about summary refs log tree commit diff
diff options
context:
space:
mode:
authorJoseph Myers <joseph@codesourcery.com>2016-03-09 00:30:59 +0000
committerJoseph Myers <joseph@codesourcery.com>2016-03-09 00:30:59 +0000
commit613c92b3b59df6a06784cde1d4f410cef0b6da96 (patch)
treeefdcc70a56baf3ed21f7278b72efed5d2e8a9b27
parent3bd80c0de2f8e7ca8020d37739339636d169957e (diff)
downloadglibc-613c92b3b59df6a06784cde1d4f410cef0b6da96.tar.gz
glibc-613c92b3b59df6a06784cde1d4f410cef0b6da96.tar.xz
glibc-613c92b3b59df6a06784cde1d4f410cef0b6da96.zip
Fix ldbl-128ibm nearbyintl in non-default rounding modes (bug 19790).
The ldbl-128ibm implementation of nearbyintl uses logic that only
works in round-to-nearest mode.  This contrasts with rintl, which
works in all rounding modes.

Now, arguably nearbyintl could simply be aliased to rintl, given that
spurious "inexact" is generally allowed for ldbl-128ibm, even for the
underlying arithmetic operations.  But given that the only point of
nearbyintl is to avoid "inexact", this patch follows the more
conservative approach of adding conditionals to the rintl
implementation to make it suitable for use to implement nearbyintl,
then builds it for nearbyintl with USE_AS_NEARBYINTL defined.  The
test test-nearbyint-except-2 shows up issues when traps on "inexact"
are enabled, which turn out to be problems with the powerpc
fenv_private.h implementation (two functions that should disable
exception traps potentially failing to do so in some cases); this
patch duly fixes that as well (I don't see any other existing cases
where this would be user-visible; there isn't much use of *_NOEX,
*hold* etc. in libm that requires exceptions to be discarded and not
trapped on).

Tested for powerpc.

	[BZ #19790]
	* sysdeps/ieee754/ldbl-128ibm/s_rintl.c [USE_AS_NEARBYINTL]
	(rintl): Define as macro.
	[USE_AS_NEARBYINTL] (__rintl): Likewise.
	(__rintl) [USE_AS_NEARBYINTL]: Use SET_RESTORE_ROUND_NOEX instead
	of fesetround.  Ensure results are evaluated before end of scope.
	* sysdeps/ieee754/ldbl-128ibm/s_nearbyintl.c: Define
	USE_AS_NEARBYINTL and include s_rintl.c.
	* sysdeps/powerpc/fpu/fenv_private.h (libc_feholdsetround_ppc):
	Disable exception traps in new environment.
	(libc_feholdsetround_ppc_ctx): Likewise.
-rw-r--r--ChangeLog14
-rw-r--r--sysdeps/ieee754/ldbl-128ibm/s_nearbyintl.c109
-rw-r--r--sysdeps/ieee754/ldbl-128ibm/s_rintl.c14
-rw-r--r--sysdeps/powerpc/fpu/fenv_private.h4
4 files changed, 32 insertions, 109 deletions
diff --git a/ChangeLog b/ChangeLog
index 73a49a06e0..a6be76213e 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,17 @@
+2016-03-09  Joseph Myers  <joseph@codesourcery.com>
+
+	[BZ #19790]
+	* sysdeps/ieee754/ldbl-128ibm/s_rintl.c [USE_AS_NEARBYINTL]
+	(rintl): Define as macro.
+	[USE_AS_NEARBYINTL] (__rintl): Likewise.
+	(__rintl) [USE_AS_NEARBYINTL]: Use SET_RESTORE_ROUND_NOEX instead
+	of fesetround.  Ensure results are evaluated before end of scope.
+	* sysdeps/ieee754/ldbl-128ibm/s_nearbyintl.c: Define
+	USE_AS_NEARBYINTL and include s_rintl.c.
+	* sysdeps/powerpc/fpu/fenv_private.h (libc_feholdsetround_ppc):
+	Disable exception traps in new environment.
+	(libc_feholdsetround_ppc_ctx): Likewise.
+
 2016-03-08  Roland McGrath  <roland@hack.frob.com>
 
 	* sysdeps/x86_64/tst-audit10.c: #include <cpu-features.h>.
diff --git a/sysdeps/ieee754/ldbl-128ibm/s_nearbyintl.c b/sysdeps/ieee754/ldbl-128ibm/s_nearbyintl.c
index 08134edd10..dfdefe3f02 100644
--- a/sysdeps/ieee754/ldbl-128ibm/s_nearbyintl.c
+++ b/sysdeps/ieee754/ldbl-128ibm/s_nearbyintl.c
@@ -17,110 +17,5 @@
    License along with the GNU C Library; if not, see
    <http://www.gnu.org/licenses/>.  */
 
-/* This has been coded in assembler because GCC makes such a mess of it
-   when it's coded in C.  */
-
-#include <math.h>
-#include <math_private.h>
-#include <fenv.h>
-#include <math_ldbl_opt.h>
-#include <float.h>
-#include <ieee754.h>
-
-
-long double
-__nearbyintl (long double x)
-{
-  fenv_t env;
-  static const long double TWO52 = 4503599627370496.0L;
-  union ibm_extended_long_double u;
-  u.ld = x;
-
-  if (!isfinite (u.d[0].d))
-    return x;
-  else if (fabs (u.d[0].d) < TWO52)
-    {
-      double xh = u.d[0].d;
-      double high = u.d[0].d;
-      feholdexcept (&env);
-      if (high > 0.0)
-	{
-	  high += TWO52;
-	  high -= TWO52;
-          if (high == -0.0) high = 0.0;
-	}
-      else if (high < 0.0)
-	{
-	  high -= TWO52;
-	  high += TWO52;
-          if (high == 0.0) high = -0.0;
-	}
-      if (u.d[1].d > 0.0 && (xh - high == 0.5))
-        high += 1.0;
-      else if (u.d[1].d < 0.0 && (-(xh - high) == 0.5))
-        high -= 1.0;
-      u.d[0].d = high;
-      u.d[1].d = 0.0;
-      math_force_eval (u.d[0]);
-      math_force_eval (u.d[1]);
-      fesetenv (&env);
-    }
-  else if (fabs (u.d[1].d) < TWO52 && u.d[1].d != 0.0)
-    {
-      double high = u.d[0].d, low = u.d[1].d, tau;
-      /* In this case we have to round the low double and handle any
-         adjustment to the high double that may be caused by rounding
-         (up).  This is complicated by the fact that the high double
-         may already be rounded and the low double may have the
-         opposite sign to compensate.  */
-      feholdexcept (&env);
-      if (u.d[0].d > 0.0)
-	{
-	  if (u.d[1].d > 0.0)
-	    {
-	      /* If the high/low doubles are the same sign then simply
-	         round the low double.  */
-	    }
-	  else if (u.d[1].d < 0.0)
-	    {
-	      /* Else the high double is pre rounded and we need to
-	         adjust for that.  */
-
-	      tau = __nextafter (u.d[0].d, 0.0);
-	      tau = (u.d[0].d - tau) * 2.0;
-	      high -= tau;
-	      low += tau;
-	    }
-	  low += TWO52;
-	  low -= TWO52;
-	}
-      else if (u.d[0].d < 0.0)
-	{
-	  if (u.d[1].d < 0.0)
-	    {
-	      /* If the high/low doubles are the same sign then simply
-	         round the low double.  */
-	    }
-	  else if (u.d[1].d > 0.0)
-	    {
-	      /* Else the high double is pre rounded and we need to
-	         adjust for that.  */
-	      tau = __nextafter (u.d[0].d, 0.0);
-	      tau = (u.d[0].d - tau) * 2.0;
-	      high -= tau;
-	      low += tau;
-	    }
-	  low = TWO52 - low;
-	  low = -(low - TWO52);
-	}
-      u.d[0].d = high + low;
-      u.d[1].d = high - u.d[0].d + low;
-      math_force_eval (u.d[0]);
-      math_force_eval (u.d[1]);
-      fesetenv (&env);
-    }
-
-  return u.ld;
-}
-
-long_double_symbol (libm, __nearbyintl, nearbyintl);
+#define USE_AS_NEARBYINTL
+#include "s_rintl.c"
diff --git a/sysdeps/ieee754/ldbl-128ibm/s_rintl.c b/sysdeps/ieee754/ldbl-128ibm/s_rintl.c
index 8c51ded1d6..e4af01c9a0 100644
--- a/sysdeps/ieee754/ldbl-128ibm/s_rintl.c
+++ b/sysdeps/ieee754/ldbl-128ibm/s_rintl.c
@@ -26,6 +26,11 @@
 #include <float.h>
 #include <ieee754.h>
 
+#ifdef USE_AS_NEARBYINTL
+# define rintl nearbyintl
+# define __rintl __nearbyintl
+#endif
+
 
 long double
 __rintl (long double x)
@@ -44,7 +49,11 @@ __rintl (long double x)
 
       /* Long double arithmetic, including the canonicalisation below,
 	 only works in round-to-nearest mode.  */
+#ifdef USE_AS_NEARBYINTL
+      SET_RESTORE_ROUND_NOEX (FE_TONEAREST);
+#else
       fesetround (FE_TONEAREST);
+#endif
 
       /* Convert the high double to integer.  */
       orig_xh = xh;
@@ -103,7 +112,12 @@ __rintl (long double x)
       if (orig_xh < 0.0)
 	xh = -__builtin_fabs (xh);
 
+#ifdef USE_AS_NEARBYINTL
+      math_force_eval (xh);
+      math_force_eval (xl);
+#else
       fesetround (save_round);
+#endif
     }
 
   return ldbl_pack (xh, xl);
diff --git a/sysdeps/powerpc/fpu/fenv_private.h b/sysdeps/powerpc/fpu/fenv_private.h
index e1b02a3f80..02ac980909 100644
--- a/sysdeps/powerpc/fpu/fenv_private.h
+++ b/sysdeps/powerpc/fpu/fenv_private.h
@@ -146,7 +146,7 @@ libc_feholdsetround_ppc (fenv_t *e, int r)
 
   old.fenv = fegetenv_register ();
   /* Clear current precision and set newer one.  */
-  new.l = (old.l & ~0x3) | r;
+  new.l = (old.l & ~0x3 & ~_FPU_MASK_ALL) | r;
   *e = old.fenv;
 
   if ((old.l & _FPU_MASK_ALL) != 0)
@@ -240,7 +240,7 @@ libc_feholdsetround_ppc_ctx (struct rm_ctx *ctx, int r)
   fenv_union_t old, new;
 
   old.fenv = fegetenv_register ();
-  new.l = (old.l & ~0x3) | r;
+  new.l = (old.l & ~0x3 & ~_FPU_MASK_ALL) | r;
   ctx->env = old.fenv;
   if (__glibc_unlikely (new.l != old.l))
     {