August 06, 2019
With the release of the 5.2 Linux kernel, you might have noticed that Collabora has done a lot of work with Rockchip's RK3399 platform, more specifically with the ROCK Pi 4 single-board computer.
Our ongoing work on the reverse-engineered Panfrost OpenGL ES driver for Arm Mali GPUs turns the RK3399 SoC into a very attractive platform to try out Wayland on ARM devices, especially since it's such a versatile platform that is both affordable to buy and is available in multiple form factors, including system-on-module (SoM) and single-board computer (SBC).
Collabora has been steadily contributing to enabling RK3399 support (among many others!) in upstream u-boot. Some of the most notable contributions include:
For the 5.2 release cycle Collaborans focused on optimizing various subsystems such as the Panfrost DRM driver for Arm Mali T6xx/T7xx/T8xx GPUs used in a huge number of platforms, and a V4L2 driver for the Hantro G1 video codec used in RK3288 and RK3399 devices, as well as NXP i.MX8M and Microchip SAMA5D4.
We have also put work into better supporting Rockchip's display controller within its atomic kernel modesetting (KMS) driver.
You can find more information about these contributions here.
We'll also be focusing on the V4L2 contributions in a separate blog post in the future.
A lot of work has been put into shaping up Panfrost leading to GNOME running on the RK3399. Alyssa recently blogged about it over here.
I'm pleased to announce that I also have Plasma running on Panfrost. With two of the biggest Desktop Environments running on Panfrost, Panfrost is maturing to be a great way to run an open source graphics stack on the RK3399 platform very quickly.
Over the course of the past couple of months, in order to iterate quickly, we've had to come up with a way to quickly deploy images and builds.
Collabora has developed debos, a very versatile tool which allows you to compose custom builds based on Debian in a composable and reproducible way. Debos allows you to take a Debian base, add particular 'overlay' software packages, combine those with a board-support package, and finally generate images which can be flashed directly to a device. Since we wanted more recent builds of u-boot, Linux, and Mesa than what is yet available in Debian, we've chosen to build those ourselves.
In order build all of these, we've used GitLab's excellent CI infrastructure, as used for Mesa upstream, and you can find all the relevant builds over here.
The kernel and Mesa builds can be reused across boards, so in order to enable more RK3399 boards one would only need to build the relevant u-boot binaries and point to the correct Device Tree file.
We've also written a debos recipe for the ROCK Pi that puts all of these builds together on top of a Debian Testing base here.
You can build images locally like so :
git clone https://gitlab.collabora.com/rockpi/rockpi4 cd rockpi4 docker run --rm --interactive --tty --device /dev/kvm --workdir /recipes --mount "type=bind,source=$(pwd),destination=/recipes" --security-opt label=disable godebos/debos --scratchsize=8G rockpi4.yml
Alternatively, pre-generated images can be downloaded from here.
These images come preloaded with Weston (a minimalist Wayland environment) and Panfrost. On boot you should be logged into a Weston session. You can also customise the images to install KDE Plasma 5 or GNOME as you see fit; both of those will work fine as well. The default username and password for these images is "debian".
A shoutout to Tom Cubie for providing the ROCK Pi 4 boards that made a lot of this possible.
The Hantro Video4Linux2 (V4L2) kernel module has gained support for another SoC! The Microchip SAMA5D4 features a single decode unit supporting…
DKMS is a framework that is mostly used to build and install external kernel modules. It can also be used to install a specific patch to…
Building GTK 4 as a Meson subproject for your own application is not only useful for Windows builds, but also for many Linux distributions…
Recently, we have been using Perfetto to successfully profile Apitrace traces in crosvm through VirGL renderer. We have now added perfetto…
As part of a project backed by INVEST-AI, a program managed by IVADO Labs, we have developed a multi-stage neural network-based solution…
Did you know you could run a permissively-licensed MTP implementation with minimal dependencies on an embedded device? Here's a step-by-step…