[OpenJDK 2D-Dev] Deadlock: sun.font.FontManager.findFont2D (6u18b07)

Janda Martin jandam at crcdata.cz
Thu Jan 21 07:03:08 UTC 2010


Yes it looks like same problem. My bug report Review ID is 1700023. I can not edit or add more information to my bug report.
Its strange to me that Bug: 6892485 has "Low" priority. According to internal rules in our company deadlock has "Critical" priority because it can lead to loss of data. (application must be killed)

I'm sorry that I don't have time to make sample to reproduce this problem. Maybe later.

  Thank you for your time
    Martin

----- Original Message -----
From: "Phil Race" <Phil.Race at Sun.COM>
To: "Janda Martin" <jandam at crcdata.cz>
Cc: 2d-dev at openjdk.java.net
Sent: Thursday, January 21, 2010 12:28:37 AM GMT +01:00 Amsterdam / Berlin / Bern / Rome / Stockholm / Vienna
Subject: Re: [OpenJDK 2D-Dev] Deadlock: sun.font.FontManager.findFont2D (6u18b07)

I believe this is the same as http://bugs.sun.com/view_bug.do?bug_id=6892485
"Deadlock in SunGraphicsEnvironment / FontManager"

-phil.

Janda Martin wrote:
> Today I discovered deadlock on linux when initializing fonts. I tried to run World Wind for Java demo applet.
>
> I reported bug to bugs.sun.com. I don't think there will be fast response/fix.
>
> JConsole dealock stacktraces are in attachment. There are several problems.
>
> *) synchronization in "sun.java2d.SunGraphicsEnvironment.loadFonts(SunGraphicsEnvironment.java:425)- locked java.lang.String at 39ea58"  against String that can be "intern"
>
> *) several lines above "loadFonts" is simple getter with description that same variable used for synchronization can be <null>  -  !!! synchronization can't be with null monitor !!!
>
> *) SunGraphicsEnvironment.loadFonts in synchonized block calls native sun.font.FontManager.getFontPath - that can lead to deadlock according to "Sun" information
>
> *) there is waiting awtLock on javawsApplicationMain and AWT-EventQueue-0 is trying to SunGraphicsEnvironment.loadFonts
>
> I hope that this help you fix this problem.
>
>   Martin
>
>   




More information about the 2d-dev mailing list