Summary

This chapter began by overviewing all the concepts involved in memory management. We then explained the implementation of each concept. The first concept we looked at was pages, which is the basic unit of memory managed by the kernel and how pages are kept track of in the kernel. We then discussed memory zones as memory partitions that are subject to limitations from hardware. We followed this with a discussion about page frames and the memory allocation and deallocation algorithm that Linux uses, which is called the buddy system.

After we covered the basics of page and page frame management, we discussed the allocation of memory sizes smaller than a page, which is managed by the slab allocator. This introduced us to kmalloc() and the kernel memory allocation functions. We traced the execution of these functions down to how they interact with the slab allocator. This completed the discussion on the kernel memory management structures.

After the kernel management structures and algorithms were covered, we talked about user space process memory management. Process memory management is different from kernel memory management. We discussed memory layout for a process and how the various process parts are partitioned and mapped in memory. Following the discussion on process memory management flow, we introduced the concept of the page fault, the interrupt handler that is in charge of managing page misses from memory.