Assignment: Context Switch and Non-Preemptive Scheduling

Context Switch and Non-Preemptive Scheduling

This assignment is an early Xinu assignment allowing the student to grow a more firm understanding of how the basics of an operating system work. This assignment is part of the Student Built Xinu track for professors that are Teaching With Xinu. This assignment should be completed in teams of two. Students should only turn in the files required for this assignment: system/create.c, system/ctxsw.S, and a README file, as explained below.

Preparation

Untar the new project files in a fresh working directory:

   tar xvzf <tar-ball location>

Copy your synchronous serial driver file (kprintf.c) into the system/ directory.

Context Switch

The new source files include two new headers, include/proc.h (definitions for Process Control Blocks and a process table), and include/queue.h (definition for a process ready queue), as well as trivial updates to include/kernel.h.

The new source files also include:

system/initialize.c				Updated initialization for Project 4.
system/main.c				A "main program" for testing scheduling.
system/queue.c				An implementation of the queue data structure.
system/create.c				A partial function for creating a new process.
system/ctxsw.S				An incomplete assembly routine for switching process contexts.
system/ready.c				A function for adding a process to the ready queue.
system/resched.c				The primary scheduling code, equivalent to yield().
system/getstk.c				A rudimentary function for dynamically allocating new stacks for new processes.
compile/Makefile				Updated rules for compiling XINU.

The create() and ctxsw() functions are incomplete and must be filled in. The file system/initialize.c contains code to test your context switch with three processes, each of which prints a process ID and then yields. Once your context switch and creation functions are working, you will see these three processes take turns running.

Some Assembly Required

An operating system's context switch function typically must be written in assembly language because the low level manipulations it performs are not modeled well in higher-level languages. If you have not worked in Mips assembly language before, there are many helpful resources available online. Despite its low-level nature, a context switch does not require complex instructions. Our context switch can be completed using only arithmetic opcodes, and the load (lw), store (sw), and move (move) opcodes.

README

There are many possible layouts for your operating system's activation records. It is critical that your create() function and your ctxsw() agree on the design you choose. Turn in a plain text file, "README" that diagrams and explains the activation records created by create() and expected by ctxsw(). Label the purpose of each field. This design document will be worth 25% of the assignment credit.

Testing

The default test case provided with the tarball is necessary, but not sufficient. Just because it switches between a handful of identical processes does not guarantee correctness. In embedded systems, details matter. (A LOT!) Students in previous terms have found that subtle bugs in this phase of the term project were responsible for nightmares weeks and months down the line. Test your code thoroughly:

• Processes can be passed an arbitrary number of parameters upon creation;
• Processes should terminate cleanly upon exit;
• Stack pointers and activation records should align properly and have known contents at critical check points.

You have a working I/O driver, a kprintf() function for formatted output, and code from earlier assignments for parsing integers. Use these to explore every aspect of the operating system structures you are building.