2019 redux: Raspberry Pi Zero support


#1

I use Manjaro on two everyday machines, and when I did a quick google I found Arch support for the Pi Zero W which I ordered.

Digging deeper, I now see the following with respect to Manjaro:

Timeline

In early 2017, This ZDnet article said:

The Manjaro ARM distribution for the Raspberry Pi 1 and Zero comes in three variations - Minimal Edition, Server Edition and Media Edition.

[Perhaps I also saw this article when doing purchasing research for the Pi Zero W]

In March 2018, Manjaro-ARM was relaunched, but support was dropped for the Raspberry Pi 1.

In this July 2018 post @Strit said:

Manjaro-ARM does not support Raspberry Pi Zero’s (armv6h) at this time.

(The Raspberry Pi wikipedia specifications table shows the Instruction Set as ARMv6Z (32-bit), and SoC as Broadcom BCM2835, the same as the Raspberry Pi 1s.)


Current

In order to hopefully attract maintainers:

  1. What is qualifications are needed to support the Pi 1 and Zero?

  2. Given the past Manjaro BCM2835 support and the current Arch support, what is the guesstimation on how many man hours would it take to support it again? Is there an old backlog of issues/tickets from the last time it was maintained?

  3. I note that the Zero W adds WiFi and Bluetooth support. Given the upstream support here, what Manjaro-ification is needed (and approx mythical man hours)?

  4. How many hours a month would be needed for maintenance?


#2

Are you saying you are willing to help contribute to manjaro-arm so that it supports that board?
Cuz that would be cool.


#3

Well, @Strit is the head of the team, we may know better and more accurate info than me.

  1. Have people that have the hardware and know to build a arch package format and willing to maintain them, also maybe someone to create the images frequently.

  2. I would say not many, just keep everything up to date when there is a free time. I would take more time to make things up running, but once running is just as much as maintaining the packages and deploying it

  3. We try to modify as few as possible from Arch-ARM, so I hope not so much

  4. May vary a lot depending the amount of packages and the compiling times of them and if there is a problem with the packaging.

Manjaro-ARM is a community effort, we’re always opened to receive new maintainers

Edit: Also, Strit made some tools that made a big reduce of actual man power and human brain power required to be successful with maintenance.


#4

For the distribution, it would require a larger server, since the armv6h repo will be around 10 GB on top of what we already have.
This would then translate to more space needed on the mirrors.

As @grayjack said. It won’t take many man hours to keep it running, when it’s first implemented again. The stuff that will take time, is building the packages.

If Upstream has support (which Arch does), then we wouldn’t need to build any of the wifi/bluetooth support packages ourselves, so that would be almost zero man hours.

That depends entirely on how many packages are needed to be built for it. Like kernels, pamac, cotopi etc.
Right now I use about 3-4 hours pr week on package building alone.
It also depends on how well I can get the tools to support armv6h as a platform, since that depends a lot on how QEMU does it. Right now, even armv7h platform is flaky at best. If it’s the same on armv6h, then some packages will need to be built on the devices themselves.

I hope this answers some questions.
Supporting a “new” platform/architecture is often about resources available and if someone wants to maintain it.
I know I don’t have the time to maintain it. :slight_smile: