Tracing Program Side-effects within the RHODOS Environment — A Case Study *

If a program behaves abnormally, or exhibits logical faults, a debugger can be an invaluable tool to determine the exact cause of the error. However, there can be occasions when a debug ger is of marginal use. If a running program corrupts an area of memory that will be used later, then the program may terminate in a procedure which is logically correct. It is then necessary to identify how the memory was corrupted and then fix the problem code. Most debuggers are unable to operate in a manner that will identify the problem code. Furthermore, when a program is run through a debugger, the fault may not appear. This paper, details a technique in which using features of the M68020 processor and the RHODOS Microkernel, it is possible to quickly isolate and correct program faults which corrupt memory.