Community
Participate
Working Groups
This is a "random instability" bug report, because this is not reproducible. I'm using the final M8 version of WSWB, and after using one Eclipse session for a while, including debugging and hot code replace (with IBM's J9 JVM), I added a field to a class and then typed "/**" and hit Enter to fill out the JavaDoc style comment. This is on Windows, and I got an access violation: The instruction at 0x000e1107 referenced memory at 0x00000001. Starting MS Visual Studio on the javaw process showed that address 0x000exxxx is outside all loaded modules, and in fact it is a data area consisting mainly of words with the value 0x0060 and 0x0020. So obviously the problem was a wild jump into data. Some addresses at negative offsets from SP lie in the address range of NTDLL.DLL and KERNEL32.DLL, but the nearest one that makes sense is in ZIP.DLL and that doesn't seem very likely.
Moving to SWT (Styled Text got recoded for M8).
Can you attach the dump file?
Created attachment 9220 [details] The javacore dump file from this crash This is the javadump file from the crash.
SSQ, the trace doesn't give much insight into what was happening in main at the time of the crash. Could be J9 related?
Please could you try the latest integration build? We have fixed a segment fault on Windows, which might be the same as yours.
This problem is not repeatable, so "trying" a new build won't help. All I can do is report if it happens again.
Please, reopen this if it happens again. *** This bug has been marked as a duplicate of 56390 ***