Kernel Debugging Feb94

This manual, "DEC OSF/1 Kernel Debugging" (February 1994, Version 2.0 or higher), is intended for system administrators and programmers to understand and troubleshoot the DEC OSF/1 operating system kernel.

The document provides a comprehensive guide to:

1. Kernel Debugging and System Crash Analysis: An overview of system crash types (hardware trap, software panic, hung system, resource exhaustion), the crash dump and recovery process, including how to save and analyze crash dumps (partial vs. full), and procedures for forcing dumps on hung systems.
2. Kernel Debugging Utilities: Detailed descriptions of the primary debugging tools:
   • dbx debugger: For examining, modifying, and displaying kernel variables and data structures, including multithreaded debugging and exception frame examination.
   • kdbx debugger: An interactive, extensible front-end to dbx specifically designed for kernel crash analysis, providing commands and predefined aliases for interpreting kernel symbols and data structures.
   • kdebug debugger: For symbolic, breakpoint debugging of a live kernel, with instructions on remote debugging setup using a serial line and build/gateway systems.
   • crashdc utility: An automated tool for collecting critical system information from crash dumps or a running kernel.
3. Crash Analysis Examples: Practical, commented examples demonstrating how to use dbx and kdbx to identify software panics, hardware traps, and debug kernel threads.
4. Writing Extensions to kdbx: Instructions for developing custom kdbx extensions to extract specialized information, covering considerations, standard kdbx library functions, data types, examples, and the process for building and debugging these extensions.
5. System Exercisers: Information on utilities (fsx, memx, shmx, diskx, tapex, cmx) for exercising and troubleshooting various system components like file systems, memory, disk drives, tape drives, and communications systems.

In essence, the manual equips users with the knowledge and tools to diagnose kernel issues, analyze crash dumps, and extend debugging capabilities for the DEC OSF/1 operating system.