java PrintCompilation output: what's the meaning of "made not entrant" and "made zombie"

When running a Java 1.6 (1.6.0_03-b05) app I've added the -XX:+PrintCompilation flag. On the output for some methods, in particular some of those that I know are getting called a lot, I see the text made not entrant and made zombie.

What do these mean? Best guess is that it's a decompilation step before recompiling either that method or a dependency with greater optimisation. Is that true? Why "zombie" and "entrant"?

Example, with quite a bit of time between some of these lines:

[... near the beginning]
42       jsr166y.LinkedTransferQueue::x开发者_运维问答fer (294 bytes)

[... much later]
42    made not entrant  jsr166y.LinkedTransferQueue::xfer (294 bytes)
---   n   sun.misc.Unsafe::compareAndSwapObject
170       jsr166y.LinkedTransferQueue::xfer (294 bytes)
170   made not entrant  jsr166y.LinkedTransferQueue::xfer (294 bytes)
  4%      jsr166y.LinkedTransferQueue::xfer @ 29 (294 bytes)
171       jsr166y.LinkedTransferQueue::xfer (294 bytes)

[... even later]
42    made zombie  jsr166y.LinkedTransferQueue::xfer (294 bytes)
170   made zombie  jsr166y.LinkedTransferQueue::xfer (294 bytes)
171   made not entrant  jsr166y.LinkedTransferQueue::xfer (294 bytes)
172       jsr166y.LinkedTransferQueue::xfer (294 bytes)

[... no further logs]

I've pulled together some info on this on my blog. A Cliff Click comment I found says:

Zombie methods are methods whose code has been made invalid by class loading. Generally the server compiler makes aggressive inlining decisions of non-final methods. As long as the inlined method is never overridden the code is correct. When a subclass is loaded and the method overridden, the compiled code is broken for all future calls to it. The code gets declared "not entrant" (no future callers to the broken code), but sometimes existing callers can keep using the code. In the case of inlining, that's not good enough; existing callers' stack frames are "deoptimized" when they return to the code from nested calls (or just if they are running in the code). When no more stack frames hold PC's into the broken code it's declared a "zombie" - ready for removal once the GC gets around to it.

This is absolutely not an area of expertise for me, but I was interested and so did a bit of digging.

A couple of links you might find interesting: OpenJDK:nmethod.cpp, OpenJDK:nmethod.hpp.

A excerpt of nmethod.hpp:

// Make the nmethod non entrant. The nmethod will continue to be
// alive.  It is used when an uncommon trap happens.  Returns true
// if this thread changed the state of the nmethod or false if
// another thread performed the transition.
bool  make_not_entrant() { return make_not_entrant_or_zombie(not_entrant); }
//...

Just as a starting place.

A newer explanation is that there can be entries like this in log:

129   72       3       EscapeAnalysysTest::second (24 bytes)
.......... some more lines
135   76       4       EscapeAnalysysTest::second (24 bytes)
137   74       3       EscapeAnalysysTest::second (24 bytes)   made not entrant

This actually means that one method (second here) has been compiled under tier level 3, then under tier level 4 it has been further more optimized and it has been made not entrant for the 3-d tier; meaning it will be replaced with code from the 4-th tier.

Here is a Gist with an unbelievable amount of information on PrintCompilation. Specifically, it says:

When deoptimization happens, if it is decided to invalidate the offending nmethod, it will be "made not entrant" first; and then, when the NMethodSweeper finds that there are no activations of it on the stacks any more, it is "made zombie";