End-of-life smartphone? Penguins at PostmarketOS aim to revive it | Hacker News Reader

End-of-life smartphone? Penguins at PostmarketOS aim to revive it

theregister.com

156 pointsrntn65 comments

> They're not all equally supported, though. Most of them can can boot, many have Wi-Fi support, but currently just two actual phones work as phones: the open-source hardware PinePhone and the Purism Librem 5.

And very little will change until something is done at legislation level: companies declaring a device obsolete, withdrawing it from market, or ending its support, should be forced to publish at least the bare minimum information necessary to allow the creation of drivers and functioning software; nobody is asking them to surrender all their IP. PostMarketOS and other projects are a wonderful effort, but they can't keep the pace of the absurd number of perfectly good devices that are thrown away each year because their manufacturers didn't want them to be reused. This practice encourages bad habits from users contributing to pollution, and it's more a political issue than a technical one.

>nobody is asking them to surrender all their IP

I agree with you, but that is how they'll see it. Some of the datasheets I've gotten for an NRND or EOL device are under NDA and still too spartan to write a driver without a lot of trail and error. Because this year's latest-and-greatest product is is 95% the same as the old perfectly-functional products they'll call trash.

For a lot of the more complicated communication chips I've used the data sheets are more aspirational than a reflection of reality anyways. I've often used registers in ways that are documented; it didn't work, and the response from the support was "oh just do it the way our example driver does it."

My understanding is that the support is less about the device OEM and more about the SoC supplier (ie Qualcomm). Once Qualcomm decides they're not supporting the SoC anymore, there's not much the OEM can do in terms of security patching. This is the understanding I've picked up from similar discussions over the years, but I very well may be wrong.

Many SoC's these days are well supported by the mainline kernel (and Qualcomm actually has way better support than, e.g. MediaTek) so some security fixes are still possible even after the hardware is out of support. Flaws in the embedded firmware could of course be problematic.

The non-cellular mediatek chips actually have decent mainline support (not just kernel, tf-a and coreboot as well). At least the ones they put in Chromebooks. The mediatek socs with modems in them seem to be a different story.

Thats what I have heard as well. You would think a giant company like Google or Samsung could go to Qualcomm and require much longer support but it does not seem like that is happening unfortunately

Qualcomm effectively has a near-monopoly on the concept of a cellular radio, especially on the high end, so they really don't have to bend for anyone, even a little.

And yet curiously no one has done an nVidia-like breach on them yet. Coming in the future?

That’s why you make it a legal requirement- if Qualcomm wants customers it has to comply.

So let's place a bet on what support on the pixel 6 generation will look like. Google has their own SOC there, right? No more excuses. I believe they promised a longer support cycle than with previous generations, which is a good start

Hell I just want them to let you unlock the bootloader.

Phone makers don't need to innovate if they can just stop supporting the devices you bought in order to support the "next-gen", even though the next gen is very similar to the last. That way you can pay more for the same thing.

Also, our wallets are the ultimate ballot. Convenience and perceived novelty clearly trump reusability and sustainability.

So long as companies have to show year over year growth in revenue, we will continue to see wasteful corporate practices and terrible designa meant to be thrown away. The incentives are all wrong, and trying to get companies to work more sustainably is akin to asking them to swim against the tide..what are they, salmon??

This is a really cool project!

I decided a few years ago to never buy a new laptop/computer again and reuse as much as possible, repair, and buy refurbished if necessary.

Maintaining a lean Linux distro for older generations of hardware is a great idea.

Great, I am trying to do the same thing because the rhythm of Change is unsustainable in every possible way (ecologically, but also, I want to master my skills on the instruments I use everyday and this is not possible when something change every year).

> They're not all equally supported, though. Most of them can can boot, many have Wi-Fi support, but currently just two actual phones work as phones: the open-source hardware PinePhone and the Purism Librem 5.

This seems like a very significant error. In https://wiki.postmarketos.org/wiki/Devices I see twelve more in the Community section with a Y in the “Calls” column (and they all have SMS and mobile data too), from various different brands—Lenovo, Motorola, Samsung, Wileyfox, Xiaomi. I get the impression from this that many in the Testing section will also support calls.

It’s possible that both are true depending on the definition of “calls”. It may be different with postmarketOS, but at least with LineageOS, there are phones where calls are supported over 3G but not VoLTE. So, while the phone and OS are technically capable of making calls, without VoLTE support they can’t actually do it because mobile providers have decommissioned their 3G networks.

Ah, USA problems. I don’t think anywhere else has decommissioned 3G just yet, though some plan to next year or so; https://en.wikipedia.org/wiki/3G#Decline_and_decommissions suggests the EU may even be planning to keep 2G alive as a fallback, which is an interesting decision.

I have used at least three phones (two Android, one PinePhone on Arch) that allegedly support VoLTE, but I have never had a phone call do anything other than drop to 3G on the Optus network in Australia (through at least three different carriers). I’m not sure quite what’s up with that, but I vaguely recall some kind of insanity about VoLTE not having any baseline codecs and carriers being able to operate a device whitelist and idiotic user-hostile things like that. I’m fuzzy on details, and have never tried asking the telco why it doesn’t work. I really dislike how opaque everything is in mobile telephony.

Keeping 2G+4/5G alive looks like a good option to me - 2G is supported by any GSM phone and has long range, 4/5G - high bandwidth for modern phones. I would not miss 3G at all if 2G and 4G will be available, but 2G decommission will turn a lot of working dump/feature phones into e-waste (I myself use multiple SIM-cards and some of them are in 2G-only phones).

2G isn't as spectrally efficient as 5G. It can't service nearly as many users in a cell so it ends up just being wasteful. Carriers do not want to waste their spectrum licenses on 2G service for some small minority of customers.

Those old feature phones are generating e-waste already as they're on their nth battery and have chargers with connectors not used by any other device.

Also downgrade attacks. 2G is insecure and needs to be deprecated. Newer protocols are the way forward.

Excellent point. Spectrum for 2G is just the start. There's also several layers of backend services needed to keep 2G working and interoperable with modern voice networks. AIUI GSM's encryption is also relatively easy to break today. So 2G service is not only wasteful but a gigantic security hole.

Meh. For low bandwidth internet applications nothing in 2G is really a problem. Everything on the internet uses strong crypto for authentication these days. It would be nice to not have to replace cellular equipment every few years and have some compatibility.

The air interface for 2G is less secure than 5G. This means everything from signaling to actual payload is less secure. In terms of bandwidth, 5G is far more spectrally efficient than 2G. A low bandwidth 5G device can have far better battery life because it's communication duty cycle is much shorter.

A 2G service can't overlap with other services. So if you've got 2G service that's spectrum you can't use for anything else. It's also expensive to run since it essentially doubles the infrastructure needed to provide the service.

Here in Japan, A career only supports VoLTE because it stopped CDMA2000 3G network for obvious reason. An other career don't support 3G WCDMA network on 5G plans because it's inefficient and LTE coverage is good enough. An other career only have 4G/5G network because it started recently. 2G network is completely gone a decade ago because no one use GSM.

We will finally able to re-use old smartphones if open source drivers will be written for their hardware. There is literally no fundamental smartphone functionality that has been lacking from 2012 onwards, and as such there is absolutely no reason why a Galaxy S2 shouldn't at least have an up to date kernel that isn't limited by driver incompatibility

Unfortunately, the S2 has very limited 4G (LTE) support, and the 3G networks are being shut down. I believe, based on the supported bands, that it would only work with AT&T. The 1G of RAM is going to make things difficult moving forwards too.

-------------------

Ultimately, we need something even beyond Google's Fuchsia OS.

First, we need hardware that is relatively standardized, and able to be identified at boot time. Just like how you aren't bothered by having an extra USB controller on your PC motherboard, or which PCIe slot you plug in your network card. We need a standardized means of identifying on-chip and off-chip peripherals available to the system, so that we can have a standard set of drivers loaded automatically.

This is hampered by all the changes (over the last couple decades) in things like power management and security (secure boot).

It is not all bad news. For example, there is relatively low diversity in USB host controllers, everyone tends to use the same standardized hardware interface.

But there hasn't been much cooperation across the industry to standardize peripherals in the embedded world like there has been in the PC world.

> The project is quite different. It uses the mainline Linux kernel, and a standard userland, to support a wide variety of devices. The theory is that not needing a manufacturer's outdated firmware or drivers means that pmOS can use more current components, direct from the various upstream Linux projects.

This clarifies a lot for me. Regarding android app support on postmarketos - I haven't tried postmarketos yet but my understanding is some have been managing this via using 'waydroid'

[Article author here]

Oh, good. Glad to hear that.

I was lucky enough to attend a talk about postmarketOS at the last in-person FOSDEM conference in Brussels, which explained some stuff. :-)

Good to hear. My personal use case that tempts me to learn how to build postmarketOS is to have a 7" tablet (arm, currently android) that can be a fast second screen for my main 13" tablet PC (Intel, Ubuntu gnome)

The key part is fast, I've tried the vnc / rdp based options wired & wirelessly and they've been slow, there's recently been some nice explorations into another approach with a postmarketOS device: https://tuxphones.com/howto-linux-as-second-wireless-display...

I have read about this tech, yes, but I lost interest as soon as I learned I'd need to use GNOME. I can't stand it. I like traditional menu bars, and desktop iconsm, and a vertical taskbar or panel -- and the top panel in GNOME is an egregious waste of desktop space.

JOOI, what Intel tablet do you have that works well with Linux?

Huh, for me GNOMEs been a decent fit as a tablet user thats used to trackpad & screen gestures. Part of desiring this tablet and PC paradigm has meant questioning and letting go of traditional PC design aspects, including ones I've been fond of from macos.

I use an HP Elite X2 G4: https://www.reddit.com/r/ErgoMobileComputers/comments/s6k1qr...

It's surprisingly supported Linux well out of the box, and has some more repairability options (and a business-tier warranty plan), just wish it had a bit more ram than 16gigs when occasionally juggling several projects/VMs. There's starting to be more Linux Intel Tablet options (Surface Pro 8 as ihci touch support becomes baked in, Asus ROG flow Z13, Dell XPS 2-in-1 for 2022, etc.) but this turned out okay as a sudden replacement need after my MacBook finally died a year ago. This is pretty much my one and only computer so reliability is crucial.

One more thing, conceptually the other intriguing thing about postmarketOS is the idea that someone can have something closer to being a 'backup computer' with a used low-cost Tablet rather than trying to own two fully-featured Laptops or Tablet PCs - to me this makes a decent difference when thinking about long-term or remote travel.

Man it'd be so great to be able to run a home server on an old phone, kinda like a raspberry pi but with a lot of accessories included.

It would be very cool indeed. I can imagine myself gathering a bunch of old smartphones people inevitably throw away as it rots in a drawer somewhere for years and years. Then proceed to create a little network of these things and... I dunno, use as CCTVs or something. If we can get up to 50 sub-par smartphones we'll be able to make some sort of resemblance of a botnet!

You can do that easily on Android with Termux. Not sure it would be very powerful though.

I have a few old Moto G1s that I'd love to put this on, but they have carrier-locked bootloaders, so that's out. I have a 2012 Nexus 7 that's pretty well supported, but it won't charge (or recognize the USB connection in any way), so that's out, too.

In some cases, you can ask the carrier to give you the unlock code, if the device is out of contract.

Ok, I'll bite. Here's what I want:

1. make phone calls with bog standard reliability with a Verizon account

2. receive and send texts

3. take pictures with the phone's camera

4. have a bog standard virtual keyboard

5. read Wikipedia

Can PostmarketOS achieve this on some piece of ubiquitous hardware? If so, which one?

My understanding is the current state is "yes on 3G, but no on LTE" which means (depending on where you live) the actual answer may be "no."

Unfortunately Verizon is disabling their 3G network at the end of the year.

Have you considered jmp.chat? You'll never have to worry about VoLTE, WiFi calling, or 3G decomissioning again.

> In case it’s been a while since you checked out JMP, here’s a refresher: JMP lets you send and receive text and picture messages (and calls)

Even for $2.99, I'm suspicious about those parentheses! :)

How is the call quality? What's the latency like?

Call quality is meh and latency is not that bad.

If it isn't "bog standard reliability" then it's simply not usable as a phone replacement.

As an aside, Alpine is a really great distro and an excellent choice to base pmOS on.

If I'd have been in the anti-systemd camp I'd consider it where I currently use other distros.

[Article author here]

It really is, and it's got a lot easier to set up as a desktop OS in recent years. I wrote about it too recently:

https://www.theregister.com/2022/05/26/alpine_linux_316_rele...

This is a really neat idea, but their coverage is not that great. I would totally by a phone that gave over details that made it easy for projects like this to adopt.

I still have a BQ E5 Ubuntu Edition laying around at home. Still waiting for any other distro to support it, despite Ubuntu Touch.

End-of-life? How about just the Samsung Galaxy line beyond one made 12 years ago. It's sort of popular in Android-land.

It seems like a lost cause because of the ever-escalating CPU demands of the present-day web. Keeping the phone running is one thing, making its browser able to keep up with the latest web bloat is another.

> The project's own wiki currently lists over 200 supported devices... Most of them can can boot, many have Wi-Fi support, but currently just two actual phones work as phones

Dealbreaker. I guess that using a phone for its intended use (for making phone calls) makes me a boomer.

I've never met someone who needs at least three types of phones at a time.

This is a project trying to provide all of these things. The limitation is device manufacturers not providing drivers or specifications for people to write drivers against. If there is a driver that supports calls then it will support calls.

mmh.

given then number of phone that reach "end of life" only because the battery dies and is not (easily) replaceable, not sure what good this will do.

I have two BlackBerry phones in a drawer with perfectly good hardware that I reluctantly stopped using because of software issues (Passport and KeyOne). PostmarketOS isn't relevant for these specificaly, but to your point, the hardware would have been fine for probably double their actual lifetime.

To add to this, Google itself just dropped the security updates for Pixel 3a (their own phone not OEM) after just 3 years of launch.

3 years of use from a piece of complex hardware that could keep going easily for 7/8 years.

This kind of narrow planned obsolescence should be an environmental crime.

(Owner of Pixel 3a) Yeah, that's a bummer. There's not something else I'd rather own right now, so I'll likely keep running it for a while.

But most other Android phones don't have a long support lifetime either. Heck, there are a lot of new phones you can buy today that have outdated OS versions, and you're not likely to get much in the way of updates either.

Indeed my example was just to point out the insanity of 3 years support on your _own_ products not some external third party. Truly a waste.

Google's horrible support might turn out to be an opening for Linux. There are a lot of Pixel owners who hate iPhones with a passion that are currently, against all of their instincts and values, considering buying iPhones because they get double or triple the functional life of an Android. They would even switch to something more difficult to use if they both could avoid buying Apple, and either extend the life of their current phone (unlikely at this point, but maybe in the future) or have confidence that the next phone that they buy would last as long as an iPhone.

If they could run Android apps on it with reasonable performance, it would be an easy switch.

> (Owner of Pixel 3a) Yeah, that's a bummer. There's not something else I'd rather own right now, so I'll likely keep running it for a while.

I have the 3a and 4a (non-5G), if you can find a 4a someone is trying to give away on the cheap it's basically a nicer version of the 3a IMHO. It's going to EOL in a year so I wouldn't spend much on it, but if one crosses your path for $100, well...

I have a OnePlus 5 (a device from 2017) that's suffering the same fate. The hardware itself is fine. It has 6 gigabytes of RAM and a SoC that's fast enough that I don't notice any slowdowns in day to day usage. Yet the officially supported software by the manufacturer has been abandoned two years ago. I'm currently running LineageOS, and the phone is working as good as ever. Running a modern version of Android with monthly security patches.

It's a shame that expensive devices nowadays are made to be disposable. After getting a battery replacement, this phone is ready to go for a few more years at least. SoCs have gotten so good that even a half a decade old chip will perform well for the vast majority of users that don't play big games on their phones. It's not like you need a supercomputer to browse social media, websites and use maps for navigation.

I have a Pixel 2XL. It's in perfect condition, works well, and still has decent battery life on the original battery. It has no problems running everything I use, and there would be little benefit I would gain from getting a new phone. Yet Google dropped security updates at the end of 2020, as with the 3a, three years after launch, and around a year after they stopped selling them. My phone now nags me to buy a new one in the settings.

Now, it might be argued that Google has no obligation to produce updates for their products, but... if I try to improve security on the phone by using third-party distributions that are updated, rather than buying a new phone, then thanks to Google's SafetyNet, the phone will be crippled and have difficulty running many banking, payment, and other apps that are at this point a requirement in many situations. Thus, the reason the phone is problematic at this point is because of Google's working to make the phone unusable, not just the lack of updates.

It is bizarre that SafetyNet will vouch for the security of an OS Google itself argues to me is unsafe because of lack of updates, while not allowing a third-party distribution. Yet I suppose I wouldn't be surprised if Google eventually decides to have SafetyNet stop functioning entirely on older devices, in the name of security, to make them even less usable.

Here you go:

https://download.lineageos.org/taimen

That device is officially supported by Lineage and as such gets weekly OTA updates. You'll get a clean Android experience not unlike that of the original Pixel, you can - but don't have to - install Google apps and can even if so required get one of the 'PixelExperience' distributions:

https://download.pixelexperience.org/taimen

Personally I'd stay away from those since they are (by design) heavily tied into the Google world while I do my best to stay free but if you're a heavy Google user this might be the way to go.

One of the issues of old phones is that the battery can't be replaced and the battery cannot run for a long time

Reply on news.ycombinator.com