Optimize fesetenv

Improve fesetenv to use an optimized implementation similar to
feupdateenv.

2014-06-24  Wilco  <wdijkstr@arm.com>

	* sysdeps/arm/fesetenv.c (fesetenv): Optimize implementation.

2 files changed, 24 insertions, 15 deletions

diff --git a/ChangeLog b/ChangeLog
index 67759941ca..b855439303 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,9 @@
 2014-06-24  Wilco  <wdijkstr@arm.com>
 
+	* sysdeps/arm/fesetenv.c (fesetenv): Optimize implementation.
+
+2014-06-24  Wilco  <wdijkstr@arm.com>
+
 	* sysdeps/arm/fpu_control.h (_FPU_MASK_RM): Define.
 	* sysdeps/arm/fenv_private.h (libc_fesetround_vfp)
 	(libc_feholdexcept_setround_vfp) (libc_feholdsetround_vfp)
diff --git a/sysdeps/arm/fesetenv.c b/sysdeps/arm/fesetenv.c
index b2ed1d3911..ac47ae2d49 100644
--- a/sysdeps/arm/fesetenv.c
+++ b/sysdeps/arm/fesetenv.c
@@ -17,14 +17,13 @@
    <http://www.gnu.org/licenses/>.  */
 
 #include <fenv.h>
-#include <fpu_control.h>
 #include <arm-features.h>
 
 
 int
 fesetenv (const fenv_t *envp)
 {
-  fpu_control_t fpscr;
+  fpu_control_t fpscr, new_fpscr, updated_fpscr;
 
   /* Fail if a VFP unit isn't present.  */
   if (!ARM_HAVE_VFP)
@@ -32,25 +31,31 @@ fesetenv (const fenv_t *envp)
 
   _FPU_GETCW (fpscr);
 
-  /* Preserve the reserved FPSCR flags.  */
-  fpscr &= _FPU_RESERVED;
+  if ((envp != FE_DFL_ENV) && (envp != FE_NOMASK_ENV))
+    {
+      /* The new FPSCR is valid, so don't merge the reserved flags.  */
+      new_fpscr = envp->__cw;
 
-  if (envp == FE_DFL_ENV)
-    fpscr |= _FPU_DEFAULT;
-  else if (envp == FE_NOMASK_ENV)
-    fpscr |= _FPU_IEEE;
-  else
-    fpscr |= envp->__cw & ~_FPU_RESERVED;
+      /* Write new FPSCR if different (ignoring NZCV flags).  */
+      if (((fpscr ^ new_fpscr) & ~_FPU_MASK_NZCV) != 0)
+	_FPU_SETCW (new_fpscr);
 
-  _FPU_SETCW (fpscr);
+      return 0;
+    }
 
-  if (envp == FE_NOMASK_ENV)
+  /* Preserve the reserved FPSCR flags.  */
+  new_fpscr = fpscr & _FPU_RESERVED;
+  new_fpscr |= (envp == FE_DFL_ENV) ? _FPU_DEFAULT : _FPU_IEEE;
+
+  if (((new_fpscr ^ fpscr) & ~_FPU_MASK_NZCV) != 0)
     {
+      _FPU_SETCW (new_fpscr);
+
       /* Not all VFP architectures support trapping exceptions, so
 	 test whether the relevant bits were set and fail if not.  */
-      _FPU_GETCW (fpscr);
-      if ((fpscr & _FPU_IEEE) != _FPU_IEEE)
-	return 1;
+      _FPU_GETCW (updated_fpscr);
+
+      return new_fpscr & ~updated_fpscr;
     }
 
   return 0;