Level 6 of 12 · about 3 weeks
Memory management
How each process gets a private address space: protection, allocation, paging, page tables, the TLB, demand paging, page replacement and thrashing.
Modules
Core: everyone needs it. Common: most jobs use it. Specialist: optional depth, safe to skip on a first read.
1
Address spaces and protection
What a process sees as its memory, how hardware turns its addresses into real ones, and how one process is kept out of another's memory
Core 6 parts · about 4 hours
2
Contiguous allocation and fragmentation
What happens when each process needs one unbroken block of memory, and why the free space breaks into useless pieces
Core 7 parts · about 5 hours
3
Paging and address translation
How cutting memory into equal pieces removes external fragmentation, and how every address is translated through a page table
Core 7 parts · about 6 hours
4
Page tables, multi-level, hashed and inverted
Why one flat table per process is too big, and the layouts that keep the table small
Core 7 parts · about 5 hours
5
The TLB and effective access time
How a small cache of recent translations hides the cost of page tables, and how to compute what is left
Core 6 parts · about 5 hours
6
Segmentation, protection, sharing and copy-on-write
Regions with their own rules, pages that many processes share, and the trick that makes fork cheap
Common 6 parts · about 5 hours
7
Demand paging and page faults
How a program runs with only some of its pages in memory, and what the cost is when a missing page is needed
Core 7 parts · about 5 hours
8
Page replacement algorithms
Which page to throw out when memory is full, and how FIFO, LRU, the optimal rule and the clock compare
Core 7 parts · about 6 hours
9
Frame allocation, working sets and thrashing
How many frames each process should get, why too few can bring the whole machine to a halt, and how to prevent it
Core 7 parts · about 5 hours