356282 - [Linux] Very long PARAM tags on APPLETs lock up Firefox [@JavaVM5::WaitingForReply]

Status: RESOLVED INCOMPLETE

(Core Graveyard :: Java: OJI, defect)

Description

[Bob Beaty]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1) Gecko/20061003 Firefox/2.0
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1) Gecko/20061003 Firefox/2.0

I have an applet that takes plain-text encoded data (think uuencoded) as a PARAM tag. When I use Firefox for XP it loads and runs fine. When I use Firefox for Linux (Fedora Core 3) it hangs. I've used 1.5.0.6 and 1.5.0.7 and 2.0rc2 and they all have the same behavior. I'm going to try and attach the HTML file after I file this bug. Firefox starts the JVM, I see the Java console come up and initialize and then Firefox hangs. No repaints, no nothing. But this is ONLY on linux. Windows is fine.

Reproducible: Always

Steps to Reproduce:
1. Load attached HTML file.
2. JVM console appears
3. Firefox hangs.

Actual Results:  
I have to 'kill -9' the processes.

Comment 1

[Bob Beaty]

Attachment: HTML with APPLET and a *very* long PARAM tag

Comment 2

[Ryan Jones-Ward [:sciguyryan]]

Takes a new seconds here but nothing really bad.

Comment 3

[Bob Beaty]

(In reply to comment #2)
> Takes a new seconds here but nothing really bad.

I don't mean to be a pain, but when I try this on a 2GB RAM, i686 running Fedora Core 3 and Firefox 2.0rc2 (as well as 1.5.0.6 and 1.5.0.7) I get it to hang.

Firefox on Win XP is *fine* - works great.

What hardware/OS are you using to get a response in a few seconds? I'd love to be able to put that together here and prove that it's on my end.

Comment 4

[Andrew Schultz]

Attachment: stacktrace during hang (SeaMonkey)

looks like a JRE bug

Comment 5

[Bob Beaty]

(In reply to comment #4)
> Created an attachment (id=242211) [edit]
> stacktrace during hang (SeaMonkey)
> 
> looks like a JRE bug

I've tried Sun's JRE 1.4.2_06, 1.4.2_07, 1.4.2_08, 1.5.0_02, and 1.5.0_08 and they all work fine in Firefox for the Sun's test of applet functionality as well as working for this example so long as the PARAM tag is considerably shorter.

Also, appletviewer for all these different JREs displays the applet/page fine. If it's a JRE bug then I'd expect it to fail on other platforms as well (WinXP in particular) as well as having the appletviewer fail.

So while it's possible to be in the JRE, I still wonder if it isn't something in Firefox on linux.

Comment 6

[Andrew Schultz]

> I've tried Sun's JRE 1.4.2_06, 1.4.2_07, 1.4.2_08, 1.5.0_02, and 1.5.0_08 and
> they all work fine in Firefox for the Sun's test of applet functionality as

I really doubt that the "test of applet functionality" is meant to be an exhaustive test of the plugin.

> well as working for this example so long as the PARAM tag is considerably
> shorter.

Which doesn't really indicate anything.  The thing that's actually hanging in the stacktrace /is/ the plugin.

Anyway, I moved this bug into the Gecko component for communicating with the Java plugin, which (if it's not a plugin bug) is where the buggy code lives.

Comment 7

[Benjamin Smedberg]

Mass-closing bugs in the "OJI" component: OJI plugin integration was replaced with npruntime long ago, and these bugs appear to be irrelevant now. If there is in fact a real bug that remains, please file it new in the "Core" product, component "Plug-ins".