Journey to Engineering

Search IconIcon to open search

How to Setup the Icestorm Toolchain on macOS

Last updated Oct 19, 2020

# Introduction

I have recently experienced the so-called FPGA Hell: you are completely stuck on a Verilog development problem, and you have no clue where to find the bug… very, very distressful.

Luckily, this problem pushed me to explore the Verilog compilation process a little deeper. At work, we use APIO to program Lattice FPGAs, but I always feel that something is missing when using “pre-packaged” tools: what I am really doing?

In this guide, you will learn how to set up the Icestorm toolchain to program the Icestick on macOS. When you are finished, you will be able to understand the basics of the compilation process and you can start to adapt it to your particular workflow.

macOS is probably not the most friendly OS for FPGA development. It is possible to make everything work but CAUTION: it can be painful.

# Prerequisites

# Step 1 - Downloading the Toolchain

There are many ways of installing the toolchain, but I have recently discovered the fpga-toolchain page on GitHub. This project creates daily builds of the toolchain for every OS.

So head over to the project release page and download the proper archive: for macOS, you should download the one for the Darwin architecture.

If you are reading this from the future (a future in which Mac computer switched Intel CPUs with Apple ARM architecture) this last choice may not apply to your particular case!

Now you have to unzip it and finding a place on your computer in which to store the fpga-toolchain directory. Personally, I (try to) use the PARA productivity system, so I am storing this into ~/Documents/resources/fpga/.

# Step 2 - Adding to the PATH

If you explore the directory content you will find the bin/ subdirectory, which contains loads of executable. Among all those programs, there are the ones that you need to program the Icestick:

and some other utilities, all part of the Icestorm project.

Nonetheless, we need to tell the OS where to find all this program. To do that, you need to modify the .bashrc file in your home folder. Open Terminal, type nano ~/.bashrc and add the following line to the bottom of the file

# Add FPGA toolchain to path and add GTKwave
export PATH="/Users/<user>/<your-storage-path>/fpga-toolchain/bin/:$PATH"

You need to fill <user> and <your-storage-path> with the proper values. In my case, this second value is equal to Documents/resources/fpga/. In the end, type ctrl + X and Y to close and save the file, then re-launch the Terminal app.

# Step 3 - Install FTDI Drivers

You have now installed the tools needed to program your Icestick. Unfortunately, since the board USB interface is managed by an FTDI chip, you need to install FTDI drivers.

To do that, head over to this download page, click on the link reported in the table corresponding to the row Mac OS X 10.9 and above and install the drivers using the downloaded .dmg file.

# Solving the FTDI Drivers Bug

Unfortunately for macOS users, FTDI drivers have a bug which can prevent the detection of the Icestick. The bug and its resolution is reported on the Icestorm project page, but you have to apply these steps every time you reboot your system.

Since this can be tedious and hard to remember, I find that the best way is to create an alias to quickly apply all the steps at once, when needed.

To do that, open the Termianl app and type nano ~/.bashrc again. At the bottom of the file, type the following lines:

# Load and unload FTDI
alias loadftdi="kextstat | grep FTDIUSBSerialDriver && sudo kextunload -b com.FTDI.driver.FTDIUSBSerialDriver && kextstat"

Again, type ctrl + X and Y to close and save the file, then re-launch the Terminal app.

From now on, in your command line you can type loadftdi, input your system password when required, and solve the FTDI driver loading problem.

# Step 4 - Install Other Useful Software

Technically, all the software needed to build and program the Icestick is now installed. Nonetheless, some other tools can be useful when developing code in Verilog.

To install the additional software, you should open Terminal and type

brew install icarus-verilog
brew install make
brew cask install gtkwave

You have installed the following program:

Unfortunately, also this software require some care to work properly.

# Using the Correct make

It can happen that for some reason you already have make installed on your system. You need to tell macOS which one to use, and to do that you need to update the PATH variable.

So open Termianl, type nano ~/.bashrc and add this line to the bottom of the file:

export PATH="/usr/local/opt/make/libexec/gnubin:$PATH"

The exact path that you must add may change over time. Luckily, brew output after the installation will tell you exactly which is the correct string that you need to paste!

Again… type ctrl + X and Y to close and save the file, then re-launch the Terminal app. To test that you are using the right software, type which make in the Terminal and check that the output match the string you placed in the ./bashrc file.

# Installing gtkwave CLI

For some reason that I do not understand, gtkwave does not work properly when invoked throught the command line or using make.

I have spend so much time trying to make it work, that now I do not really remember all the steps I have made. Nonetheless, this resources can help:

  1. follow this guide to install gtkwave command line tools;
  2. follow this other guide when changing the Makefile used to compile your program. In particular, read the Adding Alias and Makefile paragraphs.

# Step 5 - Testing the Installation

The purpose of this article is not explaining how make works and how to use Makefile to compile your project. Nonetheless, if you are here I bet that you are interested in FPGA development and you want to try if the toolchain works.

To do that, I suggest that you follow these steps:

  1. using git, clone this awesome set of tutorials by Obijuan.

    • By the way, you can use these tutorials documentation to learn FPGA development. Some of them are in Spanish, but Google Translate is magic.
  2. reach the first project using cd <your-clone-directory>/open-fpga-verilog-tutorial/tutorial/ICESTICK/T01-setbit

  3. using a text editor, change the content of the Makefile in this way:

    • add gtkwave:=/Applications/ in the first line
    • apply the following changes to the simulation stage
    #-- Ver visualmente la simulacion con gtkwave
    $(gtkwave) setbit_tb.vcd setbit_tb.gtkw
    • apply the following changes to the building stage
    #-- Sintesis
    yosys -p "synth_ice40 -top setbit -json setbit.json" setbit.v
    #-- Place & route
    nextpnr-ice40 --hx1k --pcf setbit.pcf --json setbit.json --asc setbit.asc
    #-- Generar binario final, listo para descargar en fgpa
    icepack setbit.asc setbit.bin

Keeping fingers crossed, type make in the command line. Then start debugging what is not working (PAINFUL, remember?)

# Conclusion

In this article, you set up the toolchain needed to program the Icestick (and other Lattice FPGAs). You had also your first contact with build systems (i.e. make), which I hope to explore more in future articles. Now you can start developing FPGAs like a pro!