From: Neil Booth <neil@daikokuya.co.uk>
Date: Sat, 13 Oct 2007 03:34:08 +0000 (+0000)
Subject: If the power of 5 is exact, and the reciprocal exact, the error is zero not one half... 
X-Git-Url: http://plrg.eecs.uci.edu/git/?a=commitdiff_plain;h=d1a23d573d5408be1cf52358fe03df0bfa07e16e;p=oota-llvm.git

If the power of 5 is exact, and the reciprocal exact, the error is zero not one half-ulps.  This prevents an infinite loop in rare cases.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@42950 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/Support/APFloat.cpp b/lib/Support/APFloat.cpp
index 65ec7857255..a4c1affb7a3 100644
--- a/lib/Support/APFloat.cpp
+++ b/lib/Support/APFloat.cpp
@@ -1947,7 +1947,7 @@ APFloat::roundSignificandWithExponent(const integerPart *decSigParts,
           excessPrecision = calcSemantics.precision;
       }
       /* Extra half-ulp lost in reciprocal of exponent.  */
-      powHUerr = 1 + powStatus != opOK;
+      powHUerr = (powStatus == opOK && calcLostFraction == lfExactlyZero) ? 0: 2;
     }
 
     /* Both multiplySignificand and divideSignificand return the