Skip navigation.
Home
mattfleming.com
Home

WSAD 5.1.2 server started in debug mode crash (GPF in j9dbgi21.dll)

Submitted by Matt Fleming on Wed, 2006-11-15 17:06.

I hadn't used WSAD in debug mode in quite some time. I always found that the debugger was a memory hog and quite slow. On a whim, I clicked the enable hot method replace option in my test server and lo and behold the debugger was running as fast and as lean (*cough*) as the normal runtime! Then it broke...

enable fast debugger

Everything was working great and after two days I had given up using System.out.println() statements (or trace message logging) and was only running my test server in debug mode. I was even using the debugger in JSPs with moderate success. My brain wanted to cry when today I started up the server and got: 


Unhandled exception
Type=GPF vmState=0xffffffff
Target=20030915 (Windows XP 5.1 build 2600 
Service Pack 2 x86)
ExceptionCode=0xc0000005 ExceptionAddress=0x10401ca1 
ContextFlags=0x0001003f
Handler1=0x111016c0 Handler2=0x10019730
Module=C:\Program Files\IBM\WebSphere Studio\Application
 Developer\v5.1\runtimes\base_v51\java\jre\bin\j9dbgi21.dll
Module_base_address=0x10400000
Offset_in_DLL=0x00001ca1
EDI=0x04849f18 ESI=0x01ce0d69 EAX=0x0540acd4
EBX=0x0013613c ECX=0x070ebee8 EDX=0x00000001
EBP=0x0453a378 ESP=0x047ef96c EIP=0x10401ca1

You gotta be kidding me a GPF?!?

Solutions

The last resort option is to enable suck == true by unchecking the hot mode replace option. Everything will work well because the j9 vm option and dll won't be used. But you might as well go get some coffee or take a bathroom break everytime you start the server up, or resort back to using logging.

The other solution I have cannot be confirmed, as I cannot reproduce the GPF again.. even after changing my settings back. What I did was enable servlet reloading in the web.xml (extensions tab). I had disabled it the day before, thinking it wasn't all that necessary. After I did this no more GPF and happy programming again. The other thing I was messing with at the time was adding source and javadoc to some supporting jars.

I'm not the only one struggling with this intermittent bug and was hoping to have a good scientific solution. I guess I just got lucky though.

»