RVI

A simple assembler for the RV32I instruction subset. This project in no sense is aimed at being a full assembler and the primary motive for developing this is so that it serves as an aid for the hardware development project we are doing. This assembler does not support any form of stack manipulation, space allocation and other essential features of any regular assembler and does not intend to support these 'essential' features in the near future. My primary motivation for building this assembler was to make one to one mappings between assembly level coding and the corresponding binary level coding. This is particularly useful when simulating or testing a new RV32I hardware design that only partially supports the instruction set.

Here you can perform the one to one mapping as well as see the tokenized values independently to verify whether your processor is treating the binary values as it ought to. Further, simple self contained programs can be run - those programs that do not assume a stack setup or a external call to its main or start label.

The assembler aims to allow for selectively turning off instruction and easily modify instruction opcodes - a feature currently not fully supported.

Supported Instructions

Sl. No	I Type	R Type	U Type	UJ Type	S Type	SB Type
1	addi	add	lui	jal	sw	beq
2	slti	sub	auipc		sb	bne
3	sltiu	sll			sh	blt
4	ori	slt				bltu
5	xori	sltu				bge
6	andi	xor				bgeu
7	slli	srl
8	srli	sra
9	srai	or
10	jalr	and
11	lw
12	lb
13	lh
14	lbu
15	lhu

Installation

Step 1:

The assembler works on Python3. Plese create a python3 virtualenvironment in your system. For linux systems with virtualenv installed, this is as simple as running

virtualenv -p python3 rvi

This creates a new virtual environment named rvi. For Windows, Anaconda, or other environments, plese refer to your environment specifc instructions.

Step 2:

After activating the environment created in step 1, install the requirements specified in requirements.txt. In src/assembler, run

pip install -r requirements.txt

Usage

In the simplest case, the usage is as follows. cd to the src/assembler directory. To assemble a input file INP.rvi, type

$ python rvi.py INP.rvi

where the .rvi extension has no special meaning, and any type of file will actually do.

This will create a output file named a.b with the binary code. If you want to specify a custom output file name, please use the -o OUTFILE option as follows.

$ python rvi.py INP.rvi -o OUTFILE

This will write the output to OUTFILE instead of a.b.

By default, the output is written in binary - one instruction per line. If you want the output to be in hex rather than in binary, please use the -x flag as follows.

$ python rvi.py -o OUTFILE -x INP.rvi

More options can be viewd in the below help text which you can obtaining by

$ python rvi.py -h

usage: rvi.py [-h] [-o OUTFILE] [-e] [-nc] [-n32] [-x] [-t] INFILE

RVI 0.1 - A simple RV32I assembler developed for testing RV32I targeted hardware
designs.

positional arguments:
  INFILE                Input file containing assembly code.

optional arguments:
  -h, --help            show this help message and exit
  -o OUTFILE, --outfile OUTFILE
                        Output file name.
  -e, --echo            Echo converted code to console
  -nc, --no-color       Turn off color output.
  -n32, --no-32         Turn of 32 bit core warnings.
  -x, --hex             Output generated code in hexadecimal format instead of
                        binary.
  -t, --tokenize        Echo tokenized instructions to console for debugging.

Programming in Assembly

The language I have designed for is similar to MIPS assembly and I have included some examples in the examples/ drectory. Each statement should be in one line and the new line character is used to mark the end of a statement.

WARNING I have not tested how the assembler behaves in Windows line endings. I have tried to make sure this does not become a problem but in case you face any problem, please use an editor which supports UNIX file ending like notepad++ or sublimetext and raise an issue regarding the same.

TODO

• In the parser, for each statement, check if all the tokens are valid
• In parser, move immediate checking to different function (I Type)
• recheck if the all type encoding is correct [PRIORITY]
• Write test suit
• Support labels and label based jumps
• Test with windows
• Support turning on and off of instruction selectively
• Implement jalr
• Implement remaining instructions
• Improve README
• Document installation instructions
• Support include directives

Don Dennis, 2016