339120 - Jigzone Java puzzles hang trunk

Status: RESOLVED FIXED

(Core Graveyard :: Plug-ins, defect, P1)

Description

[Ria Klaassen (not reading all bugmail)]

Steps to reproduce: 
1. Go to http://www.jigzone.com/z.php?k=2F4GIDD8 and wait until Java has stopped loading
2. Then start loading the puzzle links under Recent Puzzles:
Boat Rope
Rose-P
Shopping Carts
Golden Stamen
Herbie.
One by one and wait after each puzzle until the processor has become fairly quiet again.
3. Result: Firefox hangs while loading "Shopping Carts" but Seamonkey can load one puzzle more. But they both finally hang.
When I close the browser Java stays in the systray so I need to end the process in the task manager.

This is a regression between 1.9a1_2006051010 and 1.9a1_2006051022.
It is trunk-only.
http://bonsai.mozilla.org/cvsquery.cgi?module=PhoenixTinderbox&date=explicit&mindate=2006-05-10+09%3A00&maxdate=2006-05-10+23%3A00

I think it could be caused by Bug 326273 but it is just a guess.

Comment 1

[Hermann Schwab]

there is a new version available, but as you marked the regression time range I don't think it helps in this case. JRE 5.0 Update 7

http://java.sun.com/j2se/1.5.0/download.jsp 
http://javashoplm.sun.com/ECom/docs/Welcome.jsp?StoreId=22&PartDetailId=jre-1.5.0_07-oth-JPR&SiteId=JSC&TransactionId=noreg

Comment 2

[Ria Klaassen (not reading all bugmail)]

Not sure what you mean but it is tested with the latest Java version ("You already have the latest Java(TM) Platform on this system"). 

Comment 3

[Ria Klaassen (not reading all bugmail)]

Rechecked regression range which is correct.

Comment 4

[Ria Klaassen (not reading all bugmail)]

What also might be relevant to say is that jigzone is a leaking site (Bug 324796), *but* there were never problems before 10 May 2006.

Comment 5

[Darin Fisher]

-> me

Comment 6

[Darin Fisher]

Repeating the steps from comment #0, Firefox just locked up.  Interesting...

Comment 7

[Darin Fisher]

One possible change that could have impacted the Java plug-in behavior is the change to nsJVMManager::PostEvent with async = false.  In that case, we now process pending events on the current thread while waiting for the event to be synchronously dispatched.  It is possible that that is causing problems for the Java plug-in.  It'd be easy enough to create a patch that better matches the old behavior of waiting on a condition variable until the event runs.

Comment 8

[Darin Fisher]

Attachment: v1 patch

This seems to help a lot.

Comment 9

[Darin Fisher]

fixed-on-trunk