Greetings, I just came across this.
Posting as FYI and for any comments.
Peace.
Windows kernel drivers have a restricted set of languages they are allowed to be written in.
They're currently only allowed in C++ and C.
2 Likes
Anyone have info confirming that stack trace is actually the one that caused the problem?
3 Likes
I think they are considering rust. it's just the ddk existed long before rust and more works need to be done. see e.g.:
5 Likes
The twitter account also posted this:
There's also no way of knowing if that screenshot is related to the current bug.
2 Likes
Ah, the conspiracy theory BS strikes again ..... what a surprise (not!).
well, I would like to remind everyone of this:
7 Likes
As a long-time member of both URLO and IRLO, personally I have yet to see observable evidence of any "cabal of woke tards that are doing strange things."
But also, it's Xitter so yeah... Not exactly unexpected behavior over there.
13 Likes
To be honest, a Rust version would probably unexpectedly panic, and they wouldn't have handled that gracefully either.
They've pushed globally an untested code update on Friday. That's more of a fault of a broken organization culture, not C++.
41 Likes
The effect was felt outside the organization that pushed the untested code. The culture of the entire system is broken.
We are dancing on a house of cards.
1 Like
I am happy to report I chose Southwest for yesterday’s travel. It was surreal watching all those monitors displaying a death loop while we boarded. Our flight was 17 minutes early.
As to the stack dump… Windows no longer displays that. They would have had to have access to an actual dump file.
And, that’s not a NULL pointer. It’s an uninitialized pointer. Which, in my mind, is really sloppy work typically requiring ignoring a compiler warning.
2 Likes
From the last news I heard the problem was they pushed a driver file, some .sys something, that was full of zeros instead of actual code. When that fails to run we end up in a boot loop. I cannot imagine how that happened but it sounds like it would have happened no matter what language created what was supposed to be in that file.
1 Like
This is not related to null bytes contained within Channel File 291 or any other Channel File.
From cloudstike
However, given that they encrypt their configuration files, give them filenames ending in .sys, and put them in the System32\drivers folder, I think this confusion is on them too.
Interesting. Thanks. Looking forward to seeing what the root cause was. I hope they are open about that.
For what it's worth, Tavis Ormandy has posted a thread on X/Twitter debunking this post, though you'll need an account to read the whole thing. The short version is that it was an out-of-bounds read (which would have been a panic in Rust):
3 Likes
We can actually embed the whole thread.
10 Likes
It still seems like something that Rust culturally discourages: writing your own config language and parser.
Also where Rust nominally shines: writing parsers for untrusted inputs.
But really, organizational issue. No language can help with that.
1 Like
Yeah, the "Rust wouldn't have saved you if you're trying this hard to screw up" theme is probably about half true - the bigger effect would probably be that you're a lot more careful in general after rustc has kicked you in the goulies for a year straight.
Even if an out of range read was the cause and would have been caught in Rust in this specific case, you're eventually going to find a way to push a crash if you're trying this hard to not check your work.
As a "woke tard doing strange things" myself, I feel slighted to not be invited to the cabal. What's up with that, guys?
14 Likes
I have to check out what this CrowdStrike actually is...
offtopic:
Isn't that because we accepted strange woke strange things as normal? 
1 Like