back
9 comments
Not a kernel veteran, but I do send patches and reviews occasionally and as mentioned in the article, Sashiko is a big help. It can detect very obscure race conditions, stack leaks and other bugs that could cause a kernel panic. It's also really good at analyzing subsystem-specific nuances (in the IIO subsystem for example, it can get chip parameters from a datasheet and actually check whether the code reflects it correctly, e. g. with timing).
What percentage of those would be non-issues if the kernel used a different language?
Hard to say, different language - different problems. Nevertheless Sashiko gets the big picture and its main concerns are race conditions, bad register mapping etc., issues that concern C code style are minimal (at least in the patches I write haha)
URL changed but article from March OP;

https://news.ycombinator.com/item?id=47547849

If you get reports on what looks like backdoor and do a gitblame, does the enail that returns get traced to other projects? Like is there a pattern detection to the authors that also allows for detection of current malicous contributors?
Needs a [26th March 2026] tag. That's, like, 3 months ago. Can anything in it still be relevant?
Not everyone is a AI-firehose-of-news enjoyer.
If it was a previous year one might expect (20xx). Anything posted in the same year does not typically have the date in the submission.
Sentiment was trending negative, HN needed some positive AI news, even if a bit old.