Blue Screens Of Death Are Coming To LINUX
HTML-код
- Опубликовано: 29 ноя 2023
- The BSOD or Blue Screen of Death from Windows is often memed on but in reality it's actually a really great feature just really misunderstood by so many people
==========Support The Channel==========
► Patreon: brodierobertson.xyz/patreon
► Paypal: brodierobertson.xyz/paypal
► Liberapay: brodierobertson.xyz/liberapay
► Amazon USA: brodierobertson.xyz/amazonusa
==========Resources==========
Dave's Garage Video: • Why are Bluescreens Blue?
Phoronix Article: www.phoronix.com/news/systemd...
XScreensaver BSOD Man: man.archlinux.org/man/bsod.6.en
Devuan Video: dev1galaxy.org/viewtopic.php?...
Systemd BSOD: www.freedesktop.org/software/...
BSOD Merger: github.com/systemd/systemd/pu...
=========Video Platforms==========
🎥 Odysee: brodierobertson.xyz/odysee
🎥 Podcast: techovertea.xyz/youtube
🎮 Gaming: brodierobertson.xyz/gaming
==========Social Media==========
🎤 Discord: brodierobertson.xyz/discord
🎤 Matrix Space: brodierobertson.xyz/matrix
🐦 Twitter: brodierobertson.xyz/twitter
🌐 Mastodon: brodierobertson.xyz/mastodon
🖥️ GitHub: brodierobertson.xyz/github
==========Credits==========
🎨 Channel Art:
Profile Picture:
/ supercozman_draws
#Systemd #Linux #OpenSource #FOSS
🎵 Ending music
Track: Debris & Jonth - Game Time [NCS Release]
Music provided by NoCopyrightSounds.
Watch: • Debris & Jonth - Game ...
Free Download / Stream: ncs.io/GameTime
DISCLOSURE: Wherever possible I use referral links, which means if you click one of the links in this video or description and make a purchase I may receive a small commission or other compensation. 
Наука
![Kehlani - Next 2 U [Official Music Video]](http://i.ytimg.com/vi/MpI7ekFG8A0/mqdefault.jpg)
Linux users overreacting to a minor change that will literally not affect 99.997% of them? Whaaaat, omg, that has never happened before! I am in shock.
Meanwhile windows and mac users after their bloated OS got some extra telemetry in it for free in the recent update:
Yes i use Linux, and while the community is not the best at times, it's still leagues ahead of any private corporation driven only by profit
"We like change, but only if we are the ones doing it." - Benno Rice
Im just curiois what do you mean it wont affect most users? I imagine on most mainsteam systemd distros this service will be enabled by default, i imagine many savvy users may disable it but it will certainly still be something they encounter. Give I do agree this is a massive overreaction nonetheless
I use Linux and I agree with you to a certain extent xd
It's about the PRINCIPLE!!11!11!
Imagine getting an error like this: "Kernel panic: something went wrong :("
I think something like that would be fine on a more mainstream Linux distro like Linux Mint or Ubuntu, but on something like Arch, I would want to see the exact error, considering I would probably file a bug report (unless it was me who made the mistake).
I don't know, I'm on Ubuntu and I'd want more info that "lol, good luck ;P" of an error message. A QR code or a general "something over here borked" would be preferable vs nothing to explain why the system just froze. I have been considering switching to Arch though, I'm just bad at reading so it's taking me awhile to get through the wiki.
@@sevenredundent7256 I think it would be customizable with systemd services, so you could obviously turn it off if you want to.
Linux already has a bluescreen feature: the kernel panic.
Ignoring the visibility on GUI systems, it has *exactly* the same feature.
From what i can tell, up until now, the Linux world has *not* had any common method to inform the user of a critical error.
Nice to see that that changes.
Mind you, it looks to be only related to the boot process?
@@nowave7 That's where most things tend to go wrong, not?
journal, syslog, dmesg are the way
@@nowave7 Even if it is, that's a good start. Plus, boot failures can be a pain to diagnose. And I say that as someone who uses rEFInd. For a regular user they're almost impossible to deal with.
@@darukutsu Do they show messages to average GUI users?
I like the fact that this linux BSOD is literally named "BSOD" internally. I don't know exactly, but to extend of my knowledge, at no occasions were Windows's BSOD officially named "BSOD", despite it being called like this by virtually everyone
Yep. Microsoft always called and still calls them BUG_CHECK to this day
@@Aura_Mancerdoesn't the troubleshooter call it BlueScreen though?
@@bluetoothspeakergamingbut not OfDeath
@@microcolonel It's kinda like suicide doors on cars, everyone knows it's not literal, but having "suicide door" and "of death" in official product materials doesn't inspire confidence
@@Aura_Mancer I thought the official user-facing name was Stop error.
I think having it on desktop versions of linux is fine, but for server installations having a kernel panic that logs to /var/crash is much better. Obviously kernel panics like this are not helpful to users, as they just see the computer rebooting and are wondering why without knowledge of how to examine why. The bsod option would at least give users a place to look, maybe something to google.
Without a doubt
Does one fully prevent the other? The full logs can still be there in /var/log even if a smaller BSOD is displayed. I fully agree about the server part though.
as far as servers usually yea already logged in logs local and usually in external logging automation tools bsod would only waste time.
My understanding is this catches the kernel panic after its already been logged, so logs should exist for review after boot. That's how apple's gray screen of death works.
I don't think that this would be a bad thing, because it (i hope) won't be enabled by default. Also, the text on bluescreen of systemd would be crash log and qr code would contain crash log, which is much better than Microsoft's realization of bluescreens in Windows 10+, which contains almost nothing but link to Microsoft's bluescreen troubleshooting webpage which most times doesn't helps
The last time I checked it does try to describe the error and falls back to the old NT kernel exception code format (or whatever it's really called) if the code describing the error doesn't know what's going on, so it does provide info on what could be wrong.
But what I usually do is to just reboot and check eventvwr. I also do something similar with Linux distros that have error reporting mechanism that collects kernel crash dump (ABRT on Fedora or apport on Ubuntu).
Yeah usually my kernel doesn't panic, it just hangs with no info anyway. Would be nice to see something, we'll see if it actually displays.
One of the things in Windows I hated with a passion: Help files and wizards that assumed you were too stupid to use a mouse, or other very basic shit, but no help whatsoever when something actually went wrong with the thing you want help with in the first place! It was the "Not M$'s fault, you're just a complete idiot" method!
Well, I haven't had to put up with that nor any of M$'s other steaming pile of BS for over a decade already, and got my sanity back!
@@acerIOstream My kernel panicked once, then it tried to run out of the room, missed the door, and smacked into the wall face first! 😂
@@Bob-of-Zoid sounds about right 😂
Honestly I’ve never understood the hate for the BOSD other than theres not enough detail/many reused.
I see this as a great tool for people who don’t know how to deal with the system just locking up lol
True. My only real objection to the BSOD on Windows was when they replaced the semi-informative error codes with nothing more detailed than "Something went wrong :("
If anything, what we're talking about here just saves me time from booting my computer up on live USB when it fails to boot to check the log files for what even happened.
People blame bsod, not the underlying error that caused the crash.
As far as the layman is concerned, bsod is what caused the crash.
People are quick to kill the messenger.
@tanujrana8490
The issue is more that the BSOD doesn't tell you what the problem is and only really leaves you with the option or restarting your PC. It may as well restart automatically and display a message when it finishes rebooting telling you it had to restart because something went wrong.
Until it actually contains useful info(and a generic error code that doesn't tell you anything even after googling it is not useful) and helps users fix the problem it's an annoyance on top of the problem, of course it's going to get hate for that. It's not helping while pretending to help. There's a reason they removed the error code and added a smiley to it, they know the BSOD is a waste of users time, but they're unable to get rid of it without breaking more than it'd fix.
@@JeoshuaCollinsThere’s still something: the STOP codes. The real issue is that Windows reboots automatically, meaning you never have enough time to note down the STOP code.
It's the same BSOD in W11 and you can still enable detailed crash logs and all that stuff, just as before (and sorry but it's better than W10 now, people still blame W11 but it's a never ending cycle : when W10 released everyone said the same about it but after a while it was obvious it was better than W7 and 8)
Why are people too ignorant to understand that an error screen that doesn't look like your 14 and trying to look like a hacker is a good thing and that the color doesn't matter?
EDIT: Yes, I was that 14 year old kid and I felt very cool.
This isn't only about convenience, it's a much needed functionality for debugging. The way it is right now, when the kernel panics and you're in a graphical session, a lot of the time the screen simply freezes as it is and you won't be able to read the messages emitted in the panic. Even worse is that the kernel refuses to flush I/O buffers when it panics, which often results in any error messages that were emitted in the panic to never make it to the log file. So after you reboot, you will never find them in the log, either. They are simply lost forever.
If on the other hand the system immediately displayed a screen that says "The device driver for XYZ stopped working", you would have an immediate pointer of at least who is responsible for the panic, and possibly why it happened, which works regardless of whether it would've made it to disk or not.
As usual, kudos to the systemd developers on their great work, and screw those who've been crying wolf about this great system manager nonstop for over a decade.
finally someone said it
I have little use for this system manager. The problem with it going against the linux philosophy of modularity isn't some abstract semantic philosophical bs. It's a very real-world issue. I'm forced to buy into this huge, sprawling mess (from an individual user perspective, working as a sysadmin managing servers I don't mind systemd) and deal with it trying to replace a bunch of stuff that already works perfectly the way it is, just to take advantage of a few individual features. If I wanted to put up with redundant and monolithically interconnected implementations cluttering my system just to get a few extra functionalities, why not just use Windows at that point? In my free time I wanna be a Linux-User, not a systemd-sysadmin, and unfortunately the mere existence of systemd makes that increasingly difficult. It's just like vegans arguing that their disgusting excuse for food doesn't affect me - yes it does, it's part of our shared culture now and I have to put up with it taking up limited shelf space and restaurant menus. Every distro that relies on systemd means one less distro that doesn't. The reality is that there just aren't that many people willing to use and maintain niche distros. It doesn't matter how good it is, I will never be in favor of something that limits my choices in so many ways, sorry.
This is some great insight. I'm personally experiencing the lack of feedback myself on Manjaro GNOME - GNOME seems to be crashing sometimes and does so with zero explanation - all I get is a freeze (can tell straight away by seconds not advancing on the clock) and then a '@'@'@'@'@'@'@'@'@ on the tty, and GNOME restarts. Now, I'm a long-time Linux user (so it's not that much of a problem for myself) and started using GNOME recently only due to touchscreen support, but given how GNOME is apparently supposed to be this novice-friendly DE (e.g. limited customization options and such to minimize overwhelming the user), it really screws up the experience.
@@cyberdracewhole paragraph, not one example. what parts of systemd are restricting, and what are they replacing that is already stable and working? I've been learning more about linux recently and have no loyalty to systemd and would be interested to hear more from your perspective
thank you systemd for adding useful features such as ```systemctl is-system-running``` and hostnamed
Hopefully it actually displays useful information. I hate the "modern" BSOD that just says "there was a problem", displays a QR code that just points to the windows forums, and maybe has a single non-descript error message.
I could see it being useful if it generates a qr code that has relevant log file location and a timestamp and turns it into a url that could display that information in a different web browser.
While I do find the idea of BSOD on Linux funny, it definitely is useful.
Yes, NVIDIA drivers, I'm looking at you!
Y'know, before we get Good Nvidia Drivers, we'll probably have something far far better.
We just happen to be good at doing shit like that.
The only problem I ever have with the NVIDIA drivers is the module not installing correctly after a kernel update, at which point there is no display for BSOD to output to anyway.
fuck you nvidia
dope shout out to dave. dude has some great wizard content for pcs in general. always a blast to learn something from him
Bingo, if it makes a kernel panic more understandable to the average computer user, then I'm all for it.
QR codes are a blessing for advanced users. No more taking a photo and/or typing error messages into search engines.
this does not handle kernel panics, only systemd error logs on boot failure
if the kernel panics, systemd aint gonna be running to catch it
@Beyley I still kind of wish that was an option. Imagine running Linux and the system locks up due to a kernel panic, but you don't know it at first because there's no console showing it...
I tell people all the time: errors messages are a good thing, we like error messages. They help you find what you need to do next.
The most frustrating part of my job is when Mac or Windows clients just "Error", or "failed to do X" like that's fucking helpful at all.
honestly it is actually something i missed from older windows, it was handy having an error code to look up.
I'd rather have a plaintext explanation going "Hey I'm getting some kind of faulty reading from *insert part/program* could you check it out and everything related to it"
the no. 1 issue with all of this: why should the color be blue? --> this is the color of microsoft and windows. it is also not a warning or error color. blue tends to mean nominal or 'not an error' type of message. and for unix or posix based operating systems it does not even need to be in a color. for example (and for years now) all of the macos and linux critical system errors have been in monochrome colors. has that ever hurt anybody!?!?!! has that ever prevented people from being able to legibly read and understand the error message??? ---> no. in fact monochrome tends to be better than (for example) red. since when you photo or reproduce red color onto another device then the color red can indeed be highly problematic. but so can blue too (but to some lesser degree than the red, orange or yellow).
so this whole blue screen for systemd i totally object to the labelling and color choice here, rather than these added functionality of the feature itself. which is indeed quite a useful additiion. but until then everything referring to or labelled as such blue or bsod should be entirely rejected. (until they fix that aspect, at which point i really dont see much reason to be object at all anymore)
I wish the linux desktop had more things like this, as well as good feedback about other things that go wrong. For exemple, sometimes I try to launch an app in the gnome launcher and it never launches. I can't understand what's wrong untill I launch it trough the terminal. Users needs good GUI warnings, this is a crucial thing for any desktop or mobile OS.
It is not crucial. Why not use Windows if you miss those little annoying pop ups?
@@_sneer_ Surprise! Windows is NOT Linux! Often you can't just grab Windows and replace Linux with it
Mmm... Was running into issues where wine/proton would just fail without any sort of error message, which only seemed to go away after rebooting. Later, figured out I could kill the running wine processes and get it to work that way, but some message that wine was failing because it was already running and not getting along with the crashed running version would have been helpful, to say the least.
Honestly, wouldn't be that hard... Maybe a way to gather specific error messages or, failing that, show the command line output of the currently running desktop session? Especially on things like wine or flatpak programs where the command line can be pretty tedious compared to a double click.
@@akeem2983 Damn! I didn’t know that
@@psiah9889 “it wouldn’t be hard” you are free to add those. Honestly, it wouldn’t be hard. Same as typing ps aux | grep wine for example
Bsods were always kinda spooky ngl. Now systemd will trigger my phobias.
The biggest problem with the windows BSoDs is they last like 5 seconds and then disappear.
That's not a useful amount of time to give critical information. The default behavior should be 30 or 60 seconds displaying the error message with a count down to automatic power off/restart and a Press [key] to skip the timer prompt at the bottom.
Really? The ones I remember stayed until you rebooted yourself... But the last time I saw one on my own computer probably was 20 years ago so it might have changed since then. I don't think my gaming desktops have ever blue screened. My work Mac book crashes much more often, mostly because it overheats because of the terrible design of the 2019 model.😅
@@SteinGauslaaStrindhaug atm suffering intermittent BSOD's randomly on windows, it'll "Collect the Error Log" and then reboot, sometimes it's gone within 5 seconds, other times I do get the time to read it, it differs from time to time, but yeah it doesn't help much, as it's some random errorcode that I'll need to google to understand, and then I'll have 10 different solutions of wildly different things, and then none of them actually help...
@@Voron_Aggrav Plus you'll have forgotten what the error code was by the time you get to look it up.
@@Voron_Aggrav if it's happening fairly regularly at random you probably should take it to a repair shop, maybe they can probe the hardware to see if there's some faults.
@@SteinGauslaaStrindhaug I've got a friend who does such jobs, was already planning on asking him to give it a once over install a new OS, and see what parts can be upgraded, I know RAM is certainly something that needs it
This is pretty logical to display a QR code. How else are you going to extract the log/information when the system fucks up
Honestly, as someone who likes to tinker above my skill level it would be super helpful to get a full error message when something borks bad. I feel like too often I have done a full reinstall because trying to fix it made more of a mess then I had before.
I'm with the posters who commented that they were fine with this functionality, but coloring it blue and _explicitly calling it_ "bsod" was a 🤡 move
The main misunderstanding here: it is NOT for kernel panics. The kernel panic can not be handled by userspace apps like systemd. To work the way it does in Windows, it should be implemented on a kernel level.
Systemd-bsod is simply a graphical splash when some non-critical boot error occurs.
Let's say that the sata cable for a drive in fstab fell out and it's not the boot drive, but the system hangs because the device cannot be mounted - I wonder if this would get caught by this "bsod" implementation
@@TheExileFox probably, as fstab mounts are managed by systemd.
The feature is very good, but it being branded the same way as the windows one annoys me slightly, even though it probably can be customized/hacked to not look like the windows one. But in reality, I do not expect to see it at least as often as on windows so probably this doesn’t really impact me at all
The longer time goes on, the more the Linux world becomes occupied by people who wish they were really using Windows. It's unrecognizable from 25 years ago and a fair bit of it not in a good way.
@@chaos.corner have you ever heard of a regular user switching from Windows to Linux for any reason other than being fed up with Microsoft?
@@arsenii_yavorskyiYes. To use it for free
@@chaos.corner Yeah, it was so much better when it took you a week to figure out how to get audio partly working, had to configure the network manually, have to create your own xorg.conf to even get a graphical display on your monitor and things would just crash all the time. What an awful experience it is having a working system from first boot.
@@arsenii_yavorskyi Me, I wanted to get rid of my screen time from my parents back when I was young. So I installed linux back then.
But now to be honest, this bluescreen is good. Better than no error message. But what I can imagine is that former windows users are so pissed off with windows that they want a different design for the bluescreen.
I hope the error messages are useful. BSOD was good way back, but lately they just haven't kept up with the new kinds of errors and the information aren't as useful or actionable at least on the face. But the QR code is a good idea though and it's good that sydtemd has it as well
I want my error screens to have anime girls crying and profusely apologizing for the error.
I mean, my issue with bsod is that every time I get one it's impossible to even find out what causes the issue, with linux I can normally figure out where the issue comes from
My first kernel panic happened after I physically removed a drive from my system without removing it from fstab. I asked my friend (who is a long time linux user) for help and I kid you not, his response was "time to reinstall." Needless to say, I was not happy with that answer, so I spent the next 30 minutes reading back through the systemd log file until I found the error pertaining to fstab. Having a dedicated screen that would have straight told me the critical error was from a missing drive from the fstab file would have been very useful, and I openly welcome the addition.
‘dmesg | grep error’ would show you all errors you had back then in like 5 seconds. No need for systemd or idiotic bsod
doesn't systemd display an fstab error in the tty already?
unless you have quiet boot enabled of course
@@_sneer_That's not a friendly way to display errors, not only high-tech elits have to be the only people to use LINUX.
@@Eren_Yeager_is_the_GOAT From what I remember, the error message was more or less "Kernel panic, the log file is at /path/". Maybe it has gotten better since then, but from what I recall there was no user-friendly message.
@@_sneer_ Thanks for the opinion. You can keep enjoying your custom gentoo stack, and I'll mind my own business with my ease of use features without getting in your way.
Glad to see Blue screens going back to their roots of letting you know that there was a critical problem and what it was related to, after time as a Windows admin where win 10 and win11 you've got a 80% chance of the BSOD having no workable information ("Oops, something went wrong :
It's the same everywhere - the bearer of bad news is held responsible :D
I hope these QR codes won't end up as just urls to websites. This will be hell to debug, when wiki or other website will be down or address just not exist because QR will have outdated data.
Personally I prefer to have visible logs of what have just happen. For me its much easier to find cause of the issue thanks to them.
I hope it contains the error message or even (part of) the trace and puts that into some kind of search engine :)
@@jfolz Exactly. :) Part of trace is must have. Sometimes root cause could be deeper, than the last error message.
It could have part trace and a QR code at the same time though, so in case the wiki is unreachable, you can still get some useful information from this
Yes, but only so much info can be in a QR code, they are only a tiny amount of data.
@@Christobanistan the largest version can fit a bit more than 4k characters, less if you increase the error correction level. That should be more than enough to include a few lines :)
The problem is not that the blue screen is at fault for crashing your system, you are right that it does actually not.
The problem is how utterly and seemingly purposefully unhelpful it usually is.
When getting a bsod you are usually presented with a sad smiley accompanied by a message like this: "Your computer has encountered an error, the error code is E1234567890ABC", and upon googling what exactly the error is that would lead to error code E1234567890ABC, the code is the only hint it gives you, you find out that your hard drive has failed OR maybe the service for displaying ads in your start menu has had no internet for 5 or more minutes OR one of your loaded drives crashed, that was a thing with virtual box drivers while I was in training you could not unplug a USB mouse without a bsod, OR maybe your CPU overheated OR the upper screw of your GPU needs tightening OR windows disagrees with the color of underwear you are currently wearing.
And so even reading a bsod did not really give you help since the error codes where each covering so incredibly many completely unconnected problems.
Also, there was a time that the only thing to use Google on was the box with the Blue Screen. So you are just stuck with an error message and nothing else.
Nothing like googling an error code and coming up with no results to boot.
That's just because microsoft's documentation is terrible, probably to drive more businesses to their enterprise support contracts.
Also the "useful" qr code is a mere link to the general help site. I would have expected a real event log.
way better than just letting things malfunction mysteriously.
The worst is it will sometimes reboot before you manage to write the error code down or pull up the qr code scanner.
I don't know if it's because I'm a dev, but it's honestly incredible to me that apparently (according to you) people out there blame the BSOD for causing problems. If that's the case, I don't even think they understand what it is?
Like when a buggy program crashes and you get an error popup, you don't blame the error popup. I feel like people are smart enough to know that. I'm not sure if you're correct on how people see BSOD. I'd be curious to know if you have examples, though.
I think it was the former president of the United States that said something about "if we stop testing there won't be new cases" or something to that tune.
People still blame doctors for increasingly accurate medical diagnosis. Something something eight trillion years ago we have no diabetes and autism. Go figure.
Well, he did show some comments by some very ignorant people at the beginning who apparently that that it meant bringing intentional instability to Linux systems! :D
Lol, it's like blaming imessage because your girlfriend texted you saying "it's over".
Yes, let me rice my blue screens of death please
When I had to write a kernel driver in Windows for a project, it instantly demystified the whole thing to me. It's just a crash reporter for kernel-mode crashes. Just like an unhandled error in a user-mode thread takes out the whole process, an unhandled error in a kernel-mode thread displays an error screen and then takes down the kernel. Nothing much to it.
Yes, but this is purely for the boot process. Linux has a separate kernel panic, though which doesn't do a useful BSOD.
The "your overclocking fried everything" screen. The "random Chinese peripheral discovered" screen. The "your kernel mod sucks" screen. Probably handy.
Some people have been complaining that this is aping Windows, but honestly, there's no need to get creative here. The BSOD is a well established convention, and people know what a blue screen means. Don't reinvent the wheel just to be unique.
Unix-likes have already had "BSODs" since the 1970s. The BSOD is a Windows version of a *kernel panic*. And they came up with this because they had tried to build a system that healed itself and could recover from major errors (Multics) and it didn't work. Just crashing the system and forcing a reboot was better than trying to continue on with memory corruption or driver errors.
This could really help adoption, IMO. A HUUUGE point of failure in installing Linux on a system for newbs is boot failure. No newb is going to look up logs, so they just give up, even fairly expert power users! A useful BSOD could help those people get past issues on their new Linux system.
Sounds like a great feature, I actually resolved issues thanks to BSOD information when I still used Windows.
7:35, This way of thinking probably came from the days when the DOS-based version of Windows (3.x, 95 etc) could BSOD over the slightest little hiccup. Whereas in NT-based systems, if you get a BSOD then something has gone badly wrong and needs attention. But I think a lot of people don't realise this and just put it down to Windows being Windows.
Yeah, GPU drivers were a big issue until I think Windows 8, when they got better, and many kernel level GPU driver issues that would have resulted in a BSOD now recover properly. Basically a BSOD nowadays _generally_ means something in the kernel or running in kernel mode is so badly screwed up that the safest option by far is to halt operation and display the BSOD.
Blue screens are nice. Especially when they're more informative and user-friendly than a raw kernel panic.
Exactly, some people just seem to want to complain about anything, even when its actually useful.
I've actually slowly found systemd more and more useful, it just could be a bit more friendly navigating the journal.
BSoD may stand for Black Screen of Death as well. Make it black as the rest of the linux and it's okay
I think the best we can ask for far as neutral theming is to have the QR code generated in a corner of the screen using a special QR font with box symbols that are written in a corner of the display, and overlays that on a semi-opaque black background. Kinda how Apple shows KPs.
Maybe you get to provide a CSS.
I'm all for giving important loggers gui's and this sound's even better since it has the fixing guides
Yeah BSODs are very useful once you understand the messages its giving out. Like in the Overclocking community any time your PC is overclocked and it is unstable it will throw a BSOD once you put it under load. If memory speed and timings are too unstable that it will throw a BSOD.
I'm all for this because to be honest, boot errors are a pain, because for some reason distros have decided to ship terribly small console fonts that are hard to read, and good luck finding the log line that expalins the boot failure along all the text.
I'm glad this is getting implemented, would've saved headaches to my early linux years, and will probably do in the future
They are shipping the same fonts as always. It's simply your pixels that got smaller
If you use grub, you can change the resolution of the bootloader in /etc/default/grub. You have to run 'grub-mkconfig -o /boot/grub/grub.cfg' after any changes in order for them to take effect.
Oh, god yes. But I fully expect SystemD to use incredibly small fonts on their BSOD. I hope they do 18 point.
And it's a problem why? Like, any info relating to something acting up is undeniably a good thing. People just got the wrong idea of a BSOD. It's meant to help you find the issue, not be the issue itself
The real problem: Linux logs are not user friendly, as they are kinda hard to reach if you don't know what you are searching for. Then, when you find them, selecting what is the issue and what is the "cascade of failures" it caused, as well as bogus errors that always happen harmlessly... Then, when you throw it at google, there are post from Stackoverflow, some random forums, a bulletin message from 2004, a bunch of ad filled sites that either mirror SO or another forum... Like, what is "Error 308"? Why we don't have a repo for the mapped errors like we do with documentation? But a BSOD on SystemD feels like a hack to something that could be done better, like some sort of standard error, that the bootloader, the kernel and the init daemon, so if something critically crash, we can see the relevant error message, a link/qr-code/whatever to some knowledge base (from the distro, or the package, for example), and recognize that the system is irrecoverably crashed and have time to take note of the possible cause... 😅🤷♀😊
No, the real problem is you can't get to those logs if you can't boot the system. Mayyybe you can boot to another OS or use another machine. Or maybe the log wasn't actually written ANYWHERE because of disk failure (or a non-disk failure before the disk drivers/system are loaded).
the most obvious solution is to make it black by default and everyone will be happy
Windows did that for a little while actually tho seems they went back to blue I recall after windows 11 first came out i got a black screen of death
This would have been so usefull!!!!
I had a faulty image loaded onto an USB (the whole filesystem was read only and the /var was missing - mounting issue)
The errormessage was shown only for 0.5 seconds and then rebooted the system. I had to make a video of the boot process to even read the message log on screen. With an bsod this would have been much much easier.
as i understand it the windows BSOD can occur at any time (and gives error codes that are vague at best), systemd-bsod however is solely for boot-time errors (and linux users being linux users will likely have highly specific errors that are easy to look up, will also likely stay up until dismissed unlike windows bsod which you might get half a second to pull out your phone and scan the QR)
Yeah, this is only boot stuff. Linux does have a kernel panic, though, which needs a similar BSOD.
The issue with BSOD is it gives fuck-all information about what happened.
You get _one_ word that _MAY_ give you a _HINT_ of what went wrong, but if you didn't expect a crash and if the PC is fast enough, you're not gonna have time to read that message before it's automatically rebooting.
What do you mean "hint"? People usually search blogs and stuff, instead of going to the microsoft documentation (the developers one, not the user one) and it's clearly explained.
Sure, hard to find, but they are error codes just like systemd's ones. And honestly systemd's docs aren't exactly on plain sight either.
In my opinion, this is a great addition. I just have one fear.
What I always liked about my setups is turning off the "quiet" kernel parameter so I see services starting at boot. If something goes wrong there, I'll immediately notice. If multiple service of the same category fail (for example, multiple hard drives can't mount), I can interpret from that that maybe my USB hub is broken. Or, as they are all btrfs drives, maybe the latest btrfs update broke something.
Now, coming back to the bsod, what I fear is that it simplifies too much. How will the bsod behave when multiple services fail? Will it actually have a conclusion of both? Will it only show failure of the first or the latter one? Will it stop the actual boot process after the first service failed?
I don't think that you'll get the same level of verbosity with the bsod compared to the output we can already enable now, which in my opinion is also not that hard to understand for average users.
But at the end, it's a service. If we don't want it, we can disable it. :)
So the proponent of BOSD in the systemd project is a WORKER AT MICROSOFT. Embrace, extend, and extinguish
God forbid Linux give us info about why the computer won’t start correctly. I would prefer a black screen with no text personally.
I think it's a good thing.
Especially for newer linux users.
Because now what happens is computer freezes and never unfreezes or it just reboots without the user or a program that the user is using telling the computer to do so.
If there is a screen that tells you exactly what happened (like the Windows BSOD) at least it removed a lot of confusion, but i do hope that we can customize it cuz i want nyan cat as the background
Edit: maybe i should watch the whole video before commenting but still... If it was like i thought that would be really nice
Thanks for explaining how BSOD actually works. Personally I've never had a BSOD caused by Windows itself, it was always a nasty driver or RAM screwing up. It was massively helpful for the latter when my Mint install gave a vague error screen and nothing else.
I think having easier to understand error messages is an absolutely good thing for the end user. I shouldnt have to dig thru logs to understand why something goes wrong.
I've had windows cause a BSOD in a way that looks like a driver because a windows update deprecated the driver I purposely didn't want to update without warning.
@@colinchichester1809 Deliberately keeping drivers at older versions usually is unsupported user behaviour (other than the latest version having a bug that requires staying on the penultimate version until it is fixed). From the developer's (both driver and Windows devs) perspectives, you fafo'd.
@@Ignisamisince I was forced to take a performance hit from the new driver I'll agree to disagree
@@colinchichester1809If the driver is deprecated, that means it's no longer compatible. That's not an OS problem.
@@colinchichester1809You wanted performance, but you opted for an incompatible driver. Your or the driver writer's fault. Still not the OS.
If it wasn't blue, if it wasn't called a bsod, it'd probably be fine. It sounds like a useful function, but making it blue really does throw back to an operating system that most Linux users are actively trying to move away from. Case in point, we already have kernel panics, and we don't cry about that.
Come on, BSODs on Linux are just hilarious! Show you Linux guys can laugh at yourselves.
Oh, we can, but we often choose not to install the humour package on our systems because it was just bloat. It makes us far more efficient than those Windows jokers.
Sounds good, I'm looking forward to it's functionality. I only have one annoyance with it: why blue?
why not black? or red? Seems red would fit more with an important error, and would be a much more appropriate default for the situation. As it is now, it seems the only reason for it being blue is because that's how Windows does it.
wasn't there some talk about an emergency kernel recently? something that kexecs after a panic just to collect and store the diagnostic info persistently on the drive
BSOD is genuinely one of the best Windows features. Importing it is a good idea.
Love it. If I try to install arch manually, this is amazing for troubleshooting.
I had a windows install that BSOD'd randomly for months, and the sheer wild inconsistency in error messages and lack of events in event viewer made it ridiculously difficult to diagnose. I eventually had an event recorded that matched a number of error messages suggesting a failing SSD. The SSD in question actually had 99% drive health in SMART, passed all tests and fsck. Replacing the SATA cable fixed everything.
I'm since enjoying linux, though I do wonder how it would have handled that particular hardware issue?
it definitely can't see the Health of individual parts, so it'll probably still point towards the failing end point (SSD in your case), but I at least hope it'll be able to Specify "Hey look I'm seeing your SSD is failing on my end, Check it and everything between that and *insert last measurable point in between* for potential Failures"
Not much better frankly, I don't know of any native handling that would have gone 'aha, it's a cable' or even very clearly isolated the problem. Windows handled it poorly because it read symptoms and hasn't been optimized to be aware of this possibility. If a developer on either side made this kind of issue their focus, you'd probably see it handled better. But a bad cable is rare enough that I doubt it's ever been a particular focus to isolate.
@@ShadowOfTheSPQRIT's not Windows, it's the lack of diagnostic data coming from cables. They just can't do that.
@@Christobanistan Yes, what you'd have to do - I'm not sure how realistic this would even be - is design something around the consistent symptoms of this issue and have that as a suggested possibility for what's going wrong.
BSOD on Linux would not even be bad idea. But instead of blue screen it should be a nyan cat. That said I kind of like the log spam right before it goes cold. Usually gives quite good hint what happened. Like if there is suddenly billion gpu related errors it's a good guess to say that something happened to the gpu.
This is a good idea. Nyan cat, mocking me. :)
the issue is, that there are 2 camps of people: tech savy enough, to see what possibly went wrong during the boot process, and are able to fix it, by going though journald or something, and the not-tech savy people, who regardless of whether they see the systemd boot messages, or an imitation BSOD, will absolutely ignore everything written on there and instead just complain, or google "linux pc wont boot"
if you work in IT, you should know that even the capacity to read and understand a human-readable error message wont do shit, because most people will call their IT department with only the words "my pc doesnt work" and thats it
when will linux get ctrl+alt+del ?? when linux freezes currently have to remember some weird keyboard combo with f1 or f2 or some crap then login to a terminal and open i forget what the program is called then figure out the keyboard keys to find the frozen process then kill it? why the hell can't we have a GUI ?
255-rc3 is in the Arch test repo. Brought it in yesterday. No blue screens yet but might install KDE Plasma 6 and see.
I mean, currently I can see the log while booting, and I can get some idea about what is wrong when the system does not boot. I can go and search the web. I am wondering how much contextual information will be there in the bsod. The problem with the Windows bsod was obviously that it told you nothing about what went wrong, just some obscure error code. I am also wondering if it would be enabled by default on Arch and such systems.
I don't get the BSOD hate they have saved me from a bigger headache every time I see them. Just read the hecking error message then start your diagnosis from that point.
How exactly will this interact with verbose boot logging?
I always remove the "quiet" setting from the kernel cmd so i can see everything that's happening during bootup.
Will this BSOD just clear the screen and remove all the usefull information to replace it with some stupid QR-Code which is literally impossible to make sense of without a separate device?!
For me this sounds like it's trying to fix an issue caused by bad configuration practices.
I have seen _many_ kernel panics in my early days of Linux usage, almost on a daily basis. I came into contact with Linux through the RasPi and somehow managed to break almost any install of RasPiOS ("Raspbian" back then) within 24h of installing it.
The way I noticed that it's something else than Window$ (my father's computer ran Win7) was that I couldn't find `home` and so on when plugging the Pi's MicroSD card into the computer, and also through the strangely small volume size.
Pain serves the same purpose: it tells you that the body is damaged. Yet most of the time (or at least a lot of it), we try to power through the sensation or try to kill it immediately instead of--not in addition to--addressing the core problem.
A survival mechanism, usually. Can't stop running from the leopard!
I actually don't remember a time when bsod was useful for me, because it always just restarts before I could read anything. Haven't used windows in 5 years but still remember needing to photograph the screen so I can read the message
They should make it so if these is no useful information it can give it just puts a sad penguin on the screen.
The errors Windows BSOD gives are hard to search on. Linux should have an actual solution: "Search This:" People will not remember QR Codes, but they will remember what they type in.
I think Brodie's face in the thumbnail kinda sums up the face that on old linux sysadmin would make reading the title of this video
Still prefer the good old Guru Meditation...
I remember back when I still used Windows, the information in BSOD was absurdly vague, and googling for what was mentioned there lead to way to many different situations producing the same error code...
I just hope it’s actually blue. We don’t need Linux doing Windows’ newer Cerulean Screen of Sadness too.
I've never understood people hating BSOD or the newer GSOD. Brodie is 100% correct in saying that's just shooting the messenger.
Cool, systemd v255 should be useful, but I want to know what they're going to include in the next version, systemd v0.
This would have been a great one to save for April 1st.
Honestly, I'm kinda happy that BSOD is coming to Linux. I'd much rather get a BSOD with a debug message than have my PC freeze or silently crash without any messages whenever I do something stupid.
The number of times I've fully saturated the RAM on my PC and waited for it to unfreeze, only to give up and hard reboot without any information as to why my system froze. Took me a while to figure out it was RAM.
Why blue as default? Why not something fun like a light orange?
Thet should do black text on white background flashbang
SystemD's adoption of a BSOD will be the latest in a series of choices that make you wonder if the 'D' really stands for 'Dementia'. From its inception, it seems to have strayed far from Linux's ethos of efficiency and simplicity, ironically complicating the system it was meant to streamline...
Do you think arch would ever provide openrc as an install option?
Officially no, there's no point, but it is in the AUR with instructions on the Arch Wiki
This is good news. If the system crashes, it gives the reason without having to dig through log files. This will help us help others.
As a general rule, I consider any feature bitched about by Phoronix or Devuan users, a good feature.
I really want the name to be trolling, because it sure is ruffling exactly the right kind of people*, but what else would you call it? It's been BSOD for decades now and I find myself incapable of thinking of that screen as something else.
* the kind that instantly REEEEEE into orbit when systemd adds a very useful (and optional) feature, like making information about critical system failures more accessible.
If the information it's giving is actually useful, I like the change, especially because it's just for boot not in usage.
I wish the kernel itself would go further and scribble over the GUI if it panics. It's no good writing any sort of stack trace in panic (possibly too unsafe to write to disk) situations on a VT the user can't see or switch to.
If it's broken beyond usability I want to see why if possible, not just a freeze.
It'll be fine.
You can always disable the service if you don't want it.
Maybe being able to turn it off to use whats currently in or a way to implement both
0:57 Wait, what? Doesn't Gentoo ship with OpenRC instead of systemd by default?
when you install Gentoo you can choose between systemd or OpenRC but %99.99 of Gentoo users use OpenRC
@@Eren_Yeager_is_the_GOATYep, that's exactly what I thought!
@@Eren_Yeager_is_the_GOATyou can choose between four, the most popular being openrc
This feature is not a BSOD. The BSOD is a kernel panic that can happen any time, while this is just the init process failing while booting. Should have made it black or gray or pink, to avoid the confusion.
Anyway, a boot failure is hard to diagnose and fix, so some help in that department would be useful. However, I don't see that giving an informative error message would be the focus, so I have doubts how much this will help.
Given my own boot failures are half "unable to find root" (so handled by GRUB not systemd) and half "unable to mount partition" (handled by systemd), its not exactly going to catch all boot failures either. Still, I don't get why anyone would think it existing is a bad thing.
The biggest issue with BSoDs on Windows (at least in the XP era - I've not daily drove a Windows machine since the days of Vista) weren't that they appeared.. but that what little information they provided was cryptic at best (driver filename and internal error name like PAGE_FAULT_IN_NONPAGED_AREA) and just completely useless at worst (nothing more than the STOP code). Doesn't help that, at least in my install of XP back in the day, the default action the machine would perform on a BSoD is an immediate, automatic reboot before you could see it. Woke up to a desktop that had spent half the night boot looping (try to boot -> BSoD during -> immediately reboot -> BSoD....) from a stick of RAM going bad (which took ages to track down... memtest86 with 2-3 passes had no error... from a power cycled boot. Reboot into it via the reset switch and it started throwing errors immediately) and ended up with a corrupted system32 somehow. TRWTF, as always, is Windows.
If the implementations using systemd-bsod can provide *something* of debugging value, then they've succeeded.
IMO the XP-era BSoD was the best. The fact that it decodes the code to a string and tells you the module name is more info than most systems. In newer versions of Windows it just says "Something went wrong" and in older versions they didn't have the explanation message.
@@eDoc2020 XP, Vista, and 7 were the better BSODs.
Nothing will get your attention more than a white screen of death.
I can't wait to rice my systemd (B)SOD.