In early Unix kernels the approach used to avoid race conditions in kernel code was to use a combination of the following two techniques: (i) non- preemptive scheduling of threads when they are running in privileged mode and (ii) interrupt disabling for those short sections of kernel code that manipulate data used by interrupt handlers.
(a) Explain how and why these techniques, when used together, are sufficient to prevent race conditions in kernel code for a uni-processor operating system.
(b) Explain why it is insufficient to use one or other technique in isolation. (c) Explain why the approach of using them together is still not sufficient for a shared memory multi-processor operating system, and describe a technique that would solve the problem in a shared memory multi- processor operating system.