After discussing in the last part that Digital Signage Player hardware with Linux can be a pragmatic solution. Now the question of hardware selection arises. For PCs with Intel or AMD processors and sufficient budget, this is not a problem. However, many users only want to play HD/4K videos or view images. Small, cost-efficient devices from the Internet of Things (IoT) are sufficient for this purpose. Generally, these devices do not require movable parts such as fans and hardly consume electricity. On the market, there is a virtually unlimited number of inexpensive media players, especially among East Asian providers. Most of them run Android, but some players have the option to install Linux instead. You should keep a few things in mind during the evaluation.
Basically, no matter if you choose Android or Linux, make always sure the manufacturer offers regular updates. This is often not the case, especially with cheap devices. It’s hard to get accurate information. While some vendors specify Linux or Ubuntu support in their specifications, some vendors do not disclose details such as kernel version, video acceleration, or support periods. I. e. You may get devices that run with a kernel version 3.4 (published on May 20, 2012) or Android 4.0 or even older.
Restricted Driver Support
The main reasons for the lack of maintenance are partly because of the manufacturers and the license model of ARM Ltd. ARM Ltd is a British microprocessor supplier who does not produce CPUs but merely designs them. Companies such as Qualcomm, Samsung, Rockchip, Apple, Intel and many others license a certain processor design and adapt it for their product. The licensees then generally have their “CPUs” produced as “Systems On a Chip” (SoC) in a factory (fab).
In contrast to the Linux philosophy, for patent law reasons, the SoC producers offer non-free binary driver versions. Of course, these are not available in the source code. Therefore, they cannot be customized or compiled by others and often work with a certain kernel version. For this reason, such drivers do not enter the so-called upstream or mainline kernel. But this would be important because only from this original Linux kernel every 8-9 weeks on average new improved versions are released. A free upstream-capable driver in the mainline kernel ensures that the hardware will continue to run flawlessly for future updates in the coming years.
At the same time, device specifications are kept under lock and key for patent reasons. This especially applies for graphics and video processors (GPU/VPU), which are so important to us. ARM (Mali) and other GPU manufacturers such as Imagination (PowerVR) fear that open drivers would tell competitors about the internal functioning of their chips. This means that even drivers programmed voluntarily by the open source community can only be developed by reverse engineering.
That means a big effort. Unfortunately, video acceleration is often missing. The company Allwinner was a negative example in the past. Although their processors are widely used, software support was more than questionable both legally and technically.
In addition, many Asian low-cost suppliers simply sold the chips to a PCB and do not offer any further support. For a local media player, this may still be okay; for a digital signage network, this is an absolute No-go.
What’s the problem?
The mainline kernel is rapidly evolving. Buyers of poorly supported hardware will remain seated on obsolete non-customizable software. You will no longer be able to benefit from the improvements and bug fixes in the Linux kernel. If the manufacturer doesn’t care because he doesn’t see it as earning any more money, even serious errors are no longer eliminated. This happens frequently and affects both Android and Linux devices. However, Linux has a large open source community and this can mitigate the consequences over time through reverse engineering.
For example, BlueBorne became known in September 2017. BlueBorne is a security vulnerability in the Bluetooth stack that affects all systems (Windows, Linux, Android…). In other words, there are already millions of vulnerable devices in circulation that will never fixed because their support period has expired. The only effective protection is to turn off functionality that was once paid for.
Those who run an affected device pool and are dependent on Bluetooth now have two options: Either to live with the vulnerability and the risk associated with it or to scrap working tried-and-tested hardware. This reduces both the cost and life cycle assessments. This could be prevented by more openness and the hardware would have a longer service life.
An underestimated and dangerous problem of the lack of sustainability for drivers, especially in the IoT is security. There will be billions of networked devices in the future and thus there is a huge potential for attack vectors. Unfortunately, manufacturers often do not use enough resources to take care for the security issues. This leads to an unnecessarily high risk, e. g. by creating illegal botnets. These networks send spam, attack other computers or “mines” on the electricity bill of you or your customer’s cryptocurrencies like Bitcoins or Ethereum. In order to avoid this, quality assurance will become more and more important in the future. But such a thing is complex and expensive. With an open and free development process like in Linux, quality assurance could be implemented more effectively.
Solutions in Sight
Recently, the situation has improved in general. Also with regard to the drivers for the video acceleration, which is so important for us. Some chip manufacturers are also rethinking things because of the safety factor mentioned above.
Manufacturers with more Support Efforts
For example, Rockchip works with the open source community and releases free and non-free drivers that are easier to integrate into the mainline kernel. Broadcom (Raspberry Pi) even finances open source driver development directly. The same goes for Amlogic, who has been working with the developer company BayLibre for two years to bring their drivers into the mainline kernel.
The laborious reverse engineering, e. g. at Sunxi-Community with regard to Allwinner chips, is now bearing fruit. The functionality of the CPUs and GPUs (Mali) has now been almost completely decoded. A current Crowdfunding campaign to support video acceleration in the mainline kernel for Allwinner CPUs in February 2018 exceeded its financial target after less than 5 days. This means that from June 2018 onwards, all-winner CPUs could receive the long-awaited free video acceleration by default – even without manufacturer support.
Thanks to the Freedreno reverse engineering project, Qualcomm Adreno GPUs will be able to accelerate video playback under Linux by default. Even Google and Qualcomm are now working with the Community to get the drivers of the latest Snapdragon 845 into the Linux mainline kernel.
Effects on the Digital Signage Player Hardware Industry
Our industry will benefit from these developments in the long term. In the future, Linux will make it possible to operate far more cost-effective digital signage player hardware more conveniently over an extended period of time. Many project requirements do not always require the latest octa-core processor. An older tried and tested generations of processors such as an A10 can be perfectly sufficient.
If this is perfectly supported by software in the coming years, we will save acquisition and operating costs. For example by means of well-adapted power-saving functions. In addition, we also do something for ecological sustainability, save resources and help to avoid e-waste.
With our buying decisions, we can also force suppliers to be more sustainable and open-minded. The Raspberry Pi enjoys great popularity despite its comparably weak hardware. A reason for this is that it was almost completely open from the beginning, except for the GPU firmware. However, in my opinion, a Raspberry Pi can only be used as a digital signage device to a limited extent.
Digital Signage Player Hardware Conclusion
Running low-cost digital signage player hardware under Linux is still difficult, but the situation is steadily improving. Unfortunately, it still requires more effort and research than is actually necessary.
However, there are clear rays of hope and the future gives reasons for hope. Based on the above developments, 2018 and the following years could be interesting in terms of digital signage player hardware.
For Linux Digital Signage Player hardware with Rockchip CPU, it is recommended to consider the 4K-capable Rk3288 or the 64 bit RK3399. In future articles, I will explicitly examine player’s hardware from the point of view of Linux and digital signage. If you have any comments or questions, feel free to contact me.