Hello, and thank you for joining us today. I came at Droidcon every year in the last three years showing the progress that we had on our open source mobile phone. And today, we will give you the latest status and what Jolla has managed to do with Sailfish.

So I will start with presenting myself. I'm a software and hardware engineer. I started working on Sony devices in 2006. I created the Free Xperia project in 2010 with the ambition to create an open source phone

or to provide an open source alternative. I joined Sony as an employee in 2014 when the open devices program has been started. And I'm working as a community manager and the software architect for this program. Our wish with this program is to provide an innovation

and development platform built by community, for the community, and to provide support for as many devices as possible using the latest available sources. In this case, we support the 4.4 kernel, which is fairly new

and highly secure. Everything is built on one software branch. We use one kernel for all supported devices and one camera framework. This helps us a lot because when you try to do the things in a standard way, you have to have zero patches. And zero patches is our goal on top of Android.

Even if we do not support all the fancy features, all the basic functionality is there for GPS, Bluetooth, GSM, Wi-Fi, NFC, camera, and fingerprint. Here is our software structure. And as you can see, the open devices program

does not touch the Android layers. This approach allows us to have really easy porting on the new platforms, which means that the whole bring up for Android 8 was in less than seven days. This is how we structured the project.

You have one Linux kernel, which is common for all devices. Then you have the device configuration, which is public on GitHub. You have the vendor part, which contains the vendor binaries and the open source projects that we manage. And then you have the Google ASP,

which is not touched in any way. With our unified kernel, we provide the 4.4 kernel for Android 7 and Android 8 for all devices that are listed on the left side, which means all devices that we

produced from 2015. I will come back later with lessons learned, what happened, and how we hope that things will improve. This is the current status for the 4.4 kernel. And as you can see, the only not working feature is the camera.

Since the things cannot be always open source, there are parts on a stack that are still vendor and proprietary, and others that have been open sourced. Here is vertical through all the layers on one subsystem.

As you can see, the camera, as an example, still has the framework as closed source, but the kernel driver and the hull, the application, and the services are open source. The same thing happens with all the others.

I'm coming back to the open source camera solution, which has been designed for this project. We are not using the Sony advanced camera. We are using the Qualcomm framework, which has been adapted for our needs, and we implemented the support for our sensors.

We organized the project so that other devices, new devices, can be easily added to the project. And also, we created a custom configuration that allows you to push your own changes in one place and inherit them in all other projects.

That means that when you build for one device, you get support for all devices that are supported by the same tree and kernel. How to start working with this is really simple. You pick one of our devices if you

want to be on the latest Android, or an older device if you want to be on older Android versions. You unlock the bootloader, and then you follow our how-to. We provide all the information on our GitHub, and we try to be as open as possible by providing both knowledge and by providing support

on our forum. And also, for the ones that want to be a little more into the hardware side, we provide information how to connect the UART debugger on our devices. Unfortunately, you have to dismantle the device

to access the port. But when you're a real hacker in the hardware layer, or you write kernel drivers, you will do that. And now what we learned, because it's all about lessons learned, and we learned a lot. Unfortunately, we learned that all hardware cannot be supported

on the newer kernels, just because there are some binaries. And those binaries are no longer compatible either with the newer kernels, either with the newer Androids. So unfortunately, I'm sad, but I have to say, we had to drop some devices.

And we also had to drop the old kernels, because all the old kernels are end-of-life. And end-of-life means that there are no new features and no security updates. So we decided that we should move all the possible devices

to the new kernel and then let the other ones be there in the open source world if somebody wants to play with them. So you can take any of the old devices built by Sony from 2014 and onwards and use the old kernel

in your own custom project. Maybe you want to do repurposing. Maybe you want to use the motherboard on an IoT project. This is the status for the 3.0 kernel. And I would say that it is a really good status. We know that there are some small bugs, but all those devices can still live, even if we do not

support them any longer. This is how Android support was looking, starting from Android 5.1 and Android 6. As you can see, in Android 5.1, there was a lot of hardware support missing.

And that hardware support has been added in Android 6. But in Android 7, hardware support for all devices has been removed. And to add it back, people will have to do some nasty hacks. Why we do this? We do this as an innovation platform,

but we also value what others can do. So we had custom ROMs built on top of our open devices. As you can see, the open devices is in the lower layers. So what comes on top, it can be the plain ASP, or it can be the old Cyanogen. It can be a paranoid, or it can be only ROM.

Any custom ROM or any custom project based on Android can be based on our core. But we can think even more outside the box. And we saw that other companies became interested in innovation.

So we have Sailfish that has used the open devices. And they built Sailfish on our devices. We have Ubuntu, which has built Ubuntu on our devices. We had Mozilla, which built Firefox. And the whole project is open source.

In 2015, Firefox was proud to announce that they based their open source initiative on our devices. And then in 2016, Ubuntu has published the images at Mobile World Congress for our XPI Z1.

In 2017, YOLA has announced Sailfish for our devices. And they, for the moment, support only the Xperia X device. If you want to get hold of me, those are my contact points. You can contact me at any time if you have questions.

And I'm happy to answer questions now. And after your short session of questions, I will give the microphone to our guest, Andy, from YOLA.

Hi. So Google announced the project Treble, or House Code. I'm not sure. So having a unified base, and you can always put new Android on top. Is it somehow interesting for you to have that?

Will it be possible to support phones in the future longer than it was right now? Because then you don't have to switch the kernel. In the previous slides, I already

showed the HIDL layer, which is untouched by us. And the problem is that we do not want to support devices that cannot use the latest kernel. So it's not about just supporting devices. It's about supporting all devices in one kernel.

What we did is similar with what you do when you get a new kernel for a laptop. You build one image, and that image will basically work on any laptop. With our project, you can do the same. You get the kernel source, you build it for one device from the list, and it will work if you build for another without doing any change to the source.

So you have one kernel, one repository, one source for everything. But then you probably can update Android longer when you stay on the word kernel, because now it's abstracted more. Did I get it right from the Android? Not directly from Sony.

We support only the latest, but since the vendor binaries and the kernels are public, anyone can take an Xperia Z1, as an example, in this moment, with the old 3.10 kernel, which is considered stable. And they can port O, but it's not supported by Sony. So we can provide support only for the devices that are

actively maintained and tested by us. So what device would you recommend right now as the best device from Sony? Any device from the X or XZ or XZ1 family is good. And also, we will publish the build guide for Android 8,

because that's the hot topic this week. Thank you. You're welcome. Other questions? Andy, welcome on stage.

Hello. Are we on? Good. Hi. I work for Jolla, who maintains Sailfish.

I don't know how many people are familiar with Sailfish. I'll give a quick history of it. It's a mobile Linux, so it's sort of more in common with a desktop OS than other mobile Linuxes.

Linuses. It's glibc-based. Its history goes back to what Nokia were playing around with the N900, the MIMO and the MIGO and all of that. And a lot of those guys, when Nokia went all Windows, decided to go and form their own company to try and make their own, to continue the mobile Linux effort they were doing. The big difference between Sailfish and other ones

is that we use a lot of common open source components. So underneath the shell, you've got the systemd runs everything. You've got package manager is RPM. And you've got libzip on top of that. We use package kit. Then things like Dbus, Blue Zed, Pulse Audio, Gstream.

It's all common stuff that you'll find on Linux desktops as well, which makes it a bit easier to sort of play around with different software. It's more of a sort of a mobile computer experience than sort of an ordinary handset, which might be a bit more like a console, a games

console a bit, really. The big difference is that root is something that you deserve because it's your device. It's not something you get. You don't root a Sailfish device. You don't jailbreak it. You go into the settings, and there's a developer mode. And there's a little text field. And you write what you want your root password to be.

And then you are a root on your device. And of course, because that's by design, then applications can't really complain about a device being rooted because the concept doesn't really exist in Sailfish. On top, it's based on those Wayland and everything's QT. So we have a very nice UI framework that's

QML-based called Silica. That's not open source yet, but it's planned that this will eventually be opened once we can convince the people who paid for it to be written that that's a good idea. Even so, the bits of it that aren't strictly open source, because QML is basically a text-based markup-y thing,

then all of the apps that you find in Sailfish are all in plain text on this file system of the device. And you have access to the entire file system. So you can edit the apps if you want to. You just go in there with open the terminal, SSH in, and edit stuff and run it. There's even QML Live so that the apps will update with your new code as you write it.

You can hack about and tinker with this stuff. We do have an Android compatibility layer, because a new mobile OS without any apps these days is not really worth bothering with. This is called Alien Dalvik. And it's aging a little bit now. We have to update it. It supports the API 20 4.4.

But most stuff that you need, like your WhatsApps and that, all runs with that so far. But there are plans to bring that a little bit more up to date. And that, of course, can see your contact lists and your notifications. So it works well, but not as good as native apps, as I told the here maps people earlier. We need one of those. So to build it, we have what we call the Hardware Adaptation

Development Kit, which sounds very, very good. But it's basically a PDF that tells you which tools to use and where to get them from. To start with an Android BSP, traditionally we've used a Cyanogen one or even AOSP. And then we have certain patches on top of it.

Because we have this, the biggest sort of little thin layer, but massive hack, is that everything underneath is built with a bionic C library. And everything on top of it, all of our userland stuff, is libc. So we have this thing called Hybris, which basically puts a little layer in, which means that doesn't matter anymore.

You don't need to worry about it. So we have lots of stuff that makes all the userland and even the higher system stuff believe that it's running on a glibc system when actually underneath it's all bionic. We have a lot of people who enjoy hacking about at this stuff. There's an IRC channel on Freenode called selfish-os-porters. And they are always getting new devices

and finding sources for them and porting devices and hacking away with massive code hammers. They have a really nice time. And you can put all that together, run through the guide, and then flash your image and boot Sailfish and run things. So, the Xperia that we are just about to launch.

We've done a port, a first port we concentrated on one phone, which is the Xperia X. There it is. My kids. That is kind of ready. The official launch date is on the 11th of October.

But at the moment, as of about five minutes before this presentation started, our hardware development team have pushed the required sources to do it yourself onto GitHub. We started with the X, but we'll hopefully quickly add later devices after that.

There is a sort of a sales thing that we have with it because we like to, we need to get some cash in. How we're structuring this is that the OS itself is free. You can build it yourself. You can get the sources and you can build it. But if you want to access the store, which includes the APK compatibility layer

and certain other premium things that we don't own that are proprietary, then you need to pay, I think, 50 euros a year for access for a device for that is what we're asking at the moment. But this is very much an experiment and we'll see how it pans out, how popular it is. This is the first time we've had a device available that people can run in the United States. So, we've had some demand and there's a lot of Americans,

so we'll have to see how that pans out. So, the repo manifest, which is what you need to build it, is available now at that address there. And we've got a special branch that's just been pushed this morning which supports the Sony AOSP.

We will be adding new devices. Our developer community, our porting community has been eagerly awaiting this release and I think we're gonna be getting patches to add other devices extremely quickly. You will find the hardware adaptation development kit there. The main difference between building your own image now and waiting for the official release is that home porters are limited

to what our latest release of the OS version is. We try to maintain quite a fast release cycle for self issues. The target is once a month and sometimes we make that, but sometimes we end up being a bit stuck in the RC phase and the current version 2.1.1 has been probably about six months going. But now that's out, we've got faster cycle

and the target release version for what we're going to launch on Xperia is 2.3. So that means if you build it on 2.1, there may be some bugs, but actually there aren't very many bugs relating to Xperia in that release because most of the things are in the hardware adaptation that you can download there, which of course will always be current

because it's public. The things that need to be fixed, I've put Bluetooth on there but I think one of our wizards fixed it yesterday. This is our first device that's had this sort of configuration of having fast CPUs and slow CPUs. So certainly the way our tasks are allocated on System D isn't optimal yet, so we need to work on that. We have basic support for the fingerprint sensor,

but it's not hooked up to the hardware underneath yet. And NFC is something that we've not even implemented yet because this is the first device we've had that actually is supported in an NFC sensor. So there's lots to work on and everyone's going to have a lot of fun doing it. And that's the end. You can email me if you want to and you will find me on Freenode

on our wonderful Porter's room where there are many people who want to help anybody who wants to get into this stuff because it's lots of fun. And of course if anybody really gets into it, we've got jobs available as well because it is really fun. Right, does anybody have any questions?