From: Andreas Schwab <schwab@linux-m68k.org>
Date: Fri, 23 Mar 2012 22:24:23 +0000 (+0100)
Subject: Fix undue underflow in ldbl-128ibm version of expl
X-Git-Tag: glibc-2.16-tps~712
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=a3f61311c0280da11ac79b9fc4b9b280449457bd;p=platform%2Fupstream%2Fglibc.git

Fix undue underflow in ldbl-128ibm version of expl
---

diff --git a/ChangeLog b/ChangeLog
index 8ec88cc..c1cf74a 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,8 @@
+2012-03-26  Andreas Schwab  <schwab@linux-m68k.org>
+
+	* sysdeps/ieee754/ldbl-128ibm/e_expl.c (lomark): Adjust to take
+	denormal result into account.
+
 2012-03-25  Roland McGrath  <roland@hack.frob.com>
 
 	* posix/confstr.c (confstr): Lift RESTENVS definition to function scope.
diff --git a/sysdeps/ieee754/ldbl-128ibm/e_expl.c b/sysdeps/ieee754/ldbl-128ibm/e_expl.c
index 133adc2..5618eb3 100644
--- a/sysdeps/ieee754/ldbl-128ibm/e_expl.c
+++ b/sysdeps/ieee754/ldbl-128ibm/e_expl.c
@@ -1,5 +1,5 @@
 /* Quad-precision floating point e^x.
-   Copyright (C) 1999,2004,2006, 2008, 2011 Free Software Foundation, Inc.
+   Copyright (C) 1999-2012 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
    Contributed by Jakub Jelinek <jj@ultra.linux.cz>
    Partly based on double-precision code
@@ -74,7 +74,7 @@ static const long double C[] = {
 
 /* Largest integer x for which e^x underflows.  */
 #define lomark C[1]
--709.08956571282405153382846025171462914L,
+-744.44007192138121808966388925909996033L,
 
 /* 3x2^96 */
 #define THREEp96 C[2]