Stack overflow conditions on Linux

A stack overflow on Linux® can be identified by a SIGSEGV error, although the Java™ virtual machine (VM) might stop unexpectedly without displaying any information on the console.

If you get a SIGSEGV error when there is not a stack overflow, the diagnostic information is written to the console as in the example in: Scenarios in which dump files might not be produced

When a stack overflow has occurred, these are the two most probable outcomes:

A core file (core.pid) has been produced in the working directory of the Java process.
No core file was produced, or the core file is truncated.

In the first scenario, the core file should contain the full set of threads for the VM and can be used with the dump viewer to retrieve the missing diagnostic information. For more information about the dump viewer, see Dump viewer in the OpenJ9 user documentation. Core files will be produced in the working directory of the process at the time of the crash unless otherwise configured at a system level.

If no core file was produced, you should check the system settings to ensure that core file generation is enabled and then rerun the application. For more information, see Configuring your system in the OpenJ9 user documentation.

If the system settings already have core file generation enabled but no core file was produced, the problem is not a stack overflow. Production of core files in a stack overflow case should be identical to a SIGSEGV error when running with the -Xrs or the -Xrs:sync option. For more information, see -Xrs in the OpenJ9 user documentation.