This happend to me right noww as I tried to write a gui task manager for the GNU/Linux OS
My first programming related memory is of the QBasic interpreter.
I had written some code I was quite happy with, but not saved it yet. As part of a subroutine for sound output, I quickly wrote a loop from 20 to 20000 to output a test signal over 1 second each with that frequency via the PC speaker and hit execute.
Realizing my mistake, It being MS-DOS and thus single-threaded, I couldn’t Ctrl+C out of it without killing QBasic altogether and losing my code. I couldn’t turn town the PC speaker.
I ended up closing various doors between the PC and me and waiting it out.
What language were you using?
Python maybe?I use C and C++ and my IDEs save to disk before compiling. Makes sense to not try compiling when there are potentially 2 versions (one on RAM or
/tmp
and one on Disk) and the build system might be running multiple commands, which the IDE may/may not know of, in my case.How are you crashing your system?! Crashing program sure, but the entire system?
rm -rf <some placeholder>
Works for
.
current directory. Yay!… also works for
/
system root. 🔥 Nay!Does it? I thought / specifically was protected, and you needed to add --no-preserve-root.
That won’t crash your kernel, and I was more curious about the OPs example. Task management is basically reading some files, and sending signals, it should be near impossible to crash the system.
I believe it does crash the system eventually as important buts start to go missing?
Try it out on your own system.
:(){ :|:& };:
It’s totally possible
Doesn’t explain OPs task management example. And won’t crash the kernel, just make things unresponsive
There’s this game “HyperRougue”. Run it on Arch.
hyperrogue-git version 13.0d.r60.g27fb2d92-1
Go to
settings -> 3D configuration -> projection -> projection type ->
. Cycle through the projection types. One of them causes something good enough to call a crash.I don’t remember anymore if it was just a display driver crash or a kernel crash and I haven’t updated to a newer version (which might have fixed it).
Doesn’t even startup on my box, but doesn’t crash the kernel or system either, just regular application crash
Doesn’t even startup on my box,
It needs to startup and then go to that point (after you select the projection) to cause the crash.
It definitely caused something other than the application to get into an invalid state. Which is why I am apprehensive about trying it out again to answer your comment. Probably was the display driver, which is why it didn’t just turn off after that.
it didn’t crash the kernel, it just killed every process that isn’t run by the root user, which kind of feels like a crash
Ah, that definitely would feel like a crash. Sent kill signal to cgroup accidentally? Or just iterate over all processes and signal them all?
probably the later, but idk how, all I did was insert a string in the following command like this:
``Command::new(“bash”)
.arg(“-c”) .arg(format!(“ps -aux | grep -i "}" }’ | xagrs kill -9”, input)
.output()
.expect(“error”);``
I’ve tested the command and it worked flawlessly in the terminal, but I have no idea what I’m doing, since I’m new to rust and never worked with this library
There are rust libraries to send signals, might be better to use those rather than calling bash. eg. https://docs.rs/nix/latest/nix/sys/signal/index.html
I’m guessing if input was “”, then it would sigkill all processes? Less confident, but some functions behave slightly differently in an interactive console vs a non interactive, maybe
ps
has a different format when used non interactively?Aside, you want three backticks and a newline to get code formatting :)
thx, btw I figured it out:
I forgot to trimm the string, so it had a line break in it which lead to grep showing the processes from the term I put in + all processes that contain a space/linebreak and appearently all processes shown by ps aux contain some kind of space (makes sense, since there are spaces between the user, pid, etc) so yeah, I ended up trying to kill every process on the system, but it only killed the user processes, since I ran everything without sudo
OPs example was task management, which doesn’t require kernel modules.
Try it again
Do you know the definition of insanity?
Do you know the definition of insanity?
do you know software developers?
But did you get the reference?
I did, don’t worry
What’s really insane is that sometimes the second identical test actually works.
Then you breathe a sigh of relief, merge it with a comment of “bug fix”, write no documentation–especially about how it failed testing, and quit the gig during the inevitable helpdesk explosion; walking away from the fireball like the Michael Bay maniac you are.
This is why VM snapshotting is so valuable.
My IDE is my real workstation, and it hosts a VM in which I can plop some code, run it, crash, revert and try again.