The Duck Machine technology stack: A CPU simulator, its assembler, and a compiler the language Mallard
This repository supports three projects in CIS 211 at University of Oregon.
In this project, a simulated CPU interprets Duck Machine instructions. Duck Machine machine language instructions are encoded as 32 bit integers.
Students will work in two subfolders
- instruction_set : The (binary) instruction encoding for Duck
Machine processors. The instruction set definitions are provided
in
instr_format.py, but you must writebits.pyto support definition and manipulation of the bit fields in instruction words. - cpu : The CPU simulator. This simulator operates similarly to a hardware CPU chip. On each instruction cycle, it loads an instruction from memory, decodes it (again using bitfields), and then executes it.
Further instructions are in docs/HOWTO-bitfields.md and docs/HOWTO-cpu.md. A fuller description of the Duck Machine instruction set architecture (ISA) is in docs/Duck_Machine.md.
An assembler is a simple translator from an "assembly language" to a machine language. Each instruction line in assembly language is translated to a single machine instruction. There are two main tasks in translating assembly language to machine instructons:
- Resolving adresses (e.g., determining the "target address" for a load, store, or jump instruction)
- Encoding instructions into the binary representation of the machine language
Most of the encoding is provided in "asm_encode". Students are responsible for creating "asm_resolve", which implements the first task.
A compiler translates from a high level language like Python to a lower level language, often an assembly language. The higher language is called the "source language" and the lower level language is called the "target language". Mallard is a high level language for Duck machines. The source language for this compiler is Mallard, and the target language is Duck Machine assembly language.
A compiler is typically composed of a "front end" that reads program source code and builds a tree representation and a "back end" that translates the tree structure to the target language. (Some compilers also have a "middle end" that analyzes and transforms the tree in various ways, but ours will not.) The front end is provided. Students will build the back end.