ansaurus

Question

How could I track unknown Thread/Runnable?

Answer 1

+3 A:

You could install a security manager which allows absolutely everything, but logs out when it's asked for permission to start a thread.

Or, you could provide a replacement Thread implementation and put it ahead of the one in rt.jar by using the bootclasspath options, and have this replacement Thread log its name and stacktrace in its constructor.

dty 2010-09-15 17:03:02

The security manager idea is a really good idea - +1 for that!

aperkins 2010-09-15 17:03:41

its hard to detect where does this Thread runs. The name is Thread-23 usually. Which security manager you have talked about? Could you provide please some snippets of how could I do this stuff? Thank you.

Eugene 2010-09-15 17:04:50

The reason, that I don't want to determine what kind of Thread is this, I want to get know in which place its calling and by which caller?

Eugene 2010-09-15 17:05:43

Yes, and the ideas I've given you will help you learn where the thread is constructed and by whom. You will have to implement your own security manager (it's very very simple, Google it) if you wanted to try the first idea. For the second idea, just copy the source of Thread.java and recompile it with some extra logging and add it to the bootclasspath.

dty 2010-09-15 17:07:33

So will debug version of rt.jar will help me to find caller point of Thread? Cause I don't want to be pointed just to source of run method.

Eugene 2010-09-15 17:09:57

When the SecurityManager is asked if a thread can start, you can get a stack trace and print that out at that location, as @dty is suggesting in his comment. You can do that by doing Thread.getCurrentThread().getStackTrace() or Thread.getCurrentThread().printStackTrace() [syntax may be slightly off - doing it from memory]

aperkins 2010-09-15 17:10:39

@aperkins It seems to be that this run method is inside of 3rd party jar. So how could I call your methods outside of that class?

Eugene 2010-09-15 17:14:54

*sigh* *headdesk*

dty 2010-09-15 17:16:24

@dty what does it mean?

Eugene 2010-09-15 17:18:51

It means I'm banging my head on the desk! You're not reading what me or aperkins are writing. You're just expecting someone to tell you where the problem is, and nobody can to that. You will need to do some investigation of your own system on your own. I've suggested two techniques that might help with that investigation.

dty 2010-09-15 17:21:01

@dty Nope your dear dty, I'm trying to get all available solutions, because I'm not able to spend a lot of time to test all methods, so just looking for good solution. That's all, dear dty.

Eugene 2010-09-15 17:25:22

dty is correct however - both of those solutions are good, and will likely be the best that you are going to get. I would look up securityManager and read about it, because that is how I would approach the problem you are facing, based on what you have described.

aperkins 2010-09-15 17:43:19

the way with debug version of rt.jar from latest jdk - didn't help.

Eugene 2010-09-15 18:06:31

Nobody except you mentioned using a debug version of rt.jar - whatever one of those is. I said, recompile Thread.java with extra logging in the constructor and put it on the bootclasspath. And I said it at least three times now.

dty 2010-09-15 18:25:21

How should I do this recompiling? I didn't get your solution. Sorry. Explain please. Thank you.

Eugene 2010-09-15 18:27:43

Answer 2

+1 A:

org/tritonus/midi/device/java/JavaSequencer.java has the following code:

protected void openImpl()
    {
            ...
            m_thread = new Thread(this);
            ...
            m_thread.start();
    }

So that's one place to look for (but that answers only the specific case, not the general question how to find out where a thread is started in general).

There are also other places in the same class where this.start() is called: in setSequence(..) and setTickPosition(..). There could be calls to start() outside the class however.

Andre Holzner 2010-09-15 17:05:56

yes sounds good, but regards eclipse debugger this thread appears only at the end of song playback. Thats I want to determine why this appearing and how to try/catch it? This exception dispatching only when playback is finished. But not from code I wrote.

Eugene 2010-09-15 17:08:39

From the error message and stack trace it looks like you're closing the sequencer once too often ? Maybe you don't have to close it by hand, when the thread stops (the program ends), this seems to happen automatically ?!

Andre Holzner 2010-09-15 18:09:03

I don't stop it manually, but I'm checking each time the ticker position, in all other cases application don't touching sequencer except play/pause function.

Eugene 2010-09-15 18:10:55

By the way, Tritonus is open source. I guess you could download the source and add it as a project your project depends on. So you could set breakpoints at various places in Tritonus and look at stack traces.

Andre Holzner 2010-09-15 18:12:15

I tried, but jar I'm using is 2009 year, and there are sources in CVS, and I looked on it, but find whole sources, and have no idea how to separate a java sequencer part from there.

Eugene 2010-09-15 18:26:24

ansaurus

tags:

views:

answers:

How could I track unknown Thread/Runnable?

related questions