Use ldexp instead of pow for more accuracy.

Original patch by net147@gmail.com: https://chromiumcodereview.appspot.com/10026011/

BUG=
TEST=cctest/test-conversions/Octal

Review URL: https://chromiumcodereview.appspot.com/10103030

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@11356 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

src/conversions-inl.h

@@ -228,9 +228,7 @@ double InternalStringToIntDouble(UnicodeCache* unicode_cache,
   }
 
   ASSERT(number != 0);
-  // The double could be constructed faster from number (mantissa), exponent
-  // and sign. Assuming it's a rare case more simple code is used.
-  return static_cast<double>(negative ? -number : number) * pow(2.0, exponent);
+  return ldexp(static_cast<double>(negative ? -number : number), exponent);
 }