RFR(XXS) 8197857 fieldDescriptor prints incorrect 32-bit representation of compressed oops

[Ioi Lam]

https://bugs.openjdk.java.net/browse/JDK-8197857


When UseCompressedOops is enabled for 64-bit VMs, 
fieldDescriptor::print_on_for
prints two 32-bit integers for each object field. E.g.

  - 'asTypeCache' 'Ljava/lang/invoke/MethodHandle;' @24 NULL (0 8221b591)
  - final 'argL0' 'Ljava/lang/Object;' @28 a 
'LambHello$$Lambda$1'{0x00000004110dac88} (8221b591 1)

However, compressed oops occupy the space of only a single 32-bit
integer, so the superfluous output is confusing.

The above should be printed as

  - 'asTypeCache' 'Ljava/lang/invoke/MethodHandle;' @24 NULL (0)
  - final 'argL0' 'Ljava/lang/Object;' @28 a 
'LambHello$$Lambda$1'{0x00000004110dac88} (8221b591)

Patch:
=======================================

--- a/src/hotspot/share/runtime/fieldDescriptor.cpp    Mon Feb 12 
09:12:59 2018 -0800
+++ b/src/hotspot/share/runtime/fieldDescriptor.cpp    Tue Feb 13 
09:24:26 2018 -0800
@@ -201,6 +201,13 @@
    }
    // Print a hint as to the underlying integer representation. This 
can be wrong for
    // pointers on an LP64 machine
+
+#ifdef _LP64
+  if ((ft == T_OBJECT || ft == T_ARRAY) && UseCompressedOops) {
+    st->print(" (%x)", obj->int_field(offset()));
+  }
+  else // <- intended
+#endif
    if (ft == T_LONG || ft == T_DOUBLE LP64_ONLY(|| 
!is_java_primitive(ft)) ) {
      st->print(" (%x %x)", obj->int_field(offset()), 
obj->int_field(offset()+sizeof(jint)));
    } else if (as_int < 0 || as_int > 9) {