There's a still a big catch: normally, when a computer "runs Linux", you can keep installing new Linux distros until the hardware dies, which could be decades in the future.
You can install Linux on some Chromebooks, but it's often a lot of work to disable the "press Space to erase your operating system" prompt. It would be better if all EoL Chromebooks received an update to disable secure boot, because secure boot stops being secure when it boots you to an unpatchable OS.
> it's often a lot of work to disable the "press Space to erase your operating system" prompt
That's not what this article is about.
Yes, you can put the Chromebook in dev mode, disable secure boot and deal with that prompt.
But today you can install Debian or Ubuntu or other distros in a container without enabling dev mode, and without disabling secure boot.
Containers were only available on selected Chromebooks until now, and they'll be enabled on all new devices from here on.
You can easily share files between Linux containers and the ChromeOS filesystem, with USB and SD cards, and with Google Drive. There's one-click backup and restore of containers.
Linux apps like Firefox, LibreOffice, Gimp and VS Code are fully integrated into the ChromeOS window manager side by side with ChromeOS apps and Android apps.
I think that's what p1mrx is saying as well. You can install Debian and Ubuntu in the containers, reaping the benefits of secure boot. However, after 6.5 years the devices are EOLd and stop receiving updates making secure boot useless since you're securely booting into out of date software. p1mrx wants secure boot to be disabled once the device is EOL so that they can install a distro without dealing with dev mode.
I agree, the real problem is after (and in my opinion ethically also before) EOL, we can't sign new distro's and bootloaders...
All we need that is currently missing is for end-users to be able (by law) to sign their own boatloaders, by entering their own public key into the most minimal inspectable ring infimum bootloader...
Otherwise it's just a long and slow road towards either fab at home, or alternatively blind computation on the host, and encrypted communication with a minimal IO system attached to the host...
I really agree on the user signed bootloader part. I think we should also have this for Android devices, since most Android devices only have two years of system/security update.
One should always be able to sign one's bootloader. If it's only 6.5 years later it can be that not even Linux support for that hardware would exist then?
On the contrary, I have direct experience with more notebooks that disproves your claim.
Linux support for the hardware of the notebooks was never as good as Windows. Experienced users, to be sure that they won't have problem with hardware, tend to buy either Thinkpad or Dell notebooks. Everything else is a lottery. From three non-Dell, non-Thinkpad notebooks that I've bought since 2000 only one had enough hardware support to be even usable for more than five years.
Moreover, the first one, an AMD-based Sony (also not cheap) worked better under Linux under the versions which were less than 5 years old than the model's introduction. Only since that point the problems with the drawing on the screen became so bad that it was not usable at all for anything but being used headless. I was installing every new Ubuntu version every 6 months there, and the hardware support for the video card came to be progressively worse. I've reported the bug behavior but was never contacted to help the investigation, the bug was closed unsolved eventually.
Recently, only two months ago, my Dell notebook, 2 years old, after an update stopped working with an external monitor under kernel packed in the 18.04 LTS Ubuntu (where everything worked for a year before that). It took only days to fix that, but the bug was wide spread enough to strike most users of external monitors.
So no, Linux won't magically work on a machine which is not continuously directly tested by enough developers. It will also even less work after 6.5 years of machine's introduction, again, unless there was a constant work to keep it working. Linux is not a magic pixie dust, but a product by humans that also needs active contributions to even keep working with the old hardware.
> On the contrary, I have direct experience with more notebooks that disproves your claim.
On the contrary, I also have direct experience with notebooks across a wide variety of makes, models, and eras (though most were from the Windows XP era, since a lot of my Linux installs were from migrating folks off XP).
> Experienced users, to be sure that you won't have problem with hardware, tend to buy either Thinkpad or Dell notebooks.
Well yeah. That tends to hold true regardless of operating system. The laptops that still manage to be terrible on Linux (which in my experience is constantly shrinking, even - and especially - for old ones) - were typically worse under Windows.
> So no, Linux won't magically work on a machine which is not continuously directly tested by enough developers.
Of course not. My point is that once it is working, it's typically unlikely that it'll regress all that much unless there's an explicit push to deprecate old hardware. The old laptops ain't popular now, but that doesn't mean they weren't popular enough 10 years ago (or similar enough internals-wise to popular notebooks from 10 years ago) to end up getting enough attention by Linux driver devs to still be usable to this day. Maybe someday bitrot will set in, but my experience with Linux support on old laptops suggests that to be relatively rare.
> The laptops that still manage to be terrible on Linux (which in my experience is constantly shrinking, even - and especially - for old ones) - were typically worse under Windows.
I gave already a direct counterexample, a Sony notebook. No problems under Windows for 10 years, unusable after 4-5 under newer Linux versions.
The example I haven't given: the Samsung notebook where touchpad never worked properly under Linux. The only response from community was "use an external mouse." Also the WIFi card on that Samsung never worked under Linux. Again the answer as "switch WiFi card or use an external one."
> once it is working, it's typically unlikely that it'll regress all that much unless there's an explicit push to deprecate old hardware.
For that I have 2 examples: Sony notebook graphic drivers regressed to make the notebook unusable. The Dell notebook regeressed with the LTS, it was fixed only because the bug stroke too much notebooks at once.
But hardware support does break in Linux unless enough of effort is made.
And some issues are totally broad. Last year I've bought a box with the APU which can't play videos under Linux without stuttering. No problem under Windows.
Nobody who claims that "Linux always works" seems to have an actual experience with notebooks which aren't intentionally pre-selected as "known to work." I don't know why would such claim be even controversial, it's widely known.
Another hardware device may not work at all; if you do not pay attention to wireless devices. Most laptops comes with on-board 802.11 (a/b/g/N) wireless cards. Not all card supported so make sure you get Intel Pro series card such as 3945 or Atheros based cards. My advice is use Google to search for your driver or use specialized databases (a more or less complete listing of wireless devices with information about the chipset they are based on and whether or not they are supported in Linux) to search for your laptop card."
Who exactly are you arguing against here, who says that Linux always works? Your Samsung example is instructive here. One-off hardware that was never intended to run under Linux, never did and still doesn't. While you have had some frustrating experiences, the broad consensus, and my own experience, say that Linux tends to extend the useful life of older hardware.
For instance, I have a cheap asus notebook from the Windows 8 era. Windows 8 is EOL, so I upgraded it to Windows 10. Windows 10 takes literally half an hour to boot in the best case, which is after spending a whole day in the update-reboot loop, making sure the system is fully updated. (I do this every quarter or so.) So I dual boot Arch. Arch boots in 30 seconds.
And that's not to mention the stack of Thinkpads going back to 2002, all of which still work flawlessly. (But of course they work, because if Linux ever had a target laptop platform, it was 2000s era Thinkpads.)
The whole context of this thread is me responding to the answers to my initial claim (still hard to read, so let me repeat):
> One should always be able to sign one's bootloader. If it's only 6.5 years later it can be that not even Linux support for that hardware would exist then?
And your response finally, the way I see it, confirms exactly that: if you have some notebook for which Linux can't be normally developed and maintained, after 6.5 years the Linux won't "magically work." Which is what I claimed from the start. Full 6.5 years after the introduction, it can remain a non-target. Not as a notebook, but as a set of hardware parts which aren't openly supported. Like all non-Dells and non-Thinkpads for years were, as you also confirm. Being able to install VMs before 6.5 years are over is also not enough.
> today you can install Debian or Ubuntu or other distros in a container
This is far far from the title's promise of being a "Linux laptop"... you might as well also claim all Windows PCs are now Linux PCs because of WSL.
I know I'm not alone when I expect a computer that claims to support Linux to mean running it on the hardware, without a hypervisor, outside of a container or any other kind of virtualization.
Better resource utilisation, control, freedom, not requiring google's permission to do anything. When you are inside a container you are still under the control of another OS... the lack of a bunch of other practical limitations of running inside a container that I'm not familiar with because I don't run my frickin desktop inside a container! why would I do that, it's an unnecessary limitation and increases complexity - maybe some people do if they benefit from containers for a very specific use case, but this does not make it a "Linux Laptop" first and foremost.
A few weeks ago I tried to run Debian and Ubuntu on an older Chromebook with Crouton (also a container solution).
Systemd failed booting, because the last published Chromebook kernel for that device had an incorrectly applied backport of some patch, which made nonexistent system calls return garbage data instead of erroring out.
They're talking about Termina and the Google provided Linux solution.
It's misleading to people who haven't been in the ChromeOs ecosystem before to compare your Crouton setup to Termina. Crouton is not an officially supported Linux on Chromebook solution and requires you to turn on developer mode on the Chromebook which warns you that you are in unsupported land and things may not go as planned.
If you know how Debian works, the new solution Google has for Linux on Chromebooks should work fine for most people. There's only a few gotchas at all anymore. Snaps even work if you know what you're doing.
To be fair, there is a 15-minute mod (mrchromebox.tech) that allows for all Intel Chromebooks =<Kaby Lake to install an open UEFI that supports Windows, Linux and even Clover/MacOS in some cases.
One important difference to note. Google's policy is 6.5 years after "first device on the platform is released". Apple's policy is after last manufactured.
The Apple policy also only explicitly applies to hardware (repair parts), although software often falls not far behind (maybe supported for a year or two longer)
Yeah this. I bought a Nexus 6p on sale, just over a year after it was released. This meant it was less than two years before it was no longer supported. The battery was replaced twice and was junk anyway. Good riddance.
My Nexus 6p worked pretty well except had to do a battery replacement once after which the battery life remained excellent. After nearly 3 years of usage, dropped it and shattered the glass. Tried several places for a replacement screen but the feedback everywhere was "parts no longer available". The phone works perfectly otherwise and I had no intention of getting a new one soon. Finally gave up and bought a Oneplus 6.
This kind of forced phasing out of devices is really unfortunate.
This is eerily similar to my experience. I absolutely love my OnePlus 6. 8 GB of RAM and 256 GB storage is phenomenal, and double the resources of the pixel 3
Random question about OnePlus since I'm seeing it in a Linux conversation... do you happen to use Termux? If so do you happen to know how to run a daemon (e.g. SSH) without getting choked by BgDetect every few minutes?
Yeah I do run termux, but I don't run an SSH daemon anymore (mostly because of issues like that). I just plug in the USB cable and run `adb shell` whenever I need a shell on the phone.
Performance for file transfers is also way faster and more reliable using `adb pull /sdcard/DCIM ./` than with `scp -r phone:/sdcard/DCIM/ ./`
I use termux mostly just run my own ruby scripts or SSH to cloud servers and stuff.
Mind if I ask where the conversation is? (if it's public). That sounds like something I'd enjoy :-)
Your question was not directed towards me but would like to add that for synhing media between phone and computer, Syncthing[0] has been great for me. Once setup, no manual steps are required anymore - everything syncs automatically.
Oh, I was just referring to this post, which is about putting Linux on devices meant for other OSes. :-)
Ah I see. Yeah I use USB when I need to, but I neither always have enough data to warrant the speed (often just a handful of pics...), nor is it true that I only want SSH for file transfer. =P
The only good deal I got on Amazon was for 8GB/128GB storage - still have about 70GB left after having copied all my media over from the Nexus 6p. I think I'll be ready for a replacement by the time I get near filling that 70GB up.
Still rocking my Nexus 5X and will continue to do so until it physically dies. LineageOS works spectacularly and I still get Android updates faster than most flagships.
Another important difference: you can still install linux on a macbook after it is EOL'd by Apple. HArdware support is poor but the machine can still have an OS patched on a regular basis.
And it is worth noting that the 2012 13" MacBook Pro (non-retina, the last upgradeable MacBook) was sold by Apple at least until 2017, so it should be supported until 2024.
Those vintage macbook pros are still getting OS and software updates. Hardly "software-defined garbage". I have a 2012 retina that's still chugging along perfectly. Battery life is down from when I bought it, sure, but there's still enough juice for a few hours of unplugged work.
I use a mid 2009 Core 2 Duo. I replaced the battery on it ~6 months ago. I also installed Mojave on it, and with the exception of a weird quirk where apps that do not live in the Dock do not disappear from it when you close the app, it runs fine.
It is essentially a Hackintosh, albeit a legal one, as the hardware is Apple hardware.
I do dev work on it, though it is ruby and/or javascript, not a lot of true compiling happening. But it works well.
I'd love it to be officially supported, but I mean you have to draw the line somewhere.
Edit: I forgot to mention the SSD I put in it, or the RAM I maxed out. (I did the upgrade work, I was not about to pay the premium for Apple to do it.)
Where did you purchase the battery from, and how is the performance, if you do not mind me asking?
I also use a mid 2009 Core 2 duo, but lately I have struggled to find a source for a battery that will last more that 1.5 hours in typical usage. I assume the issue is that any authentic Apple battery must have been manufactured so long ago that its capacity is severely degraded after having been stored (at non-ideal temperature) for so long, and it seems like even many third party batteries available are old and had inferior capacity to begin with.
> I also installed Mojave on it, and with the exception of a weird quirk where apps that do not live in the Dock do not disappear from it when you close the app, it runs fine.
If you’re talking about what I think you’re talking about, this is a feature and you can turn it off by disabling “Recent Apps” in System Preferences.
What did you need to do to make Mojave compatible? I've seen a few installer patches around[0] but wasn't sure if you rolled your own solution. Thanks.
I have a 2011 non-retina that's chugging along pretty well, the battery life is good enough to stream a 3 1/2 hour NFL game without a charge. Haven't upgraded the os since 10.12, but it still works well enough.
That's impressive, I can't say the same thing about my 2013 retina.
It wasn't until the last ~6 months that I noticed a drop-off in battery performance. Not sure if it was from streaming football/hockey games or my tendency to throw it into sleep mode thinking "I'll have time later tonight" while keeping my server/db/container processes still running.
The Chromebook updates are not the same as the updates you get from Apple. Chrome gets security updates every 2-3 weeks and full OS updates around every 6 weeks. Apple's OS is updated much less frequently.
The newer Macs with T2 chips do prevent you from installing Linux, actually.
There appears to be some debate online as to whether turning off secure boot on T2 Macs allows Linux to work. It should, but there are some reports online that it does not, at least if you want to use the computer's internal storage. https://www.reddit.com/r/linux/comments/a4thsc/the_actual_fu....
I would have expected this to be sorted out by now, but that doesn't appear to be the case. If anyone knows more, please share...
Vintage and obsolete are two different categories. Vintage = 5-7 years after last manufacture vs 7+ for obsolete. Service is generally not available for vintage or obsolete products, however some places like California have service for vintage products available. Further details on Apple’s website: https://support.apple.com/en-us/HT201624
Ha! I remember buying a mac with a disk drive right after the Apple started offering macs with solid state drives. After upgrading the OS, it mac became super sluggish--Apple didn't support the computer for more than a few years.
No it was supported but it was obsolete which is two different things. But also why I don't have mac hardware any longer. One of my favorite machines is still my SamsunNC10, tiny (for it's day) 32 bit machine that runs a *nix. For what I do, which involves writing software, managing servers and browsing the web, it's great, though At some point, it will not have os support any longer for 32 bit.
This is mostly true, but you'll never get as good an experience with Linux on a Mac as on a Lenovo or Dell. There's just too much proprietary stuff. Even little things like the fan sometimes don't work
Apple doesn't always make things easy, but they do have a highly-uniform component set, which can simplify configuration so long as all elements are supported, and senior Linux devs running kit is an incentive to solve hardware issues.
I'm not implying either that all devices are supported (I don't know which are/aren't, though I've had success on Apple kit myself), or that Linus still uses an Apple laptop (he seems to have moved on). But some well-established Linux users have turned to those devices and ben satisfied.
Maybe good for me. The glass is half full but rising. Like many devs, I prefer a *nix-based client, one whose features are already working (not partially left as an exercise for the user), and that has a full range of available software. Macs have been in the sweet spot for many of us, but Apple's design ethos for the Mac has switched from "passion: best computer ever" to "courage: get used to it". I suspect the thing I'll need to get used to is using something other than a Mac.
So, every move made by MS and Google on their popular, mainstream platforms is a step in the right direction. Neither is a full glass yet, but if these moves prove popular, both companies might decide to take even bolder steps toward desktop Linux.
A lot of 7 year old computers are still totally usable, PCs aren't advancing like they used to. My 2012 retina Macbook Pro is a little bit slow in some cases, but it's still a great machine for normal tasks.
Yeah, I actually bought a 2011 MBP like 2 years ago for 200 bucks. Use it as my main driver for my home music studio. Big projects load a bit slow but aside from that everything works fine.
Only thing I don't like is that there isn't really a cheap and decent way to have 2 external monitors on it AFAIK. Which is why I built a low budget Ryzen 2200g build which is perfect for audio and coding and now share my work between the two.
There is no reason you shouldn't be able to use this hardware 10~15 years from now. People use old Apple IIs and C64s from decades ago. The trouble is the locked bootloaders, secure boot and all the binary blob drivers.
I don't know about Apple II's but some C64's died due to bad power supplies. You can still find working ones, but that doesn't mean you can count on any particular device to last. They weren't designed for that any more than today's computers.
Chromebooks are designed to preserve data. The hardware is presumed vulnerable to failure.
Yes, exactly. There are vulnerabilities to be aware of (primarily, losing control of your account, which would be catastrophic), but at least replacing the hardware is just a shopping trip.
This is much better than getting an elderly relative off Windows XP.
I don't see a similar situation happening where I'd have to do that. I can buy a new Chromebook easily, so no need to worry about hardware obsolescence, and ChromeOS auto-updates.
People still run Commodore 64s. There are demo scenes, old hardware enthusiasts that squeeze crazy performance out of old hardware, even a guy who use a Raspberry Pi to emulate an NES cartridge and get his unmodified NES to play SNES games:
Old hardware can often be repaired. If you told a C64 user back in the day his machine would break in two years, he or she wouldn't shell out the $$ on it, but today people buy $400~$800 cellphones that turn into garbage.
Unsurprisingly, there's even a few hacks out there for upgrading old C64s with newer/better components, too, if you want to get another couple decades out of it. There's even new manufacture C64-compatible enthusiast boards being produced [1], although I believe you may have to supply your own SID chip. The latter doesn't count much for this topic, but it illustrates there's some interest. Further, the parts that are likely to break over time are fairly trivial for someone with soldering skills to replace. I know of at least a couple people who preemptively replaced capacitors on old C64s they bought and refurbished.
> If you told a C64 user back in the day his machine would break in two years, he or she wouldn't shell out the $$ on it, but today people buy $400~$800 cellphones that turn into garbage.
I wouldn't put money on that. A C64's starting price was $600 in 1982 dollars. For $154 in 1982, quite a few people might have been willing.
When was the last time a Commodore 64 received a vendor-issued software upgrade? Because that's what we're talking about. Not how long will the hardware last, but how long will the manufacturer release software updates.
"If you told a C64 user back in the day his machine would break in two years..."
That was pretty much expected. Hardware upgrades were just as fast and furious as they are now, if not moreso. Backwards compatibility with years-old systems at the consumer level was a byproduct, not a plan.
I'm just saying support for a few years would typically suggest 2-4 years. 6.5 years is considerably longer than "a few years" would imply. Particularly in the context of the other common computing platform - phones.
Is 6.5 a magic perfect number? No, I'd like to see more like 10 years honestly (which matches what Microsoft supports with Windows). But let's at least use the actual numbers and not easily confused vague terms like "few"
It is still planned obselescence, even at 6.5 years.
I still use a lenovo t420 that is almost 8 years old. I bought it with the fastest processor available back then, and it is still plenty useful as a Debian desktop.
That's a perfectly fine Linux workstation for most non-GPU purposes. I use a slightly older model ThinkPad, in which I upgrade the CPU, RAM, and keyboard. (It does mean running my own GPU server, over the network, for Tensorflow, etc., which is an inconvenience compared to onboard. But "cloud" is en vogue anyway.)
First, Windows 8(.1) is supported until 2023. Second, Windows 8 drivers will usually work smoothly under Windows 10, and I see plenty Windows 8 drivers for the T420 in Driver Matrix. There's even a BIOS update from 2018-06-25.
My T430 (one generation later) received a BIOS update in February and I expect more to come.
If using outdated drivers on unsupported hardware is your idea of full vendor support then enjoy. Your mention of a newer model doesn’t change anything about the op.
My newest computer (the XPS I'm typing this on) is 7 years old. My desktop computer is about 11 years old (upgraded to quad core Phenom, 16GB) and is more than enough for anything I'll need for many years to come (maybe provided an upgrade of the GPU).
I read a some time ago about them working AltOS mode, for Chromebooks, that was going to make installing other operating systems easier. Not sure what happened to it though.
For me I won't mess with a Chromebook until they've consolidated Android and ChromeOS into a single platform (likely Fuchsia). Until then I don't expect any real support for edge cases to be adequately addressed. It will always "kind of but not really work" because Google's already working to replace it with something different.
why not just a UEFI menu where you can implicitly authorize the private key by simply entering a public key of choice? 1024 bit = 128 bytes = 256 hexadecimal digits. Totally feasible to enter manually...
Because that’s still something an MITM in your hardware logistics chain could do to
keylog you, which is what the trusted-boot sequence is designed to prevent.
It might make sense to have a the ability to enter a UEFI key that would disable the dangerous “press to erase” prompt, but I don’t think it would ever make sense to make such a modified machine boot completely silently and “out-of-the-box like” (unless you don’t believe they should have secure boot as a design goal in the first place.)
Probably the most burning in a UEFI key would do is change the “this is a dev-mode laptop” warning to a “this is a customized deployment; if you were expecting a device direct from Google, you are being attacked!” warning.
That prompt has been changed in some newer Chromebooks. Specifically, you have to navigate a menu sequence to re-enable OS verification (which then wipes everything).
Given that most of Androids stop receiving updates around 1-2 years on the market, official statement of support for 6.5 years actually sounds not bad, does it?
I didn't know about the EoL for Chromebooks. I have one on the list that will reach it next year. Yes, I've had it for ~5 years, but I didn't expect it to just stop receiving updates after a certain period of time. Especially because it is still usable. I even got an older model chromebook because it was the best in the market for the price point at the time. Little did I know I was shooting myself in the foot.
This will make me rethink chromebook purchases in the future. I feel sorry for the people who bought into pixel.
Chromebooks reach "end of life" after a few (edit: ≤ 6.5) years, effectively becoming software-defined garbage: https://support.google.com/chrome/a/answer/6220366
You can install Linux on some Chromebooks, but it's often a lot of work to disable the "press Space to erase your operating system" prompt. It would be better if all EoL Chromebooks received an update to disable secure boot, because secure boot stops being secure when it boots you to an unpatchable OS.