sisa-assembler

This is an assembler for the SISA assembly language, written in Rust!

Motivation

• Challenge myself to develop an assembler
• Provide a tool that I missed when I took the IC subject to current students

Installation

Execute the following commands:

git clone https://github.com/rdvdev2/sisa-assembler.git
cd sisa-assembler
cargo install --path .

If you haven't done it yet, add .cargo/bin to your $PATH

# At the end of your .bashrc / .zshrc
export PATH=$PATH:$HOME/.cargo/bin

Usage

Run sas -h to see the program help:

The SISA assembler by rdvdev2<me@rdvdev2.com>

Usage: sas [OPTIONS]

Recognized options:
    -i, --input FILE                Uses FILE as input (source.S by default)
    -o, --output FILE               Uses FILE as output (out.bin by default)

    --text-section-start ADDRESS    Places the .text section in ADDRESS (0x0000 by default)
    --data-section-start ADDRESS    Places the .data section in ADDRESS (right after .text by default)
    --auto-align-words              Automatically aligns words to multiples of 2 (disabled by default)
    --auto-align-sections           Automatically aligns sections to multiples of 2 (disabled by default)

    -h, --help                      Shows this help message

The language

The language (as well as the ISA) is defined by the documentation of the IC subject of the Computer Engineering course of the UPC. I won't include the specification here as I'm not sure about its licensing. I wrote the assembler following this specification as close as possible, but be aware that this is a personal project and as such the implementation may not be perfect. Report any issues you find!

Notes for current IC students

Here are some notes for IC students that may use this assembler in their study:

• The assembler flags --auto-align-words and --auto-align-words aren't part of the official specification, use .even instead
• The assembler puts the .data section immediately after the .text section by default. Ensure that this is the desired behaviour before assembling. If it isn't, check the program help to relocate the sections.
• Literals are always interpreted as signed twos-compliment values. This means that you can write .byte 0xFFFF and the assembler will interpret it as .byte -1, effectively translating a word into a byte. This is possibly not desirable when writing programs for your assignments, and you should avoid taking advantage of this feature.
• The instruction NOP may not be accepted in your assignments. However, you shouldn't need it, because it just does nothing. If you use it, take note that it can be codified using any invalid opcode. In the case of this assembler, NOP is always codified as 0xFFFF.

Project status

This is a (somewhat loose) roadmap of the project. Take it with a grain of salt, I may not implement everything in the list!

• Specification compliance
• Section relocation
• Error detection
• Warnings
  • Instructions in .data
  • Raw data in .text
  • Unused symbols
  • Unaligned data
  • Unaligned instruction
• Help messages
  • Use lo() and hi() to load words
• Indicate span of non-syntactic errors
• --enforce-ic-compliance flag
• Macro support
• Tests