System Administration Guide
Appendix D, Using the crash(ADM) diagnostic tool

Studying a system panic

Studying a system panic

A system may crash if the kernel encounters an illegal condition from which it cannot recover. In such a case, it will ``panic'' and save its memory contents to the dump device before it stops running or ``crashes''. The default dump device is the swap area, /dev/swap, but you can redefine this as described in ``Defining the default dump device''. An example of when the system might panic would be if it encountered an uninitialized pointer in a third-party device driver. The memory image of a system that has crashed due to a system panic is also known variously as a ``crash dump'' or a ``panic dump''.


NOTE: If the system was swapping when the panic occurred, the memory image may overwrite some of the pages being swapped. In such cases, not all the system information will be available to crash.

The memory dump saved after a panic contains information that can be used with any of the crash commands that are used on a live system as well as the panic and trace reports that give information about the state of the kernel when the system panicked. For more information about handling system crashes, see Chapter 13, ``Troubleshooting system-level problems'' in the SCO OpenServer Handbook.