70588 - Java plugin uses the nsILiveconnect interface as a singleton service [REQUESTOR: Sun JPI group]

Status: RESOLVED FIXED

(Core Graveyard :: Java: OJI, defect, P3)

Description

[Thomas Lendacky]

From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 4.0)
BuildID:    0000000000

Java plugin uses the nsILiveconnect interface as a service.

With the fix to bug 63466 the nsILiveconnect interface changed from a singleton 
service to an interface that requires an instance per plugin (or applet in this 
case).  It needs to be determined if the Java plugin needs to be re-designed/re-
architected to use an instance of nsILiveconnect per applet or if only one 
instance is needed for the JVM process.

This bug is being created so that this issue is not dropped or lost.

Comment 1

[Phil Schwartau]

Reassigning to OJI component - 

Comment 2

[Phil Schwartau]

Really reassigning 

Comment 3

[edburns]

Hi Joe,

It looks like Thomas has done us a favor and opened up a bug for us.

Beard: any reply to Joe's question about backing out that part of the 63466 fix 
until we can modify the java plugin to account for it?

Comment 4

[Patrick C. Beard]

This fix is crucial for correct functioning of LiveConnect on the Mac. I don't 
believe that your plugin will work any worse. I could back it out partially by 
using an #ifdef XP_MAC for the caching of the plugin instance. Would that server 
your purpose?

Comment 5

[Joe Chou]

Stanley, what do you think about Patrick's proposed change?

Comment 6

[Joe Chou]

I checked with Stanley, and he said that the change proposed by Patrick should
be OK as a temp solution for now. Eventually, probably in Java plugin 1.4
release, the fix of this problem will be updated on both the browser side and
the Java plugin side.

Comment 7

[Doug Rothert]

Joe/Patrick, 

Could we get those #ifdef changes ASAP, possibly by attaching them here? 

If we are going to change the OS/2 version of 63466 source before code cutoff 
we need to do so within the next day or so. 

Thanks, 
-Doug

Comment 8

[Joe Chou]

Patrick, would you put the patch that do #ifdef MAC for the change of singleton
nsILiveconnect instance here, as you suggested? The IBM folks are planning to
apply the patch before their code freeze. If you have other thoughts, please let
us know. Thanks.

Comment 9

[drapeau]

Marking as one of the OJI group's hot bugs.

Comment 10

[drapeau]

Removing the escalation status on this: it's important, but no longer urgent
since the next Java Plug-In that will take advantage of this bug fix won't be
available until November.

Comment 11

[Joe Chou]

*** Bug 70586 has been marked as a duplicate of this bug. ***

Comment 12

[Joe Chou]

A Sun internal bug was created (#4461467) for the Java plugin side to update to
be able to handle multiple nsILiveconnect instances, which has 
been fixed in 1.4 release by Stanley Ho. Therefore, with Java plugin 1.4, the
Java plugin side should be able to handle multiple nsILiveconnect instances.

Comment 13

[edburns]

I'll accept this for you.

Comment 14

[drapeau]

So Joe: what is the status on the Mozilla side of things?  Is there work that
you will be doing on this bug, or is the only remaining work in Java Plug-In? 
If there's work to do in the Mozilla code base, then please update this bug with
status or an ETA and description of the work involved.  If there is *no more*
work to be done in the Mozilla code base, then please close the bug as FIXED or
INVALID since it's not a Mozilla bug, the justification being that Java plugins
will need to change.  (I believe there is ample justification of this already in
previous comments, so no need to write a new justification).

Also, if there *is* work to be done in the Mozille codebase, then please add the
word "oji_escalation" in the Status Whiteboard field of this bug so we can track
it more easily, and you should probably change the priority in that case as well.

Comment 15

[Joe Chou]

At this point, Java plugin has fixed the singleton problem in 1.4 release (Sun
internal bug #4461467), so that it can handle multiple liveconnect instances.
The original plan (to make some temperary change to support Java plugin before
1.4 is released) is no longer needed. Marked fixed.

Comment 16

[Joe Chou]

Ran the test case of 70586 (the original bug for 70588) again, with the recent
mozilla trunk build and JRE 1.4 beta (containing the fix of 4461467). The result
was the same as the previous test result of 70586: the first time closing the
window seemed OK, but crashed at the second time trying to close the window
again. 
Reopening the bug. 

Comment 17

[Joe Chou]

Attachment: The original test case in a zip file

Comment 18

[Joe Chou]

Attached the original test case from 70586.  

Comment 19

[shrirang khanzode]

SPAM: reassigning all OJI bugs to new OJI QA, pmac ( 227 bugs)

Comment 20

[Steve K]

The attachment in this bug still causes crashes, even with the latest Java bits
from Sun (1.4.1-rc) on Linux.

Here's some output from the Java console:

Java(TM) Plug-in: Version 1.4.1
Using JRE version 1.4.1-rc Java HotSpot(TM) Client VM
User home directory = /home/stevek
Proxy Configuration: Browser Proxy Configuration


----------------------------------------------------
c:   clear console window
f:   finalize objects on finalization queue
g:   garbage collect
h:   display this help message
l:   dump classloader list
m:   print memory usage
o:   trigger logging
p:   reload proxy configuration
q:   hide console
r:   reload policy configuration
s:   dump system properties
t:   dump thread list
v:   dump thread stack
x:   clear classloader cache
0-5: set trace level to <n>
----------------------------------------------------
Current thread:Thread[thread applet-TMain.class,4,file:/tmp/Apache/htdocs/LiveCo
nnect/-threadGroup]
OJIPlugin: No AThread

OJIPlugin acq Spontaneous pipe=10
Trying to enter spont monitor: 0
 OJIPlugin release for:Thread[thread applet-TMain.class,4,file:/tmp/Apache/htdoc
s/LiveConnect/-threadGroup]
OJIPlugin releasePipe - exiting spont monitor
Current thread:Thread[thread applet-TMain.class,4,file:/tmp/Apache/htdocs/LiveCo
nnect/-threadGroup]
OJIPlugin: No AThread

OJIPlugin acq Spontaneous pipe=10
Trying to enter spont monitor: 0
 OJIPlugin release for:Thread[thread applet-TMain.class,4,file:/tmp/Apache/htdoc
s/LiveConnect/-threadGroup]
OJIPlugin releasePipe - exiting spont monitor
TMain started
Invoking 'ivDocument.eval( "openNextWindow( )" )' function...
Current thread:Thread[AWT-EventQueue-1,4,file:/tmp/Apache/htdocs/LiveConnect/-th
readGroup]
OJIPlugin: No AThread

OJIPlugin acq Spontaneous pipe=10
Trying to enter spont monitor: 0
 OJIPlugin release for:Thread[AWT-EventQueue-1,4,file:/tmp/Apache/htdocs/LiveCon
nect/-threadGroup]
OJIPlugin releasePipe - exiting spont monitor
  returned from 'ivDocument.eval( "openNextWindow( )" )' function...

Current thread:Thread[thread applet-TNext1.class,4,file:/tmp/Apache/htdocs/LiveC
onnect/-threadGroup]
OJIPlugin: No AThread

OJIPlugin acq Spontaneous pipe=10
Trying to enter spont monitor: 0
Current thread:Thread[thread applet-TNext2.class,4,file:/tmp/Apache/htdocs/LiveC
onnect/-threadGroup]
OJIPlugin: No AThread

OJIPlugin acq Spontaneous pipe=10
Trying to enter spont monitor: 1
 OJIPlugin release for:Thread[thread applet-TNext1.class,4,file:/tmp/Apache/htdo
cs/LiveConnect/-threadGroup]
OJIPlugin releasePipe - exiting spont monitor
Current thread:Thread[thread applet-TNext1.class,4,file:/tmp/Apache/htdocs/LiveC
onnect/-threadGroup]
OJIPlugin: No AThread

OJIPlugin acq Spontaneous pipe=10
Trying to enter spont monitor: 0
 OJIPlugin release for:Thread[thread applet-TNext2.class,4,file:/tmp/Apache/htdo
cs/LiveConnect/-threadGroup]
OJIPlugin releasePipe - exiting spont monitor
Current thread:Thread[thread applet-TNext2.class,4,file:/tmp/Apache/htdocs/LiveC
onnect/-threadGroup]
OJIPlugin: No AThread

OJIPlugin acq Spontaneous pipe=10
Trying to enter spont monitor: 0
 OJIPlugin release for:Thread[thread applet-TNext1.class,4,file:/tmp/Apache/htdo
cs/LiveConnect/-threadGroup]
OJIPlugin releasePipe - exiting spont monitor
TNext1 started
 OJIPlugin release for:Thread[thread applet-TNext2.class,4,file:/tmp/Apache/htdo
cs/LiveConnect/-threadGroup]
OJIPlugin releasePipe - exiting spont monitor
TNext2 started
Invoking 'ivDocument.eval( "notifyTNext1( )" )' function...
Current thread:Thread[Thread-2,5,main]
OJIPlugin: Retrieve the AThread

OJIPlugin acq thread=:Thread[Thread-2,5,main] pipe=23
TNextWindowClosed
 OJIPlugin release for:Thread[Thread-2,5,main]
OJIPlugin release - non spontaneous pipe
java.security.PrivilegedActionException: java.lang.reflect.InvocationTargetExcep
tion
        at java.security.AccessController.doPrivileged(Native Method)
        at sun.plugin.liveconnect.SecureInvocation$2.run(SecureInvocation.java:1
41)
        at java.security.AccessController.doPrivileged(Native Method)
        at sun.plugin.liveconnect.SecureInvocation.CallMethod(SecureInvocation.j
ava:120)
        at sun.plugin.navig.motif.AThread.handleRequest(Native Method)
        at sun.plugin.navig.motif.AThread.JNIHandleLoop(AThread.java:35)
        at sun.plugin.navig.motif.AThread.run(AThread.java:27)
Caused by: java.lang.reflect.InvocationTargetException
        at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
        at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.
java:39)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAcces
sorImpl.java:25)
        at java.lang.reflect.Method.invoke(Method.java:324)
        at sun.plugin.liveconnect.PrivilegedCallMethodAction.run(SecureInvocatio
n.java:497)
        ... 7 more
Caused by: netscape.javascript.JSException
        at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
        at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstruct
orAccessorImpl.java:39)
        at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingC
onstructorAccessorImpl.java:27)
        at java.lang.reflect.Constructor.newInstance(Constructor.java:274)
        at sun.plugin.liveconnect.PrivilegedConstructObjectAction.run(SecureInvo
cation.java:469)
        at java.security.AccessController.doPrivileged(Native Method)
        at sun.plugin.liveconnect.SecureInvocation$1.run(SecureInvocation.java:8
6)
        at java.security.AccessController.doPrivileged(Native Method)
        at sun.plugin.liveconnect.SecureInvocation.ConstructObject(SecureInvocat
ion.java:65)
        at sun.plugin.javascript.navig5.JSObject.JSObjectEval(Native Method)
        at sun.plugin.javascript.navig5.JSObject.eval(JSObject.java:137)
        at TNext2.closeFrame(TNext2.java:31)
        ... 12 more

=========================================================

Same behavior as reported earlier.  The first time you open/close the windows,
you just get the exception above.  The second time, it just crashes.

I ended up here because recently, we've begun using LiveConnect in our web
application, and on Linux, we've found that our browsers seem to crash
inexplicably, especially when we close windows.  It's really quite unusable.

Any idea when this might finally be addressed?

Comment 21

[Steve K]

Just tried this testcase on Netscape7 with it's included VM, on Win98.

Pressing the button seemed to cause the Java VM to hang (console
unresponsive/doesn't redraw, no applets run), and the browser to become
"unstable" (I could open this bug with it, and enter a comment, but I couldn't
submit).

I don't know if all of these problems that are reproduced so well by this
testcase are due to the issue in this bug, but I do know that now that we are
developing a LiveConnect-enabled application, I've been pretty unable to work
with Mozilla crashing left and right.

Can we assign the "crash" keyword here, or otherwise do something to get this
issue some attention?

Comment 22

[Phil Schwartau]

Adding crash keyword -

Comment 23

[Joshua Xia]

reassign to me

Comment 24

[Pete Zha]

remove crash, adding hang since attachment 42787 [details] hangs for me

Comment 25

[Joshua Xia]

->kyle

Comment 26

[Kyle Yuan]

fixed in jre 1.4.2.