/fpu/math_private.h
+++ b/sysdeps/powerpc/fpu/math_private.h
@@ -25,26 +25,17 @@
 #include <fenv_private.h>
 #include_next <math_private.h>
 
-# if __WORDSIZE == 64 || defined _ARCH_PWR4
-#  define __CPU_HAS_FSQRT 1
-# else
-#  define __CPU_HAS_FSQRT ((GLRO(dl_hwcap) & PPC_FEATURE_64) != 0)
-# endif
-
 extern double __slow_ieee754_sqrt (double);
 extern __always_inline double
 __ieee754_sqrt (double __x)
 {
   double __z;
 
-  if (__CPU_HAS_FSQRT)
-    {
-      /* Volatile is required to prevent the compiler from moving the
-         fsqrt instruction above the branch.  */
-      __asm __volatile ("fsqrt %0,%1" : "=f" (__z) : "f" (__x));
-    }
-  else
-     __z = __slow_ieee754_sqrt(__x);
+#ifdef _ARCH_PPCSQ
+   asm ("fsqrt %0,%1" : "=f" (__z) : "f" (__x));
+#else
+   __z = __slow_ieee754_sqrt(__x);
+#endif
 
   return __z;
 }
@@ -55,14 +46,11 @@ __ieee754_sqrtf (float __x)
 {
   float __z;
 
-  if (__CPU_HAS_FSQRT)
-    {
-      /* Volatile is required to prevent the compiler from moving the
-         fsqrts instruction above the branch.  */
-      __asm __volatile ("fsqrts        %0,%1" : "=f" (__z) : "f" (__x));
-    }
-  else
-     __z = __slow_ieee754_sqrtf(__x);
+#ifdef _ARCH_PPCSQ
+  asm ("fsqrts %0,%1" : "=f" (__z) : "f" (__x));
+#else
+   __z = __slow_ieee754_sqrtf(__x);
+#endif
 
   return __z;
 }

diff --git a/sysdeps/powerpc/powerpc64/fpu/e_sqrt.c b/sysdeps/powerpc/powerpc64/fpu/e_sqrt.c
deleted file mode 100644 (file)
index 08d5f69..0000000
--- a/sysdeps/powerpc/powerpc64/fpu/e_sqrt.c
+++ /dev/null
@@ -1,30 +0,0 @@
-/* Double-precision floating point square root.
-   Copyright (C) 1997-2014 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, see
-   <http://www.gnu.org/licenses/>.  */
-
-#include <math.h>
-#include <math_private.h>
-
-#undef __ieee754_sqrt
-double
-__ieee754_sqrt (double x)
-{
-  double z;
-  __asm __volatile ("fsqrt %0,%1" : "=f" (z) : "f" (x));
-  return z;
-}
-strong_alias (__ieee754_sqrt, __sqrt_finite)

diff --git a/sysdeps/powerpc/powerpc64/fpu/e_sqrtf.c b/sysdeps/powerpc/powerpc64/fpu/e_sqrtf.c
deleted file mode 100644 (file)
index 598e7cb..0000000
--- a/sysdeps/powerpc/powerpc64/fpu/e_sqrtf.c
+++ /dev/null
@@ -1,30 +0,0 @@
-/* Single-precision floating point square root.
-   Copyright (C) 1997-2014 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, see
-   <http://www.gnu.org/licenses/>.  */
-
-#include <math.h>
-#include <math_private.h>
-
-#undef __ieee754_sqrtf
-float
-__ieee754_sqrtf (float x)
-{
-  double z;
-  __asm ("fsqrts %0,%1" : "=f" (z) : "f" (x));
-  return z;
-}
-strong_alias (__ieee754_sqrtf, __sqrtf_finite)