zlacker

Linux phones are more important now than ever

submitted by wicket+(OP) on 2025-09-16 00:33:12 | 750 points 494 comments
[view article] [source] [go to bottom]

NOTE: showing posts with links only show all posts
◧◩◪
11. JoaoCo+45[view] [source] [discussion] 2025-09-16 01:27:41
>>dmbche+E4
Portuguese government app: https://play.google.com/store/apps/details?id=id.gov.pt

Allows you to have a digital copy of your ID and sign in to government sites/services (there are alternative methods).

22. debo_+Z5[view] [source] 2025-09-16 01:36:01
>>wicket+(OP)
I didn't know syncthing-android had been discontinued. I use that app heavily.

(Linked from the post: https://forum.syncthing.net/t/discontinuing-syncthing-androi...)

◧◩
52. beefle+18[view] [source] [discussion] 2025-09-16 01:54:54
>>loughn+q7
You can install debian on a pinephone https://wiki.debian.org/InstallingDebianOn/PINE64/PinePhone

But for a "normal" linux environment on a phone I recommend postmarketOS. They make an effort to support a variety of user interfaces, init systems, devices.

Still, it is important to consider that the hardware and driver support is the limiting factor here. The camera is very bad on the pinephone because it doesn't have the image processing capability to record video in realtime. It also has no OpenGLES3 or Vulkan. Very poor lima GPU.

74. jaunty+wa[view] [source] 2025-09-16 02:18:20
>>wicket+(OP)
I fully think an amazing consumer-targetting device could take over like a storm if done well, if ambitiously done, with an aggressive software stack.

But. I think what we should ask for now should be simpler. Let this be an alpha geek toy, let folks fiddle with some basic devices boards that can do the thing. The work on PinePhone, Mobian, others is good pioneering work, alas largely held back by there just being so few decent devices for folks to play with. The driver situation keeps making hope here impossible.

It's not a high hope, but Qualcomm has a QCM6490 chip is maybe a rare hope. A chip that is somewhat buyable by regular makers, an extended life version of the Snapdragon 778G. It's pretty modern, and comes with very featureful connectivity hardware. We're seeing variants like non-cellular Radxa Dragon Q6A in the field. Particle has a new Tachyon board you can buy with it. https://www.cnx-software.com/2024/07/31/tachyon-business-car...

It's just stunningly rare alas that folks can make systems with vaguely modern cellular chips. The cores are just not available generally. Sure it's be great to have a well produced Linux phone that is super consumer acceptable with a great OS build out, a new or revived Maemo or a Jolla Sailfish: folks who can go sign the NDAs and make a consumer device but have it be Linux. But I think for this dream to really take hold, humanity needs to be afforded some possibility to have an honest shake, some chance to be a little closer to the machine than typical cellphone bargain. The lack of cellular chip availability has been so so damning to this quest. And here is one counter-example, a crack in the wall, where I see flowers and hope grow.

There was some real nice moments where it seemed like maybe some Snapdragon cellphones in general we're getting Linux support to some level, in mainline, just for the base stuff. No cellular. Unclear to me but it seems like maybe those were just the very barest of beginnings; whether any peripherals at all work or whether there was even a screen is unclear. The trickle of releases also seems to have died off. FWIW though, I will note the previous Fairphone 5 does use the above QCM6490. https://www.phoronix.com/news/Linux-6.1-Arm-Hardware

◧◩
87. defros+tc[view] [source] [discussion] 2025-09-16 02:38:54
>>7e+3c
https://makerplane.org/

(not a plane) https://www.youtube.com/@Ground-Effect/videos

Trains - not so hard, it's getting legit real track time that's the issue - and you can always 'cheat' with a Hi Rail Pickup Truck modification.

Automobiles - .. you are kidding, right? You've never built (or met a builder of) a road certified car, truck, or other vehicle?

◧◩◪◨
111. heavys+Sg[view] [source] [discussion] 2025-09-16 03:21:50
>>salawa+Id
Ask the FCC

https://www.infoq.com/news/2015/07/FCC-Blocks-Open-Source/

◧◩
129. righth+qk[view] [source] [discussion] 2025-09-16 04:04:43
>>loughn+q7
Plasma Mobile[0] on Mobian[1] is getting pretty decent.

[0] https://plasma-mobile.org/

[1] https://mobian-project.org/

◧◩◪◨
152. pabs3+qp[view] [source] [discussion] 2025-09-16 05:10:30
>>salawa+Id
It doesn't have to be, the PinePhone's modem runs a proprietary Linux distro, which you can replace with an open source Linux distro. That is only the ARM processor of it though, the Hexagon one is all proprietary.

https://github.com/the-modem-distro/pinephone_modem_sdk/

◧◩
158. pabs3+Np[view] [source] [discussion] 2025-09-16 05:13:21
>>akagus+l4
Those apps usually require Google's version of Android (or Apple iOS), and block alternative Android builds using attestation:

https://grapheneos.org/articles/attestation-compatibility-gu...

◧◩
159. fabric+dq[view] [source] [discussion] 2025-09-16 05:17:27
>>jaunty+wa
PostmarketOS seems to have decent support for the Fairphone 5: https://wiki.postmarketos.org/wiki/Fairphone_5_(fairphone-fp...
◧◩◪◨⬒
183. pabs3+cx[view] [source] [discussion] 2025-09-16 06:26:32
>>ux2664+Vk
I get the impression they shut it down, but Sony had/have the Xperia Open Devices program. They were close to having their devices running purely on the mainline Linux kernel:

https://developer.sony.com/open-source

◧◩◪◨⬒⬓
232. kuuchu+tH[view] [source] [discussion] 2025-09-16 07:58:30
>>cenamu+tA
According to their FAQ (https://furilabs.com/faq/), yes

  > FuriOS allows for running apps inside a container running Android codenamed Andromeda. This container has complete integration with the host and makes all Android applications work like native applications
◧◩
241. Garden+wJ[view] [source] [discussion] 2025-09-16 08:13:53
>>dvh+uD
Quaero was a disaster - https://en.wikipedia.org/wiki/Quaero

So many committees - so little progress.

◧◩
273. deivid+DQ[view] [source] [discussion] 2025-09-16 09:18:15
>>liendo+ry
Have you tried scrcpy? https://github.com/Genymobile/scrcpy

I ran this with a custom fork to expose the device screen as a VNC server for years, no problems

◧◩
275. Sayrus+NQ[view] [source] [discussion] 2025-09-16 09:19:20
>>ivanje+UP
Manufacturers can prevent you from unlocking bootloader. Whether they are Apple or not. Samsung used to have a hard fuse that broke when you unlocked, it seems they now forbid it entirely.

Apps developers can decide to require Play Integrity so your Android fork cannot be used to run their apps.

Google can decide to not support or explicitly exclude your custom fork. Due to Play Integrity used on their own products, you cannot run Wallet on most forks where Google is not running as root.

Google can decide to delay or not publish source code so your Android fork cannot be maintained anymore.

Manufacturers, Google and developers can alter that deal at any point in time. Recently:

- Delayed patch of AOSP unless your are a partner: >>45158523

- Wall of shame of manufacturers locking bootloader: https://github.com/melontini/bootloader-unlock-wall-of-shame

Those "annoyances" are only one of the attacks made, and not all of them can be easily defended against without having the manpower to actually maintain your own hardware and software stack.

◧◩◪◨
277. jeroen+8R[view] [source] [discussion] 2025-09-16 09:21:11
>>bluefl+rP
UBports (the maintained fork of the dead Ubuntu Touch project) runs fine with systemd/upstart/wayland.

Hell, my watch runs Tizen and that's running a bog standard Wayland + PulseAudio + systemd setup: https://docs.tizen.org/platform/porting/system/#systemd

With the right kernel drivers, configuration, and tweaks, with a well-configured userland on top of that, you can run the "normal" Linux stack in a mobile device.

Getting applications to conform with an API that won't let them drain the battery in the background to make sure notifications don't arrive two seconds too late is much harder. Desktop applications don't really like being suspended/resumed the way mobile applications do.

◧◩◪◨⬒⬓
282. Freak_+OT[view] [source] [discussion] 2025-09-16 09:44:58
>>_heimd+B8
They are required for just about everyone in a lot of countries. In the Netherlands a bank account is actually a right¹, and as far as the tax authority is concerned, required.

1: https://www.belastingdienst.nl/wps/wcm/connect/nl/intermedia...

◧◩
317. jay_ky+Z31[view] [source] [discussion] 2025-09-16 11:22:08
>>cbold+MK
Yeah, I always wanted to pink up one of those tiny GDP machines. They would be cool to play with!

https://gpdstore.net/product-category/gpd-handheld-gaming-pc...

◧◩◪◨⬒⬓
318. type0+t41[view] [source] [discussion] 2025-09-16 11:25:17
>>hilber+BB
Many drivers already install those apps

https://www.usnews.com/insurance/auto/how-do-those-car-insur...

Only a question of time until it becomes mandatory

◧◩
325. jamesn+L51[view] [source] [discussion] 2025-09-16 11:37:04
>>7e+3c
People can and do build their own automobiles all of the time. See: https://www.youtube.com/@GrindHardPlumbingCo
◧◩◪
341. layer8+1d1[view] [source] [discussion] 2025-09-16 12:25:32
>>jeroen+NP
See https://source.android.com/docs/core/connect/uicc. UICC is the software component of a SIM card.
◧◩
349. dvrj10+Wg1[view] [source] [discussion] 2025-09-16 12:51:27
>>karel-+IC
> Android that work without Google Play,

https://github.com/microg/GmsCore, so basically you cannot. It's monopoly in daylight.

◧◩◪◨⬒
357. nani8o+ok1[view] [source] [discussion] 2025-09-16 13:10:11
>>mschus+bQ
There's already a better way to check whether an Android phone is secure enough and it is independent of any proprietary OS certification: basicIntegrity [1].

Most banking apps in Germany use this API and thus work on GrapheneOS and other non-Google controlled ROMs with a locked bootloader.

PlayIntegrity is unnecessary and mostly offers vendor lock in to Google's ecosystem.

[1] https://grapheneos.org/usage#banking-apps

◧◩
363. fsflov+0m1[view] [source] [discussion] 2025-09-16 13:18:53
>>codeth+sZ
> I share the concerns about Android

> Why not try to fork AOSP or GOS

Which concerns do you share? Both AOSP and GOS must follow the Google development strategy, they aren't exactly independent on Google (which is a problem: >>45208925 ). Nevertheless GOS on Librem 5 or Pinephone would be a nice idea, except the GOS developers are against that: >>45101400

> Linux's (nonexistent) security model

Linux's security model is based on trusting the software you're installing from the FLOSS repositories, and it works very well.

◧◩
366. fsflov+Ym1[view] [source] [discussion] 2025-09-16 13:23:23
>>t1234s+vh1
You can run Android apps with Waydroid. If the app doesn't want to run in a non-approved environment, then you can have a problem with Android forks, too.

See also: https://puri.sm/posts/closing-the-app-gap-momentum-and-time/

378. redbel+Eq1[view] [source] 2025-09-16 13:41:34
>>wicket+(OP)
Unrelated question: What is this feddit.org site? It seems like reddit but I have never heard about it! I checked and it appears this is the very first post ever posted here on HN: https://news.ycombinator.com/from?site=feddit.org
◧◩
389. fsflov+Lv1[view] [source] [discussion] 2025-09-16 14:07:49
>>Panzer+HT
> What's the hardest part about an open phone?

>>21656355

◧◩
390. eigens+mw1[view] [source] [discussion] 2025-09-16 14:11:44
>>redbel+Eq1
Lemmy[1] is a federated Reddit-a-like running on the ActivityPub protocol. feddit.org is one particular instance of a Lemmy server that's hosted in Germany.

Personally, I'm pretty enthusiastic about the whole Lemmy project, and I think it's actually a much better fit to the federated ActivityPub model than Mastodon.

[1] https://join-lemmy.org/

◧◩
392. fsflov+5x1[view] [source] [discussion] 2025-09-16 14:14:59
>>cbold+MK
https://puri.sm/products/librem-11
◧◩◪
394. poetas+ey1[view] [source] [discussion] 2025-09-16 14:21:09
>>jones8+Xe1
Ah, I just abandon the idea of recovery on these phones and use the ubports installer once. Thereafter, I just flash directly. I have done the MTK tools dance, but don't any longer. Sorry :(

Ah, I have to admit I don't know much about porting :) We did put up a community wiki and the HADK is up there: https://sailfishos.wiki/books/hadk/page/hadk so that you might get some idea of the process ... I'm not sure to what extent this is different from that 'halium' builds that piggz makes. It makes sense to ask him directly, since he has a stream lined process that targets, among others, the phones you have.

As to why the hardware adaption layer is needed (ie. the android one) it's where all the binary blobs are :) Among other bits.

◧◩
403. fsflov+fI1[view] [source] [discussion] 2025-09-16 15:00:23
>>butz+mE1
What are you talking about? The telephony part has been solved in Librem 5 and Pinephone. Also, this question has already been asked here: >>45257448
◧◩◪◨⬒
407. j45+tX1[view] [source] [discussion] 2025-09-16 16:10:06
>>SahAss+zJ
I went to go check my memory :)

Android is based off, and runs off a linux kernel: https://en.wikipedia.org/wiki/Android_(operating_system). Android being modified in places including userspace is a fair point, but it's linux derived.

MacOS reminded me of it's BSD intermixings. Although the BSDs are separate from Linux, both were Unix alternatives. MacOS was downported to iOS. While iOS is not Linux at it's core, it's definitely based on an alternative to unix, similar to linux.

BSD references: https://developer.apple.com/library/archive/documentation/Ma...

411. sailma+282[view] [source] 2025-09-16 16:58:27
>>wicket+(OP)
Shameless plug for our association.

We focus on helping end users adopting "alternative mobile OS", we compiled a list of those alternatives on our website: https://sailmates.net/actors/

we also provide phones at-cost (aka without making benefits) for users interested in those alternatives but without the technical knowledge to flash one

https://sailmates.net/get_your_phone/

◧◩◪
412. ysnp+fg2[view] [source] [discussion] 2025-09-16 17:41:12
>>fsflov+0m1
>Nevertheless GOS on Librem 5 or Pinephone would be a nice idea, except the GOS developers are against that: >>45101400

GrapheneOS are against using their development resources on a platform that is drastically, significantly worse than the hardware platform they already have, which is quite fair. It is not about Pine64 or Purism's products specifically. They would not be against them if they met 95% of the requirements detailed in https://grapheneos.org/faq#future-devices. It would more sense to explain which of those requirements you think are unreasonable.

◧◩◪◨
413. deivid+lh2[view] [source] [discussion] 2025-09-16 17:45:47
>>liendo+3Z1
I had this old patchset: https://github.com/DavidVentura/scrcpy/tree/dev

Which added vnc support directly onto scrcpy, so I could leave a tablet plugged in to my headless server and access the tablet remotely via VNC

The alternative was to use X11vnc but it was janky and had some issues, plus higher cpu usage

◧◩
415. Katzen+wj2[view] [source] [discussion] 2025-09-16 17:56:44
>>Variou+Xf
Hi. I have a Google Pixel 3a running PostmarketOS https://postmarketos.org/ and it holds up pretty well. My phone lost 20% today with light usage and will maybe lose 60% if I scroll social media a lot.

I was actually surprised it is this good. I reinstalled recently and before the reinstall I had much worse battery life (Maybe 8 hours with normal usage). I think it was because of Syncthing running in the background.

It is also possible to use s2idle suspend which will improve the battery life even more but you will not be able to receive calls during suspend (though that may also be fixed in the future)

424. schnat+OB2[view] [source] 2025-09-16 19:21:23
>>wicket+(OP)
I am surprised that no one criticized kernel architecture as base for a secure mobile OS, yet.

Let me heat up the discussion using this quote

> The Linux kernel has atrocious security. It has an anti-security architecture, implementation and culture. The Linux kernel is not a good base for building any new operating system with a focus on privacy and security

https://grapheneos.social/@GrapheneOS/114665594121762341

◧◩◪◨
447. sterli+hF3[view] [source] [discussion] 2025-09-17 02:08:26
>>Nursie+R6
> People here seem to think this is some sort of Orwellian attempt to control them, but the reasons are more mundane and technical - many of them (mine included, from two countries) use security facilities on the phone to secure your accounts.

My GrapheneOS phone fully supports such facilities. I trust your app works on it?

here's all you need to do, if not: https://grapheneos.org/articles/attestation-compatibility-gu...

◧◩◪◨⬒
467. jovial+dz4[view] [source] [discussion] 2025-09-17 10:59:41
>>fsflov+Ks1
https://en.wikipedia.org/wiki/XZ_Utils_backdoor
◧◩◪
478. codeth+2Q6[view] [source] [discussion] 2025-09-17 23:50:14
>>phendr+cD5
Uhh no? One of the biggest differences probably is that Android applications are heavily sandboxed:

> Applications are security principals. The main difference to traditional operating systems that run apps in the context of the logged-in user account is that Android apps are not considered to be fully authorized agents for user actions. In the traditional model typically implemented by server and desktop OS, there is often no need to even exploit the security boundary, because running malicious code with the full permissions of the main user is sufficient for abuse.

https://dl.acm.org/doi/fullHtml/10.1145/3448609

And then there's also:

- No root access (by default)

- Verified Boot

- Hardware-backed key store and hardware attestation

- User profiles are encrypted independently of each other

- …

◧◩◪◨⬒⬓⬔
491. microt+wuh[view] [source] [discussion] 2025-09-21 16:02:45
>>nicman+iC7
They certainly can. Since Android 8.0, apps can check if a key is stored in a hardware-backed store:

https://developer.android.com/privacy-and-security/security-...

This works by signing an attestation using a hardware-backed key (which is in turn signed by Google). So, there is no way to emulate this in software, because your ROM simply does not have the private key to do so. Part of the attestation is information on whether the booted operating system was signed:

https://source.android.com/docs/security/features/keystore/a...

Again, since this is all hardware-signed, you could only fake this information if you were somehow able to extract the private key from the secure element. The primary weakness is that you could try to patch out the part of the application that asks for this attestation. But they found a solution to that, remote attestation. Instead of the app asking for the attestation, e.g. Google's servers or your bankcan ask for the attestation and for the reasons outlined above, your custom firmware is not able to fake this. If your bank, etc. implemented remote attestation, you can simply do not do banking on your phone anymore.

[go to top]