Opening ceremony 🥳
Imagine all the things you could accomplish in ten years if each week you spend a significant amount of your time on them. You could choose a small scope project and drive it to perfection. You could probably make a lot of money, live large, get a big house and five cars. Instead of that yours truly spent his time on his favorite Linux Mobile distribution. This talk gives a look back from how it all started until now, the technical changes postmarketOS went through, how our community got bigger and better and how the mainstream operating systems got worse over time. I'll already spoil the conclusion here, what we have been building together is more meaningful than ever in 2026, I'm proud of what we have accomplished and I'm so happy I didn't do something else instead!
As a self-hosted enthusiast and a person who cares about green tech, I have recently started a journey to self host more and more of the services I use frequently.
At the last Open Source Summit Japan, I discovered PostmarketOS, a real Linux distribution for phones, which made me start experimenting with the possibilities reused hardware could bring to a more sustainable future.
In this talk, I strive to describe the process and approach I had to setup a reliable, resilient Kubernetes cluster using only old phones with PostmarketOS and k3s, as well as giving tips to find more sustainable uses for said technologies.
MediaTek devices have a vulnerability which allows us to replace LK2 (Little Kernel 2, first-stage payload) with our own. We exploit it to clean-up the boot chain and gain full control over our devices.
Despite the long-term trend to automate tests as much as possible, software developers are faced with the reality where this is theoretically possible but practically hard to achieve. One good example is the Mobile Linux distribution postmarketOS, where pre-release testing is conducted manually by device maintainers and the project's community. Promising efforts to automate the process have seen work and remarkable progress for years, yet, it is unlikely that they will effectively close all gaps, obsoleting the need for manual testing.
A key challenge for manual testing is the management of the process. Testers need proper guidance by well-designed test plans while developers and release managers need meaningful testing reports. While several tools exist for this purpose, most are too complicated and too heavy-weight for community projects. With KiwiTCMS and SystemTestPortal two open source systems are available that were designed to fill this gap. However, like any server-based system, employing these systems requires infrastructure setup and maintenance.
This talk presents the conceptual idea of "SystemTestPortal Lite", an alternative to SystemTestPortal that runs completely in the web browser without the need to install or operate services. Instead, projects only need to provide data files with the test plans in their repositories. Test reports can be uploaded as files to issue trackers or the repositories so they can be aggregated before release decisions are made.
This talk covers the reasoning behind automated Hardware in the Loop (HiL) testing of standard pmos devices.
Recap of the design process and preliminary setup of a hardware in the loop (HiL) testing rig designed for scaling into a test farm with dozens of devices.
The talk and slides will be in English.
A 30-minute lightning talk session. Present your projects and ideas!
To ensure that final releases of projects are stable and don't contain regressions, we have to start testing these releases before the actual release. That's what we're doing with nightly packages, where we build the development states of GNOME, KDE, Phosh, systemd and more, (hopefully) every night.
A brief introduction into porting Alpine Linux to new architectures
FOSS projects are often focused on technical work. However, what brings people together is rarely just technical, and very often strongly related to our emotions and ways of seeing the world. Unfortunately, we get a lot less training on communication and emotions than we get on writing kernel drivers or learning Rust. The goal of this worshop is to take a bit of time to try to collectively work in those tasks.
The current postmarketOS infrastructure team has been managing the services people use every day when interacting with postmarketOS, and services that are available for the team internally.
Let's go over everything that's running, how it's set up and some untold details.
In this talk i want to give an overview about my journey from after installing postmarketOS and then become a hobby app dev with a small community. The project is a bitwarden password manager which works on mobile devices.
From the first idea, to the first commit, to the first PR from some random person on the internet, to what we have now.
Starting off day 2!
The current architecture for building Alpine and postmarketOS packages has served us well. But it's time to reevaluate our requirements and design a new system.
Software malleability, or in other words, the idea that a user can tweak their software to fit their use with a quantity of code proportionate to the complexity of their custom setup, is really what I believe free users, as tweaking most FOSS actually often requires quite a lot of technical knowledge to be done easily, even by experienced developers... while it's widely assumed that using Emacs is a slippery slope into wanting your whole system inside it, a fantasy dream that can be lived through EXWM on Guix, that's not what this talk will be about!
I got a weird idea (of course inspired by HP Palm/LG webOS and Firefox/KaiOS): what if my window manager was actually a webview kiosk? What if I could draw native apps pixel inside <canvas>? and then the whole interface could be written as you would code a web app (and the shocking reality is that there are more web developers on earth than desktop manager ones)? and that could actually load within a second, and render at 60 fps while consuming only a few megabytes of RAM? I enjoyed experimenting with that a lot because it actually helped me draw ergonomic interfaces for devices like feature phones (I own a Nokia 8110 4G) that had none maintained for PostmarketOS (RIP Bananui), and iterate really quickly with live reloading and all the great developer tooling that web browsers offer us.
River is a non-monolithic Wayland compositor that provides a stable window management protocol rather than hard coding window management logic in the compositor process. This architecture is proven for desktop use-cases with a growing number of river-compatible window managers.
This talk explores building mobile interfaces using the river window management protocol. There will be an overview of how the core protocol works followed by a discussion of the features especially important for mobile interfaces such as touch gesture support and on screen keyboards. Finally, we will examine the current limitations and how they can be resolved in the future.
I believe that river's modular architecture can greatly benefit developers building FOSS mobile interfaces, it's a lot less work to test innovative new ideas if one doesn't have to implement an entire Wayland compositor as well!
Alpine Linux began as a small, diskless distribution for routers and security appliances. It was not designed for containers, cloud computing or smartphones, yet it eventually found its way into all three.
In this talk, Alpine’s founder looks back at five lessons from more than twenty years of building and maintaining the project: how real constraints can produce unexpected opportunities, why simplicity must include the contributor experience, how being different can pay off, why neglected work eventually becomes technical debt, and why a friendly community is part of a project’s technical quality and security.
This is not a complete history or a celebration of everything Alpine got right. It is a candid account of decisions that worked, mistakes that became apparent years later, and what it takes for an open-source project to survive beyond its original purpose.
Smartphone bootloaders must be unlockable to make projects like postmarketOS even feasable. Software like Play Protect is also anti-competitive. Let's get some attention from EU policy makers by submitting a citizens initiative.
We have already written a very rough draft, but together we can improve the text!
The postmarketOS governance has evolved through the years. In this talk we will analyze how we got to where we are, and what is laid ahead of us.
This Qualcomm SoC - code-named Kodiak - is used for many use cases. With its SC7280 variant it's used in laptops and was planned to be used in Chromebooks, the SM7325 variant is used in smartphones, QCM6490 is the IoT variant used in plenty of SBCs and some smartphones, and perhaps we'll see yet another new variant of this chip?
Let's talk about the history of this chip, with a strong eye on Linux kernel development and device bringup.
Alpine Linux is underfunded critical infrastructure, and the Sovereign Tech Fund is designed to give funding to exactly this kind of project. Let's discuss with Alpine Developers and maintainers how to prepare an application
Let's take a short look at Phosh's immutable nightly image builds, why we do them and how they can hopefully be beneficial for postmarketOS Duranium and immutable mobile Linux in general.
A 30-minute lightning talk session. Present your projects and ideas!
How hard is it to break your computer or phone? That is the question we'll try to answer here, and not with a bench press but just by tinkering with the software. Most mainstream OSes advertise themselves as being "immutable", with the claim that it should require a lot of determination for a user to break the stability of their, e.g., Android or macOS devices.
I've been using NixOS personally and commercially for more than a decade now, and it's really often a question that audiences ask: How does it compare to RPM-OSTree? to pinned Alpine Docker images? etc. and with the release of PostmarketOS Duranium earlier this year, I think there is a good opportunity to look at the diversity of strategies employed to make systems more robust and to answer questions like "what are the escape hatches we can use to still break those devices?"
We start off the last day of the conference!
In 2011, Canonical announced Ubuntu touch but failed later. And before Canonical, some companies tried and also failed.
So in this talk, I, as a user who's been using GNU/Linux on desktop computers for almost 10 years and Android/Linux since 2009, tries to give my personal view about why last time we failed and why this time we'll have a bigger opportunity to succeed.
All around the stack things improve, but what are the challenges in GNOME, and most importantly the shell? What is the current state? How did we get here? What are the changes we need to land? How will we achieve converging with upstream? Is there a way for you to get involved? Markus raises interesting questions and pretends to have answers.
What it's like to run postmarketOS day-to-day, and the various shenanigans you can get up to.
A practical look at SELinux and mandatory access control and the challenges of bringing it to Alpine Linux and postmarketOS.
Have a question about developing for the Linux kernel? Come to us and let's chat!
There is initial support for a few Virtual Reality devices in postmarketOS, including Oculus Go, Lynx R1, META Quest 3S (panther) and the upcoming Steam Frame. I also own a OnePlus 6 which runs Godot Native Mobile VR with a Cardboard-compatible smartphone holder. Combining this with a Vive Tracker and libsurvive gives you six degrees of freedom in Godot Beep Saber VR.
Trying to solve the problems of working with imperative package definitions, one crate at a time.
xiotrace hooks the XNU kernel used on certain mobile devices in order to trace I/O operations such as SPI transfers. This can be useful for understanding the protocol used by peripherals like the touchscreen.
I will present the previous work leading up to this project, my approach, and a few neat tricks I used in the implementation.
Be surprised!
A few words of farewall to all the participants