[PATCH] Check rvc_normal in real_isdenormal.

[Aldy Hernandez via Gcc-patches]

[Jakub, thanks for pointing this out.  OK?]

[-Inf, -Inf] is being flushed to [-Inf, -0.0] because real_isdenormal
is being overly pessimistic.  It is missing a check for rvc_normal.
This doesn't cause problems in real.cc because all uses of
real_isdenormal are already on the rvc_normal path.  The uses in
value-range.cc however, are not.

This patch adds a check for rvc_normal.

gcc/ChangeLog:

        * real.h (real_isdenormal): Check rvc_normal.
        * value-range.cc (range_tests_floats): New test.
---
 gcc/real.h         | 2 +-
 gcc/value-range.cc | 5 +++++
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/gcc/real.h b/gcc/real.h
index f9528d765ec..306e9593866 100644
--- a/gcc/real.h
+++ b/gcc/real.h
@@ -290,7 +290,7 @@ extern bool real_issignaling_nan (const REAL_VALUE_TYPE *);
 inline bool
 real_isdenormal (const REAL_VALUE_TYPE *r)
 {
-  return (r->sig[SIGSZ-1] & SIG_MSB) == 0;
+  return r->cl == rvc_normal && (r->sig[SIGSZ-1] & SIG_MSB) == 0;
 }
 
 /* Determine whether a floating-point value X is finite.  */
diff --git a/gcc/value-range.cc b/gcc/value-range.cc
index 6b4f3dddcd5..ee15eb35df8 100644
--- a/gcc/value-range.cc
+++ b/gcc/value-range.cc
@@ -4014,6 +4014,11 @@ range_tests_floats ()
   r1.clear_nan ();
   r0.intersect (r1);
   ASSERT_TRUE (r0.undefined_p ());
+
+  // Make sure [-Inf, -Inf] doesn't get normalized.
+  r0 = frange_float ("-Inf", "-Inf");
+  ASSERT_TRUE (real_isinf (&r0.lower_bound (), true));
+  ASSERT_TRUE (real_isinf (&r0.upper_bound (), true));
 }
 
 void
-- 
2.37.3