Symmetrical Multi-Processing

One of the easiest (read, cheapest) ways to improve hardware performance is to put more than one CPU on the board. This can be done either making the different CPUs take on different jobs (asymmetrical multi-processing) or by making them all run in parallel, doing the same job (symmetrical multi-processing, a.k.a. SMP). Doing asymmetrical multi-processing effectively requires specialized knowledge about the tasks the computer should do, which is unavailable in a general purpose operating system such as Linux. On the other hand, symmetrical multi-processing is relatively easy to implement.   By relatively easy, I mean exactly that -- not that it's really easy. In a symmetrical multi-processing environment, the CPUs share the same memory, and as a result code running in one CPU can affect the memory used by another. You can no longer be certain that a variable you've set to a certain value in the previous line still has that value -- the other CPU might have played with it while you weren't looking. Obviously, it's impossible to program like this.

In the case of process programming this normally isn't an issue, because a process will normally only run on one CPU at a time^12.1. The kernel, on the other hand, could be called by different processes running on different CPUs.

In version 2.0.x, this isn't a problem because the entire kernel is in one big spinlock. This means that if one CPU is in the kernel and another CPU wants to get in, for example because of a system call, it has to wait until the first CPU is done. This makes Linux SMP safe^12.2, but terriably inefficient.

In version 2.2.x, several CPUs can be in the kernel at the same time. This is something module writers need to be aware of. I got somebody to give me access to an SMP box, so hopefully the next version of this book will include more information.