Improve root set used by hat

[Keith Randall]

I'm working on tracing down a ClassLoader leak in a web application I'm
working on (see
http://blogs.sun.com/fkieviet/entry/classloader_leaks_the_dreaded_java for a
description of the kind of leak I'm talking about).  I'm using hat to find
sources of the leak, and I have run into a problem where hat generates a lot
of false positives of a particular form.  When I list the reference chains
from the rootset for a ClassLoader, it lists lots of chains which start at a
static field of a class which was loaded by that classloader.  This isn't a
real leak, however, because if the ClassLoader is otherwise dead, the class
containing that static field will be GCed.

I propose that hat shouldn't consider static fields of classes which aren't
loaded by the root classloader as GC roots.  These roots will be found
during the normal heap walk (classloader -> class -> static field).  Unless
of course the classloader isn't reachable, but that is exactly when we don't
want to consider static fields in these classes as roots.

Does this sound like a reasonable thing to do?  It works for me, just want
to make sure it won't break some other use case for hat.

Possible patch (to openjdk6 build b16):

--- com/sun/tools/hat/internal/model/Snapshot.java
** 331,336 ****
--- 331,345 ----
          System.out.println();
          for (Root r : roots) {
              r.resolve(this);
+             if (r.getType() == Root.JAVA_STATIC) {
+                 JavaClass clazz = (JavaClass) r.getReferer();
+                 if (clazz.getLoader() != nullThing) {
+                     // this "root" will be discovered through a reference
+                     // to its classloader, so we don't need to treat it as
+                     // an explicit root.
+                     continue;
+                 }
+             }
              JavaHeapObject t = findThing(r.getId());
              if (t != null) {
                  t.addReferenceFromRoot(r);
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/jdk6-dev/attachments/20091006/a5e070ea/attachment.html