Ask HN: What is the actual state of Linux phones?
Given everything happening with Android, what is the current state of Linux phones and Linux phone development? I have a lot of C experience, how can I get involved?
Given everything happening with Android, what is the current state of Linux phones and Linux phone development? I have a lot of C experience, how can I get involved?
AOSP is still the better OS. It is fully open source. It has a much better protection model. It is production ready and optimized. It has a working build system. It supports reliable updates.
I find the idea of Linux phones pointless. Instead of trying to create an app ecosystem that will never compete against proprietary-ecosystem optimized and well-isolated one, we can leverage it. I wouldn't want an OS where the banking app can peek into my browser. Desktop OSes are still like that and that's crap OS design in 2025. It has been crap since 90s. Linux being popular has significantly hindered OS innovation in open source world.
If you would like to help, I think helping projects like LineageOS or GrapheneOS is better. You can also try joining reverse engineered driver efforts for open source drivers like Freedreno. You can help porting device-specific kernel drivers to mainline. So we can boot whatever kernels we want on normal Android phones with Mesa OpenGL.
"A lot of C experience" doesn't really tell anything. Have you worked with cross-language systems? How much you know about ABIs? How about interface definition languages (IDLs)? Have you actually written a production driver for Linux systems? Have you implemented any system-level service that got deployed a number of nodes? You need to join somewhere and improve things marginally.
The Linux kernel runs AOSP. Linux has nothing to do with the design decisions that you mentioned.
The pinephone hardware feels like a prototype due to the old A20 cpu (heat problems, poor battery life).
The lesser-known SHIFT6mq has an actual mobile chip and apparently works correctly as a phone under postmarketOS (only GPS, NFC and Camera missing): https://wiki.postmarketos.org/wiki/SHIFT_SHIFT6mq_(shift-axo.... There are pointers on the wiki to hack on the camera if you want to help the cause.
The Fairphone 5 is another good candidate for hacking: calls and camera are not working: https://wiki.postmarketos.org/wiki/Fairphone_5_(fairphone-fp...
Pine64 announced recently that they drop the PinePhone Pro, but hope to continue with the PinePhone for 2 more years. To be honest there is not much of a community left when I last checked, and I dont remember when the PP was last in stock, only the convergence? package.
But there are a great series of tutorials from Lupyuen where he gets Apache Nuttx running on the PP. He fills in a lot of the missing documentation and magic smoke details.
For example: https://lupyuen.org/articles/dsi#initialise-lcd-controller
I'm currently following the same reverse engineering process, but using Zig. The A64 is quite fun to play with, but I think the hardware might be better suited as a feature phone vs full linux. I'll let you know in 10 years time hehe.
To be fair, when I bought my Pinephone, I got the convergence one despite being unsure if I'd be able to use it as a daily driver. The price diff isn't much and well, I would rather have the slightly better specs. And I would assume most made the same choice. So I doubt there's much pressure to stock the lower spec model PP.
You can currently buy a Librem 5, PinePhone, or Furi FLX1. I’m not aware of any others at the moment.
Librem 5 looked promising but last I looked into it, the company seemed kinda sus: https://youtu.be/wKegmu0V75s?si=dyM-FP6JBP_6emm9
Librem-5-sitting-in-my-closet owner here.
I haven't watched that video, but I've seen the comments from unhappy people over the years.
I don't know if we'll ever get the full story, but I imagine it was either that or go insolvent. I'm guessing they tied up all the preorder money, and the delays put them between a rock and a hard place.
They handled it poorly, but I think they're on top of everything now.
Jolla C2. Sailfish OS is still alive (though barely).
1. assuming 7 inch device still phone
2. and phone is phone of it can do phone calls
3. it holds battery for one day at least
there is possibly oss phone https://mastodon.social/@mntmn/111387899138111367
so seems need some video demo to show me before i buy
It's ultimately a pointless endeavour because you don't have access to the source code or even the ability to binary patch and flash your own firmware on the most of the phone components. The modem in particular is a black box. Might as well stick with Android and run a Linux environment inside an app like Termux.
The modem is always going to be a black box but I would caution letting that one thing prevent investing in development.
https://pine64.org/devices/pinephone/
This pinephone project was the last thing I remeber looking into. Not sure where they are at now. But I think they have a number of OSS you can look to contribute to.
I had a pinephone pro that I got to pretty good working state on postmarketos. The phone was still a bit under powered but after lots of tinkering it could do most everything I wanted it to. Battery life was the biggest issue, I could get about 5 hours max out of it. It would also get a bit hot at times which while not a huge problem was annoying. Problem is I dropped it on a run and broke it right before they announced they are discontinuing it. So I'm trying to find a new one, but the options don't look super great. If you are willing to use an old downstream (android) kernel, then there are quite a few options, of which i like the FLX1 best. But that makes running some up to data apps harder. I'm having a hard time finding good options mainline options though. The pinephone is too underpowered for daily use. The latest shiftphone and fairphone seem promising but full mainline support still isn't quite there and they pretty pricey too compared to the ppp.
I think instead of a particular phone or brand what we need are open standards for a mobile platform.
- RISCV processor
- Standard Logic board
- Standard Screen
- Standard Battery
- Standard Wifi & Bluetooth modem
- Probably ditch cellular and use something like LoRa
- Standard OS (Linux)
- Standard package manager
Edit: Imagine a phone that you can just swap and mix components from different manufacturers. You could buy the components yourself and assemble, like a standard PC. If one manufacturer tries to lock you in, or you don't like their component you simply replace it.
So no camera?
I'll be generous and add in "standard camera" for you.
But of course camera tech is moving fast. So either the standard camera gets left behind, or the "standard camera" is continually updated.
Rinse and repeat for sll your "standards". The modem needs updating (new mobile specs coming out from time to time.) Battery tech goes very fast. Every time I eat there's a cpu update. Screens change. Form factors change.
Armies of people at Samsung, Apple et al are managing this all the time. Naturally your Open Source phone would need to keep up.
So instead of listing all the common modules, perhaps it's more useful to explain what sort of budget this effort eould need (sustainably) then list the ways the project will generate income to meet that budget.
Without this your list falls into the "what we need are flying cars" box, which is nice, but not exactly constructive. [Astute readers will also note that we really don't want flying cars...]
Yes, I forgot to add a standard camera on the list.
I think if there's an open standard that different organizations can contribute to things will move faster and will be cheaper than having single organization trying to build a new Linux phone from scratch, it is also more reselient with little centralization. Different groups can specialize on different components where they have expertise.
I chose RISCV because it's a open standard, we could build around that ecosystem.
The standard can be updated, having different generations/versions when it's absolutely necessary. But it shouldn't move too fast especially if the interfaces are clearly defined. Phones from 10 years ago are powerful enough it's just that brands create new devices to force people to upgrade.
How do you move past the initial hurdle of the at least several first iterations of this device being inferior to the competition in every single way, though?
And even if you solve the software issue, a “modular” phone like that will be bulky, heavy and have permanently outdated hw without really offering anything in return for that for >95% of its potential user base.
> to force people to upgrade.
So if they had to pick between a modern device with closed hw/software and something allegedly good enough for them but has an inferior screen, camera, battery, is heavier and has poor quality software. How do you force them to buy it?
> How do you move past the initial hurdle of the at least several first iterations of this device being inferior to the competition in every single way, though?
I think it will initially be a phone for hackers, and they will develop the ecosystem. It will also be competitive in that different manufacturers will try to develop the best components for the market.
While I'd love to see a successful ATX for phones (It's number 1 reason my desktop is x86, 2nd is Steam library).
1) The form factor really fights this.
2) Something needs to change with ARM, cause modular hardware would require device independent ISOs, and that only seams to exist for x86.... IDK about RISCV
> - Probably ditch cellular and use something like LoRa
This is absolutely a non-starter. LoRa is an extremely low bandwidth protocol - in a best-case scenario (e.g. two nearby devices communicating on an uncontended frequency), it's in the realm of 50 kbit/sec. It's designed for low-rate data like remote telemetry, not real-time communications.
I don't agree with any of this outside of having a standard screen and networking modem. The rest of this is why I think Linux phones are being held back.
I want it to work, but I gave up on it as my main device for now. I developed a few apps for Ubuntu Touch, but it's always very far behind Ubuntu. They are about to release 24.04.
I will probably continue developing my apps, but it's not really moving forward.
> I developed a few apps for Ubuntu Touch, but it's always very far behind Ubuntu. They are about to release 24.04.
That's actually interesting and good to know. Last I looked at Ubuntu Touch they were so far behind I assumed they weren't active. I didn't realize they just lag a lot.
I thought Ubuntu Touch forked from Ubuntu and was called UbPorts?
The best hardware is probably something with SD845 as this can run mainline Linux. A second-hand Poco F1 running PostmarketOS is pretty cheap and realistic.
Ehhh, Ubuntu Touch and KDE Plasma Mobile is around, i am just waiting until apps like YouTube, Instagram, etc, jumps to these platforms as well.
It's not great, IMO.
I'm a Librem 5 owner, but I haven't really been able to use it, due to the atrocious (unusable) battery life, and broad incompatibility with most of the things that I need to do with a phone.
I've been a proud Linux user since the 90s, so by rights I should be exactly the market they're trying to capture -- slightly paranoid, FOSS-idealogue, willing to trade off some (or even a lot of) usability for freedom (and these days, a degree of protection from that thing that is happening in the USA and elsewhere, which is no longer safe to call by its name).
Yet here are the things that I cannot do on my Librem 5 that I basically need to be able to do in order to subsist:
- Go for more than 5h without charging
- Google Maps. I know -- on my GrapheneOS handset, I have OSMAnd+ and OrganicMaps (installed via F-droid no less) but as it happens, I live in a city with a lot of bridges and even more traffic, and if I want to get anywhere without a +/- 1h variance due to traffic, I depend on live-traffic-informed maps.
- Signal. (note: It's been a year or two since I checked in -- is this available yet on Librem 5?)
- Parking apps. My city has invested heavily in integration against just one of these apps, and it happens not to run on Librem 5 (or, for that matter, GrapheneOS -- why does a _parking app_ want attestation? smh). This is becoming increasingly inconvenient as competing apps seem to be in retreat and I am getting used to circling, waiting for spaces that I can successfully pay for.
- Ability to run local government app, which is available for both Android and iPhone, but obviously not Librem. This was more urgent during the pandemic, when it was required for various things like crossing borders and proving Covid status, but it's still hard(er) to get service from some agencies without it.
I have my Librem 5 in a drawer with my other fun tinkering toys, like my Raspberry Pi collection and 5" touchscreen modules (and breakout boards etc.) I still have plans to install it in my car at some point (it would be a good fit for an always-charging scenario.) Every once in a while I take it out and admire it and dream about the possibility of a future without domination. Then I put it away again.
I'm still bitter that they never refunded me for my canceled pre-order, despite promising to at the time. It's been years and I never got any money back (or a phone, for that matter). I consider Purism to be an untrustworthy business as a result.
You're not the only one Louis Rossman has a video about this very topic that I saw when I was considering placing an order.
> and these days, a degree of protection from that thing that is happening in the USA and elsewhere, which is no longer safe to call by its name
I actually don't know what you are referencing.
I suspect he is referring to the current political climate in the USA which many call a fascist takeover.
I mean that is what it is. Is there some rule that you can't call it that? Or is it just fascist don't like being called fascist and might go after people who call them fascist? Which is very fascist thing to do by the way.
Sooo Amsterdam? or Pittsburgh?