OpenBSD on the Lenovo ThinkPad X1 Nano
jcs.org
jcs.org
More consumers need to be made aware of this and everyone should boycott such products. Especially for the price they are paying ($1000+ for this model). If we just give in without protesting about it we are just giving up our rights to repair our devices.
I've recently purchased a used T530, which was the last model to feature a removable battery, removable CPU, unsoldered RAM, etc etc(for the pedants - yes, later models still had some of these features, but none of them had all of them). It's a fantastic machine that runs everyday tasks without a hitch and with having the option to easily repair it I hope it will last at least another 10 years. The only disadvantage I have compared to the newer T490 model that I use is that it's about 1kg heavier.
I think it's reasonable that some consumers would prefer a lighter and thinner machine, even if that includes sacrificing modularity/repairibility, but I doubt those make up the majority and most are just apathetic and willing to go the way of disposable consumerism that corporations are pushing to drive profits further.
The only two components in a laptop that are essentially guaranteed to fail in a short-ish timeframe are flash storage and the batteries. As long as these two are replaceable, I think a device is reasonably "repairable".
While there's a definite chance of the CPU and memory etc. failing, I think for most people the trade-off in size and weight to accommodate sockets aren't worth it from a repairability perspective.
That's where I draw my line, anyway. Flash storage and batteries are essentially guaranteed to fail in 5-10 years of use, and absolutely need to be replaceable. As long as the build quality is decent, the rest can be integrated.
On the whole, I agree with you that at the very least the battery and RAM should be easily replaceable. I just happen to also think that the increase in weight/bulk is worthwhile tradeoff for the option for me to decide when I want to stop using the machine, not the manufacturer.
So naturally OEMs are making laptops as extensible as Atari, Mac, Acorn and Amigas used to be.
Want to extend it? Plug an external device, for everything else buy a newer model.
This seems to be the future as classical desktops fade away replaced by cloud instances and NUCs.
Obviously I can't predict the future, but I feel like AMD will release a new socket within a generation or two, which fits within the "5/10 year" mark.
Also, 16GB sounds awfully low to me. Why has the laptop industry decided nobody needs more than that? It seems to be the limit for laptops with soldered memory, with few laptops coming with more than that. Anyway I voted with my wallet and bought a laptop that has a Ryzen CPU and upgradable memory and SSD, and a very manageable weight.
Even with AM4 based CPU, I'm under the impression that recent CPUs cannot always use old mothercards.
Why do you think one has to be exclusive to the other? Easier upgradability means easier to repair.
And the rare failure of the soldered RAM makes what should have been an affordable repair a very expensive affair as they force you to buy a new motherboard!
Note that another major issue with soldered RAM / SSD / CPU / unremovable batteries is that it makes you, the consumer, more vulnerable to price gauging. For e.g. If you don't want a 4GB or 8GB RAM, you are still forced to buy the 16 GB RAM from the laptop maker and have to pay them what they want, even if it is available for far cheaper from other reputable vendors.
I don't want to repair my device. I don't want to upgrade my RAM, I want websites, OS and apps not to bloat. There is no reason that Facebook and Twitter don't load instantly and are extremely snappy other than that they can.
If you want upgradeable machines, buy a desktop. You gave up most of the upgradeabliity when you brought a laptop.
I don't know what model of ThinkPad you have, but on most of them for at least the last 10 years, if the RAM is the right type it will work.
ThinkPads traditionally have had PSREF and datasheets that specify the max RAM based on the modules available when they were released. But generally they will also work with larger memory modules of the same type. They just don't list them because they weren't available for testing at the time.
Which ThinkPad model do you have? Perhaps it is one that I or someone else may be familiar with.
https://www.crucial.com/compatible-upgrade-for/lenovo/thinkp...
So I don't know, that may well be the limit. OTOH, if there a larger compatible module from another manufacturer, it wouldn't surprise me if it just worked. Lenovo generally doesn't put arbitrary restrictions on RAM size.
Look at "Memory specification" part.
Even if you follow along models with a similar physical footprint as each series has incrementally shrunk over generations (say, compare the 2020 T14 to the t440s).
I've had enough Thinkpads that I lost counter over the years but my next laptop will be of another brand.
Hopefully I won't need a new one until one of the more open alternatives come with a decent 13-14.5" model with 2k display.
To me, using a trackpad or having to carry a mouse around when trackpoints exist would be a massive downgrade in comfort and usability, comparable to e.g. not having a battery.
Thus, as trackpoint uptake from other brands is almost nonexistent, I'll swallow whatever crappy decision Lenovo comes up with in new laptops. They would have to do something exaggeratedly egregious (no battery? no USB ports? That kind of thing) for me to even consider the competition.
It's sad, but as bad as it is that Lenovo makes these changes, to me it's even worse that other brands don't even bother to offer a TrackPoint in at least some of their lineup, considering that it takes negligible space and the inconvenience for people who don't use it is near zero.
Add to that the braindead half-size arrow key design of most models (often paired with a loss of Home/End/PgUp/PgDown keys, thanks a ton Apple) and already there's a comfortable gap between ThinkPads and most other laptops. HP seems to be an exception here, and you'll have to go larger than the 12"/13" range to even find any options there.
I have been keeping an eye out for options of operating the pinter device from a mech without having to ove from the home row.
There are rotary encoders, sensors you can put on the keyboard itself if you make it small and mobile enough, I think I saw something trackpoint-like at some point as well, but nothing fundamental so far.
Even in my many years of having desktop PC's, I think I've only once upgraded RAM separately and never upgraded the CPU without replacing the motherboard.
If you're talking repairability, than I just haven't had to replace my RAM sticks due to wear.
These new products are crap, to be honest.
Sandy Bridge era ThinkPads accepted 16GB. How long ago was that? Is it 10 years already? Granted, it was total, not per slot. Wasn't officially supported, but it worked. Tried personally, memtest passed, all good. IIRC W520 even took 32GB(4x8GB). Ten years of progress and ... still no progress. Who even buys this crap?
>If you're talking repairability, than I just haven't had to replace my RAM sticks due to wear.
Somewhat agree with you on this point. Safe to say that if RAM works now, it will keep working. Test it after buying and you should be good. You can't scavenge soldered RAM when the device fails though.
As someone who splits his work between a 64GB workstation and a 16GB ultraportable, there are a ton of use cases where 16GB is a perfectly acceptable amount of memory. That said, 32GB of soldered memory is feasible in a ultra-portable these days so it would be nice for Lenovo to get with the times.
This applies to normies too. I mean, it's nice to have (web/g/hot)mail, slack, skype, messenger, element, discord open in the background while surfing the web and I doubt that 16GB will be enough for long.
What if I want to run photoshop and illustrator at the same time? Heaven forbid, they rewrite those power tools to run on electron.
Only for doing Docker/Kubernetes at the coffee shop, from my point of view.
Similarly developing mobile apps is more convenient on a laptop as one can interact with an emulator with proper touch input. But then ide and emulators take memory.
On the other hand developing a web-related code on a laptop locally can be even harmful, as opposite to using ssh or remote extension in vscode. Things are too fast locally and one may become aware of performance problems only after fixing those requires a big refactoring.
And just because you have replaceable parts, and the option to buy from elsewhere, doesn't mean the principle of caveat emptor ("let the buyer beware") goes away.
Ultimately it's all about balancing the profits of the company vs the rights of the consumers. While both will have to compromise, removing options from the consumer and allowing the company to dictate terms outright really harms our consumer rights.
Even so, laptop manufacturers have brand reputations to protect and don't want customers blaming them for issues that are outside of their control. It's not just a question of who's legally responsible or whether they're obliged to provide support.
There's also the electrical engineering perspective on this. In a low power device, sending high frequency signals over contact-based electrical connections is something you want to avoid if at all possible.
>While both will have to compromise, removing options from the consumer and allowing the company to dictate terms outright really harms our consumer rights.
For the vast majority of consumers there is no compromise. Soldering the RAM is just better, period. (The vast majority of consumers never upgrade their RAM, and RAM sticks coming loose is a surprisingly common failure mode.)
It does make sense for some forms, but look at all the offerings of Lenovo, especially for home users - even the bigger laptops all feature soldered RAM (or even the weirder combination of soldered RAM + 1 DIMM slot at a more expensive price!).
(Then there's the other factor - unless we force them to explore upgradability in such form factors, they will continue to avoid doing so because the current option is more profitable for them).
From a consumer point of view, soldering RAM / SSD / CPU etc. to prevent easy upgrades and make it very difficult to repair takes away control from consumers on how they want to use their device. It leaves you vulnerable to price gauging and pigeon-holes your device to specific limited use-cases only (for e.g. a new recent trend now with home users is to create a "RAM disk" for heavy torrenting, to reduce the wear and tear on the SSD or HDD - obviously with limited non-upgradeable memory that 4 or 8 GB RAM will no longer be seem sufficient and even though your machine is more than capable, your options are now limited to an expensive upgrade by changing the whole motherboard with a bigger capacity of soldered RAM or buy a newer device with more RAM, hopefully upgradeable this time). Apple is a great example of the price gauging that happens to consumers when you have to depend on them for more RAM or SSD "upgrades option" when you are buying a new system.
Most of the common disparage against my post here is that 4 / 8 GB is more than enough (yeah, just like 640k was during its time /s). Perhaps most have forgotten that one of Microsoft's business model is to sell their OS bundled with the machines (OEM license). Thus, Microsoft obviously has a selfish interest to ensure the OEM's well being (which is obviously in selling more devices for more profit). And so they help them by making their OS more and more resource intensive, thus forcing its users to consider upgrading their hardware (which with soldered parts now means buying a new device). Ofcourse, hardware has outpaced software in the last decade, and with Microsoft also considering a change in its business model (renting softwares), we haven't noticed this trend in the last decade. But that's another threat we consumers have to live with.
I used to have a client who got into a legal tussle, and at one point they called in various key staff to impound their disc drives to archive possible evidence. How do you do that if the SSD is permanently soldered in place, and can't even be readily attached to another machine for cloning or analysis?
I'd also imagine difficulties with end-of-life disposal; you can't just shred the discs and sell the rest for refurbishing or donate them for PR/tax write offs anymore.
So yeah, not being able to repair/upgrade the RAM sucks balls. But if you can live without it, soldered LPDDR4X sucks less battery power and lets especially the integrated GPU do its best work. For the kind of device they're selling, it sort of seems like the right choice.
https://man.openbsd.org/OpenBSD-5.5/bluetooth.4
I'm not sure why the article says it doesn't support bluetooth.
Bump 5.5 up to any higher number and you'll get "not found". Bluetooth was removed around ~2014 in OpenBSD.
https://github.com/openbsd/src/commit/b8042ed98e3e7a691133b4...
I love this word.
I'm apparently behind the times, though: It's on Urban Dictionary and I love their example
https://www.urbandictionary.com/define.php?term=do-ocracy
> “Jeez, why does Mary get to decide what everyone eats and when they work? Who put her in charge?” Older and wiser heads will say, “This is a do-ocracy. If you think you can do Mary’s job, and you want to, then get up there and do it. She’ll probably be relieved. If not, don’t be a jerk and make a big stink about it, or else she’ll stop working so hard and we won’t have anything to eat!”
... which is apparently taken from:
https://communitywiki.org/wiki/DoOcracy
... which is another wiki I'll have to dive into.
As an openbsd user, I will say that bluetooth is pretty far down the list of things I care about.
This is getting really old. Not all users are programmers.
That's because enough people decided to put enough effort to support as many use cases as possible, until the OS got enough momentum to be hard to ignore.
The BSD approach has instead always been "you want it? then code it yourself and open a merge request". That sounds legitimate, but that's the real reason why OpenBSD has never taken off outside of its small niche of geeks - despite being an amazingly designed OS under the hood.
And I know many of those geeks who are quite proud of being part of a tiny niche that scuffs at Bluetooth, USB 3.0, QHD displays or anything that a "normal user" might want - had it been for them a 640x480 screen with a working session of xterm would have largely sufficed. But that's also the reason why there aren't many "normal people" using their OS.
At some point one should also ponder why we support and contribute to open-source projects (especially considering that we mostly do it in our spare time, that, in the case of developers, is often limited and precious). Is it because we want to make the (IT) world a better place with more free and cool products and attract more people, or is it because we like our own well-curated niche and we don't want to let anybody in?
For an example of the last point, transgender care in the UK -- it takes 2 appointments to get treated (hormone therapy), it can take up to 4+ years to get the first appointment (those are the smaller waiting lists), and another year or longer to get the second before you are finally treated for it. This isn't done out of bad-will, it is done out of intent to not mistreat people. However, the effect of this is that it gatekeeps people who can have access to trans care to those who are able to afford it.
Gatekeeping does not require intent, nor does it require malice. It can simply be the result of a cultural artifact creating what is percieved to be a hostile culture.
Both? OpenBSD's niche is being so secure it (almost) hurts. Curating that is worthwhile in a Research OS sense: How many knobs can we tweak on a POSIX system to increase security while explicitly and loudly not caring about much of anything else? Keeping everyone who doesn't share that vision out is part of the plan.
They can be the security pioneers, and the rest of us can see where they get scalped so we don't repeat their errors.
The difference is that OpenBSD BDFLs (Theo et al.) have not been willing to compromise their vision of what the OS should be and how it should be developed just to chase popularity. Look at how they still use CVS, ship their own httpd and ssl libs, dropped sudo years ago (and then rewrote it)... they prioritise consistency and reliability over ubiquity and “the new shiny”. Chances are that, even if you wrote a BT stack yourself and submitted it, it wouldn’t be merged unless it fits their philosophy.
That’s the real difference: Linus and his generals have been willing to accomodate and support a higher number of features just for the sake of it, because it was cool; they were more accepting of incoming developers; and they were much friendlier towards business interests, accepting binary blobs and so on, which is somewhat ironic (Linux is very hard-GPL “inside” but then gave up when it comes to drivers; OpenBSD is, well, BSD everywhere, but they push super hard for manufacturers to open their drivers).
OoenBSD makes IT better too, but it does it on its own terms, and that’s fine.
Then pool resources via a funding platform, get the word out (here on HN, social media, mailing lists) and you may get it done quite soon and also at a low relatively low cost per user.
But like, that's actually _also_ a lot of effort! Especially for projects that are "abandoned" or have no obvious owner, since then you're in "hire a random X programmer to do this for you".
I feel like more OSS projects should go beyond "Donate" and have a "buy a feature/bug fix" button.
You could run into a problem with a lot of OSS no-warranty policies this way.
I am curious what they are?
This is what I had in mind. This obviously impacts some useful gdb features, but such is the nature of OpenBSD.
Perhaps prohibiting access to user-space except GDB would be a reasonable compromise. Also, debug registers are not unique to x86: most (all?) CPU architectures have them. So calling it a "fad [...] Intel cpu feature" is a bit unfair.
There are larger forms of this but Creative's form factor is very small.
This is probably why no-one has written it yet.
Thet also have a BT-W3 that's USB-C now.
So, my guess is nobody who works on OpenBSD a lot cares about bluetooth on OpenBSD, because they're not using bluetooth much, and don't see any use cases for it that would help them.
Well, that's not true. People in the Apple bubble say that Apple devices work with Apple headphones. But that is it.
Saves everyone a lot of frustration.
It's an important commentary on bluetooth that it cannot be included in such a system. In that sense OpenBSD is a valuable filter, even if that makes it impractical to run on a laptop for your use case.
And it’s a laptop, lots of people almost never use bluetooth on a laptop. You might, I do sometimes, but it’s hardly essential.
I'm happy to make this trade-off, given how consistent and reliable OpenBSD is. Not a deal breaker for me.
One reproducible freeze, one sloppy coding (ksh), and some leftover cruft in the kernel.
Previous times I've used OpenBSD I've encountered bugs in what feels like any place I've looked. MPLS, time parsing in libc, …
But it's definitely more "if it exists, it works" than the Lennartware that's been taking over Linux for the last several years.
Some thoughts from 2019: https://blog.habets.se/2019/10/OpenBSD-in-2019.html
Lol at the state of hardware.
For the uninitiated, without S3 closing the laptop lid will either continue to drain battery pretty damned quickly (S1) OR cause the OS to take much longer to resume completely from disk (S4 aka hibernate). As SSDs get bigger/faster/cheaper in comparison to RAM, S4 will start to seem more equivalent, but to make this sensible on an ultrabook you might want to opt for that 1+TB NVME so you can hibernate on lid-shut without eating a significant fraction of your disk on swap.
But that requires pretty heavy power saving code
(It's plenty at that point, but hardly seems to be the yes/no on the whole deal)
With enough whining (ex: the Thinkpad community), the manufacturer can be convinced to replaced their half baked solution by something slightly better.
If you do not care about TPM, you can also patch your ACPI tables yourself.
my work macbook will close all the network connections/compile/rendering when i close the lid to walk from meeting to meeting (or used to. seems like the universe closed this bug as 'WONT FIX')
I have no decent UX to disable that when i need my network connections/compile time/render time/etc... It's either sleep enabled and i can't risk touching the lid, or complicated arcane steps to break sleep and allow me to carry it.
Amphetamine (for your Mac!)
Every single little obvious feature depend on a shaddy shareware app running on your taskbar.
want to close the lid? amphetamine. Want to copy paste history? copyQ. Want per-window alt-tab? Switch. Want to be able to change the volume? soundflower.
i have some twenty icons there now. And that's not even counting the chrome ones to disable features i don't want on the browser. sigh.
First, try disabling TPM (because of a TXT attack from S3), cf https://forums.freebsd.org/threads/acpi-entering-s3-cant-wak... and https://www.darkreading.com/vulnerabilities---threats/new-at...
If S3 is not supported, it is usually due to laziness, to prevent an ACPI GPE storm and causing wakeups in S0ix. People who glue together BIOS code and ACPI tables are not the best and the brightest: instead of using a context dependent EC mask, they simply disable the GPE altogether.
You can add that back with a DSDT patch: all you need is a vector for PRW (power resource wake), making sure the GPE for EC wake interrupt is not masked.
I will probably replace it with another X1 because I am a masochist. But I do wish it were a bit more sturdy. Other than that, I like the size and hardware support.
It seems like the particular model of SSD I got had a lot of issues for many users. The USB C port coming loose is also something I found other people facing.
Maybe X1 Carbon is more shoddy or fragile than other lines.
Like most of my ideas it withered on the vine of bills and rent but every time I look at an Apple Mac there is still a twinge that says "Lenovo makes 100,000 laptops with exactly same components for huge corporate A and B, so there is room to tack on 1,000 more to keep the hardware base stable and release a run of perfectly stable machines, that just work.
Or ... frankly my dell refurbished laptop runs three flavours of linux quite fine thank you. maybe I just cannot go back to BSD ... which is a shame.
I would very much appreciate who might be able to weigh in their experience who has used both a Thinkpad and a Macbook book as primary machines. It's been a while since I last purchased a pc-based laptop and would like to explore options again. Another I was interested in the past include Alienware.
Recently I upgraded to a Ryzen-based Dell but it is hot and loud. I am considering going back to MacBook depending on what they release next.
If I'm not using the mouse I double tap for click+hold.
I remove the red nub too because I don't like how it feels when I'm typing. I just wish a touchpoint-free keyboard was simply an option. Other than that I think ThinkPads are perfect though I lament the lack of ac real Ethernet port on recent models
... I am so over modern Thinkpads.