104089 - Cannot reattach context to its thread because of the bug in Context class

Status: VERIFIED WORKSFORME

(Rhino Graveyard :: Core, defect)

Description

[Vitaly]

From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 5.0)
BuildID:    

The following code:

// reattach cx to its thread
Context current;
if((current = Context.enter(cx)) != cx) {
    System.out.println("debugger error: cx = " + cx + " current = " + current);
}

from function org.mozilla.javascript.tools.debugger.Main.interrupted(Context 
cx) doesn't work because of the problem in Context.enter(Context cx) function:

else if (cx != null) {
    synchronized (cx) {
        if (cx.currentThread == null) {
              cx.currentThread = t;
              threadContexts.put(t, cx);
              cx.enterCount++;
         }
    }
    current = cx;
}

When cx.currentThread != null it's not replaced by the current thread as 
attempted.
The fragment above should be replaced with something like that:
else if (cx != null) {
    synchronized (cx) {
        if (cx.currentThread == null || cx.currentThread != t) {
	    if (cx.currentThread != null) {
                 threadContexts.remove(cx.currentThread);
	         cx.enterCount--;
	     }
             cx.currentThread = t;
             threadContexts.put(t, cx);
             cx.enterCount++;
        }
    }
    current = cx;
}



Reproducible: Always
Steps to Reproduce:
1.Start debugger without parameters.
2.Using browser, run a script containing Context.getCurrentContext() call.
3.The call returns null

Actual Results:  My script cannot find the right context because it 
still "attached" to another thread.

Comment 1

[Robert Ginda]

Rhino bug -> Rhino product

Comment 2

[Igor Bukanov]

The bug description is not applicable any more due to changes in
Context.enter/Context.exit, but as the test case works fine now, I suggest to
close this as fixed.

Comment 3

[Phil Schwartau]

Resolving as WORKSFORME -

Comment 4

[Phil Schwartau]

Marking Verified.

Vitaly, if there is a problem that still exists, please reopen - thanks.

Comment 5

[Igor Bukanov]

Targeting as resolved against 1.5R4