systemd's newest contribution to FOSS is the BSoD - but this time, new and improved, with QR codes!
Not a joke, truth. QR codes!
The systemd-bsod component is currently listed as "experimental" and "subject to change." But the functionality is simple: any logged error message that reaches the LOG_EMERG level will be displayed full-screen to allow people to take a photo or write it down. Phoronix reports that, as with BSODs in modern Windows, the Linux version will also generate a QR code to make it easier to look up information on your phone.
New FOSS chant? "Stay free, use BSD, never see B-S-o-D!"
This discussion was created by hubie (1068) for logged-in users only, but now has been archived.
No new comments can be posted.
BSOD in Pipeline for All Major Linuxes
|
Log In/Create an Account
| Top
| 40 comments
| Search Discussion
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
(1)
(Score: 5, Insightful) by Runaway1956 on Sunday December 10, @12:14AM
Trick Linux into throwing up a BSoD, with your own custom malware QR code. The sucker scans the code, you can tie his phone life to his computer life, while prompting him to download whichever malware you want him to run. NSA is right on top of this, with the FBI close behind.
(Score: 5, Insightful) by Snospar on Sunday December 10, @12:35AM (13 children)
I've been running Linux for more years than I care to think and the number of real "hard lock", nothing-for-it-but-the-reboot moments are vanishingly small. I always thought that the main "excuse" for Windows BSOD was poor quality "third party" drivers that somehow managed to cripple the fine software that Windows isn't. That sort of nonsense doesn't impact most Linux installations where we can rely on the high quality hardware drivers in the main Kernel tree to do almost all the heavy lifting. Unfortunately many of us are still tied to the non-free binary blobs from Nvidia but other than that I'm struggling to think of the last time I had to compile a third party driver.
Is this really something that the systemd crowd need? Why? I'm systemd free (and loving it) but I am genuinely struggling to think about Linux crashes being so common that I would need this type of "diagnostic tool". Maybe systemd is the source of instability, that would explain why I'm not experiencing it. Lucky me.
Huge thanks to all the Soylent volunteers without whom this community (and this post) would not be possible.
(Score: 5, Insightful) by RedGreen on Sunday December 10, @01:02AM (2 children)
"Is this really something that the systemd crowd need? . Maybe systemd is the source of instability, that would explain why I'm not experiencing it. Lucky me."
I have yet to experience any related crashes due to it, unlike that piece of garbage pulse audio also by Pottering. Play any video file when using it and my computer would lockup within seconds using that trash. They need it to complete the windows experience from the undercover mole who started it, to foist it upon GNU/Linux, like that Gnome weasel De Icaza who came before him. Both have moved on to their true calling at Microsoft now. With friends like them and Redhat the lockdown of GNU/Linux with proprietary tools is coming as soon as they can engineer it. The do one thing and do it well philosophy has been thrown out the window with the systemd tentacles reaching for everything it can assimilate just like the Borg.
"I modded down, down, down, and the flames went higher." -- Sven Olsen
(Score: 0) by Anonymous Coward on Monday December 11, @08:50PM (1 child)
Use pipewire instead. Pulsaudio is on its way out.
(Score: 2) by RedGreen on Tuesday December 12, @02:56AM
"Use pipewire instead. Pulsaudio is on its way out."
I did it already when searching for the solution to the garbage that is pulse audio. Pipewire has never failed me once after the install it just keeps giving me sound any time I need it. Pulse within seconds of a video starting hard hang of the machine with no option but push power button to get it back, within couple of songs same thing. I would add the old OSS, ALSA all those supposed inferior needing replacement had always worked too, they may have needed tweaking to configs, but never seen nothing like it until that Pulse trash got on my system.
"I modded down, down, down, and the flames went higher." -- Sven Olsen
(Score: 2) by Tork on Sunday December 10, @01:13AM (2 children)
Windows users are right there with you. The daily BSOD died for everyone who moved on from windows ME. A lot of ppl on the green site still haven't caught up to that lill nugget of info, tho.
🏳️🌈 Proud Ally 🏳️🌈
(Score: 1) by Runaway1956 on Sunday December 10, @02:33AM (1 child)
I think blue screens were still pretty common with XP. Service Pack 1 seemed to help. Service Pack 2 made XP pretty damned stable, IMO. Win 7 was pretty stable, as well. I can't speak for Vista, Win 8 or Win 10, but my Win11 installations are stable. But, I most definitely remember the first iteration of XP doing crazy things for no good reason.
(Score: 2) by Freeman on Tuesday December 12, @03:53PM
I've had plenty of blue screens on pretty much every version of Windows. Aside from the earliest versions that didn't have a blue screen. (Win 3.1, etc.)
Win 95 and 98, most definitely.
Win 2000 definitely, but I only used this on school computers.
Win NT can't say I actually used it. I've not used much in the way of Server versions of Windows either.
Windows XP most definitely as well. It was very flaky at the start, but probably the most annoying thing was devices not having driver support.
Windows Vista I got the unpleasant experience of using this to help parents with stuff. As they got the latest and greatest and it sure wasn't great. BSOD to the pain and again horrible driver support at the start.
Windows 7 seemed to be the most stable for me, though it certainly had the occasional BSOD.
Windows 8 I was thankfully able to just outright skip. Which is what my place of work did as well. Seriously, why make your OS tablet centric, when your main clients are desktop/laptop users? Pretty much only the expensive laptops had touchscreens as well.
Windows 9? I mean, they skipped 4, 5, and 6, so may as well skip 9.
Windows 10 was a return to sanity, for the most part. Fairly stable, but now they want to shove Windows Accounts down your throat. BSODs are a thing, but at this point it's more like an old friend.
Windows 11, it's essentially Windows 10, but with a few dumb UI changes and even more forceful with the Windows Account adoption. No, I don't want the Start Menu in the center of the screen. Put it back on the left side like you've had since nearly forever. Also, BSOD, why change it?
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 4, Touché) by driverless on Sunday December 10, @07:22AM
Given that systemd really, really wants to be everythingd, the answer is almost certainly "yes". I'm surprised no-one's found a flight simulator [iu.edu] in it yet.
(Score: 2, Insightful) by pTamok on Sunday December 10, @02:33PM (1 child)
(Score: 2) by loonycyborg on Sunday December 10, @06:27PM
I remember those cases of "swap frenzy". Once you run out of ram it's long waiting time for frozen system, or reisub. But those are fixed in recent linux kernels at least. And bsod won't help with them anyway because they don't result in emergency messages in syslog.
(Score: 3, Touché) by owl on Sunday December 10, @04:38PM (1 child)
[sarcam]
Why of course yes, now that the systemd creator and chief maintainer works for Microsoft, yes, Linux needs to be much more windows like in every way, including throwing BSOD's five times a day.
[/sarcasm]
(Score: 2) by turgid on Sunday December 10, @08:08PM
So the Windows kernel is going away to be replaced by Linux? We already have WSL and VSCode.
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].
(Score: 2) by mcgrew on Sunday December 10, @09:14PM
I've used Linux since Mandrake and had three lockups, all of which were caused by hardware; the computers all needed major repairs, like a new CPU.
mcgrewbooks.com mcgrew.info nooze.org
(Score: 3, Interesting) by Immerman on Monday December 11, @03:31PM
Excuse being the active word. Maybe the drivers really are the problem - but it's not like Microsoft is actually going to admit "our software is buggy crap" either, they've been blaming drivers for their own problems since the Windows 3.1 days.
Without reading TFA it doesn't sound like this is necessarily a "hard lock" screen, you might just click past it - what's the current response to a LOG_EMERG? It could just be a "something really bad just happened" alert, the equivalent of your car making a Bad Noise, that might normally go unnoticed unless you're in the habit of checking your log files.
Honestly though, if a "hard lock" is already the response then it becomes even more important to actually provide as much information as possible - *especially* if it's very rare that anyone will see it. Random crashes are pretty much the most impossible problem to debug, in large part because they can cause logging systems, etc. to abort before committing the relevant details to disk. Any information not provided at the time of the problem may go completely unseen.
(Score: 2, Insightful) by khallow on Sunday December 10, @01:02AM (4 children)
Or we could be logging that all into text readable files so you can copy/paste the text of the error message directly into web search. You don't even need to configure yet another wacko add-on if you want more reported than the default.
(Score: 5, Insightful) by darkfeline on Sunday December 10, @01:50AM (3 children)
> Or we could be logging that all into text readable files so you can copy/paste the text of the error message directly into web search.
That doesn't work if the LOG_EMERG is because the system is actually unusable. You would very likely:
not be able to read the file
not be able to copy
not be able to paste
not be able to open a web browser
not have network connectivity
Now, you could power off the system, extract the drive, mount it on a rescue system, check the logs and then copy paste into a web browser.
Or you could just whip out your phone and scan the QR code.
This is of course assuming the LOG_EMERG is a LOG_EMERG, but if it isn't, that's a problem with the application and not this daemon.
Join the SDF Public Access UNIX System today!
(Score: 4, Insightful) by jb on Sunday December 10, @06:13AM (1 child)
It works just fine, so long as your console is a real teletype (or these days more commonly a serial mux with a history buffer, ideally streamed from thence to persistent storage on another device).
There's a reason that any sane syslog.conf(5) contains a line that reads: *.emerg *
If on any given system emergency log entries actually matter then having a console that disappears altogether on fatal error is outright negligent.
(Score: 2) by Immerman on Monday December 11, @03:42PM
Lovely. And how many people actually have an external logging device connected at all times, so that one-in-a-million crash can get diagnosed?
Close enough to zero as to be an extremely rare treat for anyone trying to find the cause of a rare intermittent problem based on user error reports.
Critical errors that may disrupt the logging system are the most important time to display the messages "live", and since it's practically free to do so, why wouldn't you? It's not like they're doing this *instead* of logging, you're just in a situation where logging may not be working, and it's important to keep key information from disappearing entirely.
(Score: 2) by Rich on Sunday December 10, @03:02PM
All that could be achieved if future UEFI images contained a port of Microsoft Edge :)
Not to give them ideas... but they probably already do similar stuff with WSL2, it's just that the Kernel Panic goes straight into Window "Telemetry" there and is scooped up by M$.
(Score: 4, Informative) by darkfeline on Sunday December 10, @01:46AM (4 children)
Let's start by dispelling some FUD.
Linux has always had a BSOD (Black screen of death), for kernel panics/crashes. You rarely saw it because Linux is stable, not because Linux "didn't have the BSOD feature".
This is a new systemd component, which is to say it's a standalone program with the systemd "brand". It's up to distros whether or not they ship it, enable it, etc, assuming it makes it past "experimental" and "subject to change".
Second of all, all this does is pop up a screen when a program syslogs LOG_EMERG. It's up to the distro/DE and/or various software to decide what gets logged as such (again, assuming the distro decides to ship it).
The QR code is a red herring, is anyone actually arguing that being able to copy the error text via camera scan is bad? God forbid not having to type in a long string on a smartphone screen.
I suspect this is mainly for POS/kiosk deployments, where a technician can scan the QR code and service the terminal.
I might be tempted to enable this just to see what, if anything, actually gets logged as LOG_EMERG. I sure as hell don't regularly check the syslogs on my laptop so I wouldn't know if there were any LOG_EMERG. If there were something that is correctly LOG_EMERG, then it'd be good to get notified.
Join the SDF Public Access UNIX System today!
(Score: 4, Touché) by coolgopher on Sunday December 10, @02:52AM
I don't know about the rest of you, but for quite some time now I've been able to pull out my phone, take a photo of the console crash log, and have the photo app OCR that no worries. No QR code needed, and no risk of someone throwing up a malwared QR code. This really feels like a solution in search of a problem.
(Score: 4, Touché) by jb on Sunday December 10, @06:16AM (1 child)
Great, so now any user can bring down the system with no more than `logger -t user.emerg foo`?
(Score: 2) by Immerman on Monday December 11, @03:47PM
And where do you get from "pop up a message" to "bring down the system"?
If a legit LOG_EMERG message has been reported, then the system is already coming down. And if it's not actually coming down, displaying an alert isn't going to change that.
I mean, I suppose it's possible that they give no way to close the alert and lock down the system in response... but why would anyone do that?
(Score: 2) by driverless on Sunday December 10, @06:22AM
I think the QR code is just a way of doing something less useless than the Windows standard ":-( Something went wrong", combined with a hex code you can Google to find hundreds of other people all with the same problem that no-one knows the answer to.
(Score: 2, Interesting) by Anonymous Coward on Sunday December 10, @05:41AM (2 children)
Is anyone seriously surprised by this?
Whatever, it's perfect for the crowd that systemd serves - at least there're better off with systemd than they are with a commercial OS.
We never missed BSoD debug information 'back in the day' when I recall last seeing Linux lose its shit to the extent that you'd have needed that.
Instead, we had (and still Have, provided your crummy distro hasn't disabled it!) the Magic SysRq key, which is *far* better than just a BSoD.
For instance, if your graphics driver failed, locking up and taking the screen with it, you'd hold down alt, and type 'SysRq r e i s u', you'd wait for the HDD led to stop blinking, and then hit 'b'
This would safely reboot the system, ensuring all files have a chance to be properly written, that the files were flushed to disk, that the filesystems were remounted read-only, and you waiting for the HDD activity to go away ensured that they were finished before forcing a reboot.
End effect: We could have plain text files in /var/log/ which we could then go open, read and find out what broke.
They'd be complete and the file system wouldn't be corrupt, because the Magic SysRq key would allow you to organise a safe shutdown/reboot.
In practise, this was far more useful than a BSoD you'd have to get information from.
So, BSoD for Linux? Sure, why not.
It's not like you'll be seeing those happen much, not apart from the occasional hardware failure, anyway.
The only thing which I would consider a step better, is what old classic macs could be made to do: Drop in to a programmer so you could go poking around memory to be able to look directly at what was going on.
Personally, this seems somewhat of a killer-app for a resurgence of Openfirmware, which surely would make for a better peripheral hardware protocol standard than the clumsy written-by-committee mess that USB is. Just do it over PCIe, even for usb-alike things, and be done with it. PCIe's physical and logical layer, Got It Right (tm).
When you have to drive hardware to do things in the here and now, Forth is kinda a super-power, and still in use like a secret weapon in a surprising number of places. An interactive forth prompt useful words loaded would shit all over a no-interactivity BSoD any time. Even if you'd most likely just do the things we used SysRq for. Forth works well in this app, because you can put a whole usable system into like 2 pages of ram that can stay resident in the L1 cache.
Anyway, this reminds me how Windows had world-leading failover clustering server OS support long before anyone else.
Because they *needed* it. And it's evil genious to sell more computers and thus licenses to their customer-victims, whilst still maintaining the product as a perfect-yet-somehow-perpetually-broken-and-in-need-of-maintenance-and-yet-still-essential thing.
So what if we're getting this improvement so late, now? We weren't the worse off for missing it (hah!) and we won't be any worse off in real terms with it. (Expect to see every instance it's ever seen in public posted to social media: much like UFO, bigfoot or unicorn sightings).
Hell, go listen to that old ex-Microsoft YT content guy - BSoD were Blue because they needed to be able to notice it happening across of room full of computers soak-testing software to try to find when it breaks.
Not exactly a very engineer-grade way to do things, but I guess when your own technology is so bad you don't know how your own kernel works, you have to do *something*.
(Score: 0) by Anonymous Coward on Sunday December 10, @04:22PM
I think DEC, IBM and Tandem had that before Windows.
As for why Windows needed it, it is not so much due to instability and BSODs - later versions of Windows Server (2003 onwards) were fairly stable (at least on decent server hardware).
It's because of the Windows conventions that in normal cases makes it hard for you to update files that are in use (open files are usually locked). So you end up having to reboot for a higher percentage of scenarios. Reboot = noticeable downtime. So you may need another server if you don't want users to notice.
The Unix convention allows files in use to be deleted/renamed. So in more cases you could just update+restart services without rebooting.
(Score: 2) by DannyB on Monday December 11, @08:44PM
I remember the Mac programmer's switch you could snap on to the side. Most people did, because it had two buttons,the programmer's button and a Reset button.
I knew how to use that mini-macsbug, and I memorized a few favorite Mac Traps.
One time I was at my friends. He had a word processing file that he had not yet saved. He switched to another program for a moment (word processor now in background), then he got the dreaded Mac Timebomb screen (the Mac counterpart of the BSOD). A timebomb dialog box pops up on screen.
He was about to reset, giving in to losing his entire document.
I hollered WAIT!! Try this . . .
I had him press the programmer's button NOT the reset button. And told him to be careful because everyone is used to using the reset button.
The mini-macsbug appears on screen. I tell him to type in two commands:
1. SM 0 A9F4
2. G 0
The first command puts the 16 bit value A9F4 into memory location zero. A9F4 is the trap instruction for Exit To Shell. The second command tells mini-macsbug to resume execution at memory location zero. The A9F4 is executed, which is an invalid 68000 instruction, traps to the executive which looks it up in a software table to execute a ROM routine, and exits the crashed program (which was in front).
I told him to immediately save his document and then immediately reboot.
Saved the day!
Of course, to him, it seemed like the magical hacker incantations in hollywood movies. It really was something I happened to know. I was a classic Mac developer. I wrote Timbuktu and a few other things. I preferred TMON to Macsbug.
When trying to solve a problem don't ask who suffers from the problem, ask who profits from the problem.
(Score: 5, Funny) by namefags_are_jerks on Sunday December 10, @07:38AM (4 children)
It should resemble the Amiga's Guru Meditation screens.
That is all.
(Score: 2) by looorg on Sunday December 10, @01:10PM (3 children)
It's where it showed its graphical superiority by having that red blinking border compared to just a static blue/black screen with white text.
(Score: 3, Funny) by jb on Monday December 11, @05:17AM (2 children)
Like most things Amiga, it was also full of in-jokes.
For example, the last line usually displayed in English: "Press the left mouse button to continue" or something like that. But (at least up until Kickstart 1.2; don't think I ever saw it with later releases) the "worst" guru meditation errors would display that line in German instead: "druecken eine Maustrate..." (which as it happens is slightly more accurate, since hitting either the right or left button triggered a warm boot, regardless of which language the message appeared in).
No doubt the systemd equivalent's jokes will be far less subtle. I imagine the vast majority of QR-code generation for the systemd bsod will involve some sort of Rick-rolling...
(Score: 2) by looorg on Monday December 11, @11:03AM
There was a lot of things with it that left things to be desired as far as debugging or help, some messages even just translated into HELP in hex. After all most, if not all, was unrecoverable. Later on even the recoverable system errors usually made the entire system so unstable you better restart it anyway. But overall it was just a message telling you it crashed. But it might not have been what actually brought the system down as it was really just the last thing it managed to trap, which may or may not have been related to the crashing part. A lot of them was on the same level of helpfulness as the once on (early) Macs -- there was an error cause there was an error ...
(Score: 2) by Immerman on Monday December 11, @03:53PM
The mouse button message sounds like a variant on the old adage to always have your pause messages read "Hit a key to continue" instead of "Hit any key to continue" because the latter WILL reliably generate a huge number of "I don't have an 'any' key" calls to the help desk.
You want reading comprehension fails to still reliably guide the users to an acceptable course of action, if only for your own sanity dealing with the "bug" reports.
(Score: 2) by turgid on Sunday December 10, @09:20AM (3 children)
The great thing about Linux is that it is FOSS. There are many different distributions to choose from. You don't need to be held hostage by any of them, or by any one piece of software (apart from the kernel, by definition). That includes systemd. There are still modern, stable, reliable, usable distributions that do not have systemd, Slackware [slackware.com], for example. You still have a choice. Exercise it.
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].
(Score: 2) by bart9h on Sunday December 10, @08:21PM (2 children)
Yes.
No.
Or is it? I haven't used Slackware in the last 15 years, but for what I recon I would not call it a modern distro.
(Score: 2) by turgid on Sunday December 10, @08:38PM
It's very modern. It has all the latest stuff.
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].
(Score: 1) by CaptNemo on Tuesday December 12, @10:58PM
It's apparent that you haven't used it.
Slacware-current has newer kernels than other more popular distros. It's completely up to date.
But, even 15 years ago, it was as current as other distros.
(Score: 5, Interesting) by Mojibake Tengu on Sunday December 10, @09:25AM
I consider this a critical vulnerability.
Not only that, it's easily spoofable by a python script. Or even shell script. QR code is a transfer that crosses network air gap, another vulnerability. Serves for extraction.
If an intrusion operation needs time, confusing user into mentally passive state watching the screen and "solving problem" is the very first tactics to be employed.
Respect Authorities. Know your social status. Woke responsibly.
(Score: 3, Informative) by Snort on Sunday December 10, @09:01PM
Systemd head Lennart Poettering now works for Microsoft.
(Score: 2) by Zoot on Monday December 11, @04:27AM
As someone famously said about 20 years ago...
The great thing about open-source software is that you can have any color Screen of Death that you want!
Z.