The $199 Pinebook Pro ARM Laptop Is Closer To Running On The Mainline Linux Kernel

[phoronix]

Phoronix: The $199 Pinebook Pro ARM Laptop Is Closer To Running On The Mainline Linux Kernel

For those that have managed to get their hands on the Pinebook Pro as the $199 ARM 64-bit laptop powered by a Rockchip RK3399 SoC and with 4GB of RAM, 1080p panel, 64GB eMMC, and other decent features for the price, mainline Linux kernel support could be in order possibly even for Linux 5.7...

http://www.phoronix.com/scan.php?page=news_item&px=Pinebook-Pro-Kernel-DT

[bisby]

I have 2 pinebook pro's. and they are pretty decent. the only downside is after putting manjaroARM on one, I can no longer watch netflix. because widevine only has an arm7h release (pulled from a chromeOS image) and no aarch64. and manjaroARM only supports aarch64, and widevine doesn't.

Also, annoyingly, the ARM steamlink client doesnt work, because it does a raspberry pi version check. I mostly wanted this laptop to be able to SSH into my desktop from my couch, so while disappointing, hasn't really ruined the device for me.

Everything else functions about as you'd expect for an ARM laptop. Lots of stuff doesn't work because closed source binaries compiled for x86, mediocre performance (although pretty solid for a $200 laptop), and handles any sort of web browser just fine.

I think I would prefer an x86 laptop purely for the purpose of software compatibility, but the Motile M141 covered on phoronix before is $250 now.

tl;dr - pretty decent laptop but definitely know the limitations of ARM before diving in.

[Jannik2099]

bisby for widevine on aarch64 you could use an armv7h container via docker, podman or systemd-nspawn. This is best discussed on the PBP channels though.

Michael the article is kinda misleading: while the dts is the last thing missing that's needed for having the mainline kernel boot properly, there's still the eDP display panel and cw2015 battery controller missing from mainline last time I checked - you wouldn't get the display to light up on mainline, which makes using a laptop kind of difficult. Tobias Schramm is working on mainlining all of that.

[chithanh]

widevine only has an arm7h release (pulled from a chromeOS image) and no aarch64. and manjaroARM only supports aarch64, and widevine doesn't.

I don't quite understand. If you already pull armv7 widevine off the Chrome OS image, then why not just pull the armv7 Chrome browser along with it?

Lots of stuff doesn't work because closed source binaries compiled for x86

I'm curious, what exactly did you encounter that would not work? Is qemu + binfmt_misc an option there?

[RussianNeuroMancer]

I tried to run i386 and amd64 binaries on ROCKPro64 so I can answer:

I'm curious, what exactly did you encounter that would not work? Is qemu + binfmt_misc an option there?

It's deb dependency issue. I can add another architecture to system, but attempt to install any (big enough) i386/amd64 software will fail due to unresolved dependencies (you could try to install spotify, steam or i386 version of wine). I know about chroot way that could resolve this, but:
1. Using it is not much user friendly.
2. It will run whole (chroot-ed) system on qemu, instead of running only necessary bits.

I also didn't managed to deploy WINE Hangover or find aarch64 version of ExaGear Desktop distribution. This is problem actually, as Windows 10 ARM have performant i386 compatibility layer, that not only well integrated to OS, but also call to native system arm64 binaries from x86 executable (in result emulated apps run much faster that it possible with any emulation method available on Linux today). Such level of integration to OS allow to run simple (or just old) 3D games from Steam or heavy apps (many people run 32-bit version of Google Chrome on their Windows ARM laptops, which is kind of stupid, but this is amazing that Chrome actually work just fine on Windows i386 emulator) and this is something that currently impossible with ARM Linux distributions.

[discordian]

For any owner of those laptops: hows the panel (viewing angles, usable in well lit rooms) and keyboard?

It's deb dependency issue. I can add another architecture to system, but attempt to install any (big enough) i386/amd64 software will fail due to unresolved dependencies (you could try to install spotify, steam or i386 version of wine).

you know that apt can pull in those dependencies?

dpkg --add-architecture i386

dpkg -i stupid-drm-client_i386.deb
apt-get install -f

At some point you might run in problems, when the different files are installed from multiple architecture under the same name. It should also be able to transparently call amd64 binaries from arm code (whether that's a good idea when those apps communicate over pipe/local sockets is another topic)
But it should be generally possibly.

[oiaohm]

I know about chroot way that could resolve this, but:
1. Using it is not much user friendly.
2. It will run whole (chroot-ed) system on qemu, instead of running only necessary bits.

There is a problem with the logic here.

I also didn't managed to deploy WINE Hangover or find aarch64 version of ExaGear Desktop distribution.

ExaGear desktop is dead. ExaGear desktop was also using a whole chroot-ed system behind their version of virtual machine. So as qemu performance has improved the differences between qemu with a full chroot and ExaGear desktop have reduced massively.

Hangover is still a major work work in progress.

This is problem actually, as Windows 10 ARM have performant i386 compatibility layer, that not only well integrated to OS, but also call to native system arm64 binaries from x86 executable (in result emulated apps run much faster that it possible with any emulation method available on Linux today).

This turns out to be a lot of work this is what hangover is working on with wine and is requiring modifications to the way qemu works to make using mixed binary types without disaster. So I would not say faster than any methods for Linux today. I would say faster than production ready options on Linux. Hangover is quite fast if you have applications that work with it and are able to install it.

The reality is with Linux due to the fact most programs of a distribution are built native on arm to have a functional system has not as been dependant on emulation as Windows so the effort has not gone into working on the emulation as much as what could be..

[Guest]

The $199 Pinebook Pro ARM Laptop Is Closer To Running On The Mainline Linux Kernel

Is $199 part of the title still relevant? It's out-of-stock, and who knows, if it comes back at $199 price,...

[danmcgrew]

From Michael Larabel's opening article:

"...While the Pinebook Pro is happily running various Linux distributions and BSDs...

The MAIN problem with the Pinebook Pro, and indeed any of Pine Microcomputer's products, is that they are, most definitely, NOT complete and 'Ready-To-Go (RTG) products out-of-the-box (OOTB), no matter what you might read, and from any source.I would definitely like to add the caveat to this statement, "...as of TODAY, 29 February, 2020."
************************************************** **************
I started out as as an almost virulent fan of the PBB when it was announced--very similar to my reaction to the Asus EeePC 701-4G when it was first announced, and I bought one, sight unseen (I NEVER buy a computer 'sight unseen'; and without a lot of solid reviews behind it). I was very actively laying plans as to how I would definitely ensure my entry in the queue to get one of the first of the Pinebook Pros, along with the free upgrade to 128 GB eMMC storage.

tl;dr--

Discretion set in; I decided to wait until MUCH more info was available on absolutely-working software--operating systems--for the unit (there was absolutely none at the time); and it seems as though the correct decision was made. I found it increasingly hard to find any status information--even from Pine64 itself--which did not indicate one serious problem or another which needed to be solved...still.
There is not, to this day, any review or any objective comment of the PBB which indicates that one can purchase the device, under any set of circumstances, and get a very nice (almost) error-free Linux computer--using any Operating System--(such a device does not exist) OOTB, ready to go to work and become your "daily driver".

Pine Microcomputer's problem is that it (they?) solves only half (some would say much less than half) of the problem: they design hardware which has 'dynamite" specs, but then expect the "community" to "...finish the job..." by designing the software required of a high-quality total system.

It simply doesn't work that way. It simply will never work that way. Pine Microcomputer needs to re-think their strategies for bringing a high-quality product to the marketplace. If Pine thinks that they will be successful by catering to the individual whose mentality is that of a reader of "Hackaday", or "instructables", and that "...hacking around with a device until it becomes usable is THE way to go...", we, being desirable of OOTB-usable devices, need to look elsewhere.

It has never been my experience that consumers of any product were content to buy unfinished product.

[oiaohm]

Is $199 part of the title still relevant? It's out-of-stock, and who knows, if it comes back at $199 price,...

Pine as company historically has never increased price of a product. Just with the distruptions in china and thinking their stuff is made in china they have some supply line issues.

Pine Microcomputer's problem is that it (they?) solves only half (some would say much less than half) of the problem: they design hardware which has 'dynamite" specs, but then expect the "community" to "...finish the job..." by designing the software required of a high-quality total system.

This is not understanding some basics. The reality is even major OEM like Acer, Dell, HP.... solve much less than half of the problem. So the work pine is doing is normal level. There is a reason why the result has not been as good.

It simply doesn't work that way. It simply will never work that way. Pine Microcomputer needs to re-think their strategies for bringing a high-quality product to the marketplace.

Its really simple to miss Pine Microcomputer strategy.

Fun enough pinebook pro and rockpro64 are exactly the same soc chip. Why would something like this be important. Some that is simple to over look is the importance of sale volume. Microsoft with their windows mobile hit a brick wall of support due to low sales volume. If you look closely Pine Microcomputer is using 2 vendors of cpu being allwinner and rockchip this is so they can have two vendors fighting over the market share they have..

IIt has never been my experience that consumers of any product were content to buy unfinished product.

There is a particular subset customers that will be happy with a incomplete product normally those wishing to develop/experiment or collectors. There are kind of important customers as they start giving volume of sales.

Once you have volumes of sales you can then start asking your suppliers or parts like soc chips for things like upstream Linux kernel support.

The hard reality is you cannot make a quality product as Windows Mobile found out if you don't have enough volume of sales that the companies providing you with parts will put in the effort to provide you with decent drivers.

If you cannot get decent drivers from the upstream hardware makers you have two choices that are both horrible.
1) Price your product at a non competitive price to cover cost of developers doing the driver development.
2) Price your product at a competitive price and leave it to the end users/community to fix it.

Remember you have to pick the path that you think will give you the most sales volume to improve your treatment by hardware vendors on future products.

Also remember dell.... don't make Windows or Android either they out source this. The Windows the crapware offsetting cost of windows and the discounts they can get for mass purchases windows is effectively free to them. Android other than certification is basically free for Android makers. So there is really no room to add markup to cost to make the final interface complete.

out-of-the-box (OOTB) is a really hard place to get to. Part of getting there is enough sales volume.