Hacker News new | past | comments | ask | show | jobs | submit login

Same sentiment, and that's how I found Linux. I've now made it a solemn vow to convert every daily driver I have to dual-boot Win/Linux when it's a >100GB HD.



Ah right, Linux, the platform where suspend and resume doesn't even work and hibernation is a mess.

AMD https://www.phoronix.com/scan.php?page=news_item&px=AMD-S2id...

s2idle is broken and I have an AMD thinkpad with deep sleep and even that doesn't suspend 80% of the time.

Intel (search for 'deep' and 's2idle') https://community.frame.work/t/ubuntu-21-04-on-the-framework...

I already know the replies I might get but just wanted to get this out there. My laptop is now on 24x7 because I don't know if it will suspend or completely freeze.


Linux has reliable sleep on many more laptop models than mac os. But the rule is that you need to buy known good hardware that supports Linux to have a frictionless experience. (Yes this means you often have to buy something that's been on the market for a bit of time already.) Anecdotally I've had less sleep problems than mac-using coworkers, over a long time and many laptops.


> you need to buy known good hardware

What is this "known good hardware"?

I was thinking of somehow getting my hands on the FrameWork laptop since everyone and their mother keep singing praises about it but going through its forums makes it clear that it also doesn't do suspend and resume reliably.

I've already seen plenty of people reporting that their AMD laptops don't do suspend resume and break with every new kernel release. I assumed Intel would be better but reading about the same issue on FrameWork forums isn't encouraging at all.


> What is this "known good hardware"?

Any hardware that ships with Linux, for a start. For another, anything whose quirks are documented on a popular community site (e.g., the Arch Linux Wiki).

It's important to choose a distro release a little bit newer than the hardware.

The Framework is brand new and doesn't ship with Linux, so I wouldn't expect it to be a trouble-free experience.


One way to know is reports on the web like you are already reading. Business ThinkPads with integrated GPUs tend to have many good models, for one concrete direction if you haven't found a starting point yet. Asking your sales rep, certifications, trying a live USB stick of your preferred distro at a friend/coworker/shop/it-dept owned machine can work too.


In my experience, "known good hardware" is ">2-year-old hardware" when it comes to Linux.

But, I'm a bit cheap, and I treat optimizing older tech as a challenge, so I think I've just declared myself both an outlier in general and a stereotypical Linux user.


My AMD Thinkpad was released a little over 2 years ago and it still has serious flaws. And if the AMD gitlab bug tracker and bug reports on forums are anything to go by, I'm not optimistic about those flaws going away in the future.

I have no idea if the situation is just as bad on Intel.


I never understood why I should be hibernating instead of suspending so I can't talk about that - but I haven't seen a laptop unable to suspend for years. That could be luck and not buying laptops with terrible linux support but I wouldn't paint the situation as this tragic.

The experience out of the box is still crap because the ACPI signals from the hardware are often a complete mess - but disabling most of them generally works.


> and not buying laptops with terrible linux support but I wouldn't paint the situation as this tragic.

Please tell me the name of a laptop which has excellent support for Linux. At this point, I'm ready to go out of my way to spend more money than I can afford if that gets me excellent Linux compatibility.

I know, for a fact, that AMD based ThinkPads don't qualify. The Intel based FrameWork laptop has issues with suspend resume as well.


I have a Thinkpad X270, it has an Intel CPU. Never had any problems with suspend/resume on Linux.


I obviously won't buy a 5 year old laptop. I guess I should change my question to "Are there any laptops released within the past year that have excellent Linux compatibility?" Because from what I've read on forums, suspend resume doesn't work for a lot of people and the scenario keeps changing from one Linux version to the other.


I can't confirm my experience is excellent but my partner's 2021 Acer Spin Pro 5 is on par with various Lenovo laptops (I was surprised when it mostly just worked) and better than Asus mobo desktops that I have used.


Dell XPS 9360


Not usable for me because it needs fractional scaling.


This has been my experience as well, with a fairly wide variety of nixes/bsds/etc and a half a dozen different machines.

Now I just disable it completely and tell it to shut down when I close the lid, and wait like 15s. If the fans and lights don't stop "soon", I know it's having problems turning off. It's much better than the random freezes, or boiling alive in a backpack before draining all battery power.


My XPS 13 on Ubuntu while sleping randomly turns on in my backpack and tries to melt a hole/burns the battery to 0. So I think at least SOME of it might be hardware related.


As I understood it, it's an ACPI configuration, more or less 'forced' by Microsoft, to ensure this cool new S0 is the default. See [0] [1] and [2]. The fix/work-around is to tell the kernel to not do that, and just use traditional S3 sleep.

Adding `mem_sleep_default=deep` to your kernel cmdline should fix it. Been doing this on my XPS13 for 3 years now and it's fine.

[0] https://bugzilla.kernel.org/show_bug.cgi?id=199689

[1] https://www.dell.com/community/XPS/XPS-13-9370-battery-drain...

[2] https://www.reddit.com/r/Dell/comments/8b6eci/xp_13_9370_bat...


S3 is no longer available on newer XPS notebooks (e.g. 9310). S0ix works fine with Linux but I've found it typically requires tweaking and testing.

dmesg | grep ACPI | grep supports [ 0.193967] ACPI: (supports S0 S4 S5)

sudo cat /sys/power/mem_sleep [s2idle]

https://01.org/blogs/qwang59/2018/how-achieve-s0ix-states-li...

https://01.org/blogs/rzhang/2015/best-practice-debug-linux-s...


Dell has "fixed" that problem by disabling S3 sleep on certain laptops. Not even kidding, the ACPI tables don't have S3 sleep as an option and it's not even some matter of OSI string trickery either.


The takeaway from this thread is that there is no platform safe from this issue.


Should airplanes forbid PC laptops?

It’s more serious than laptops committing suicide in bags. It’s, anything with a high-energy battery can short itself and cause a fire. Worse, it could be malware or hardware. At this point I am surprised the vulnerability hasn’t been used by anyone.


Airplanes have bags/boxes to throw li-ion batteries into and extinguish any particular fire.

The law is that you can only bring aboard Li-ion batteries of size 100 watt-hrs or smaller on any airplane (https://www.tsa.gov/travel/security-screening/whatcanibring/...).

I think the airline crews are confident they can handle 100 watt-hours worth of burning, but no more than that!


You can't extinguish a lithium ion battery fire by smothering it. The fire creates its own oxygen. You can contain the fire, and you can immerse it in enough water that it cools below the burning temperature and shorts out through the water.


https://youtu.be/oa_yao1DC1U

https://youtu.be/LCFsmHLDuyg

I'm not sure if they are smothering the fire. The strategy seems like it's just a safe place to keep the fire until later.

You won't have buckets and buckets of water on an airplane. You need a solution to safely contain the fire and keep the passengers safe.


The probability of the issue happening seems to be meaningfully far less on certain operating systems than others.


I don't see how you can possibly make any meaningful conclusions about relative probability from this thread.


It is not from this thread, but my experience over the last 15 to 20 years. I have seen lots and lots of people, including myself, close MacBook lids, toss it in a bag, and start walking. But I do not see that with non MacBooks. It was so notable to me that it led me to decide to look into switching to MacBook Airs.


And you believe your experience is statistically significant.


n=1, but we still weight it x20 because we have the most data on it.


I'm really exhausted by this trend of people adding the word "probability" to their argument because they think it makes it stronger. Just no.


*except Macs. I daily drive my desktop with Linux and occasionally game with Windows (it’s set up to be my driver as well with WSL2, but that happens rarely). I’ll only use Mac laptops. I’ve screwed around with others, but for about 18 years they have been the only ones that have been reliable, with good build quality, and no molten backpacks yet.


There are some people reporting problems with Macs on this thread...

What is bad, because if there were one platform that would avoid this problem it would be Macs. But anyway, my phone does that once in a while too... Phones also shouldn't do it.

It's not even a hard problem to solve. There is a single piece of code that wakes a device up, you just have to not call it everywhere. If you don't control all the code, just require some kind of permission, and don't go granting it to the team that writes the system updater.


I've had nothing but problems with Apple's "Power Nap" functionality. I remember three discrete issues with my 2014 MBP (Quartz would randomly crash out of naps when connected to external displays, the topcase frequently felt mysteriously warm and my media controls would freak out forcing me to close spotify/firefox before closing it) Ironically, the only time I've seen it behaving as-intended was when I had my T460p running MacOS with photo analysis disabled. I'm guessing it's an ACPI issue, since Apple's track record with the technology is shaky.


I've had the opposite experience with Macbooks. WiFi randomly dropping out, battery dying overnight while the laptop is closed, external display settings not being persisted, randomly switching from my external microphone to the built-in one halfway through a call... Sometimes reading through these threads I feel like the only person in the world who has somehow had three faulty Macbooks in a row.

I've now had three generations of XPS 13 with Ubuntu. They're not perfect (the battery drains over 3 or 4 days instead of overnight) but overall my experience has been much better.


but very easily fixed as written above :)


I used to disable all ACPI signals (keys and lid) for exactly this reason on my Macbook 2015 running linux.

Not the most comfortable, as you have to manage sleep manually, but definitely the safest.


Same here on an XPS 15, damn thing was so hot I was worried about it catching fire.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: