views:

70

answers:

1

Hi all, An eclipse plugin for IBM RSM launches a java process to make some computations (realized in C/C++ via JNI). The problem here is when the process is started, it crashes...but only on windows. Moreover, if the process is started using Oracle/Sun jvm, it works fine.

Under linux, java -version gives

java version "1.6.0"
Java(TM) SE Runtime Environment (build pxi3260sr7-20091215_02(SR7))
IBM J9 VM (build 2.4, JRE 1.6.0 IBM J9 2.4 Linux x86-32 jvmxi3260sr7-20091214_49398 (JIT   enabled, AOT enabled)
J9VM - 20091214_049398
JIT  - r9_20091123_13891
GC   - 20091111_AA)
JCL  - 20091202_01

and under windows:

java version "1.6.0"
Java(TM) SE Runtime Environment (build pwi3260sr7ifix-20100202_01(SR7+IZ67488))
IBM J9 VM (build 2.4, JRE 1.6.0 IBM J9 2.4 Windows XP x86-32 jvmwi3260sr7-20091125_47905  (JIT enabled, AOT enabled)
J9VM - 20091125_047905
JIT  - r9_20091123_13891
GC   - 20091111_AA)
JCL  - 20100202_01 

My question here:

Anyone knows what are the differences between these two jvm, and eventually could tell me the reasons of this crash ? Is there a bug to report to IBM ?

I can post a dump of the crash if necessary

EDIT: After some work, we have replaced JNI by JNA, and everything works fine...

A: 

Can you post the dump and your interface code?

I think examining your JNI code will help you figure out what's going on; J9 is brittle to begin with, JNI is most likely destabilizing it somehow.

Brandon
Thx, but I don't know what part of the dump I should post (it's about 150KB...I don't think it's readable to post the complete file).For the JNI interface, I've about 50 files...Your understand now why we try to find another way (ie JNA) to solve the problem, instead of trying debug the crash... (by the way, to build our solution for 64 bit platform is very easier :) )
Sylvain M