This is how security issues get introduced.

Just for fun, I looked at how Apache handles this.

It turns out that fork and exec aren't the only way to create processes on new threads. And, if you care about performace, you will use other system calls in addition to fork and exec. This is what Apache mostly does and calling exec after calling fork is rare. I didn't read the code. AI tells me this, but it sounds reasonamble.

However, regardless of the foregoing, introduced reading from some "cache" (read place I put my attack vector) to improve performance on functions that people who care about performance don't use, sounds like the ideal code backwater that a hacker would exploit.

Unless you actually are a hacker and have plans to use this potential vulnerability later.

As far as Kernal development goes, my perspective, and something I think Kernel developers need to hear is , "Just don't do something, sit there!"

On Sun, Jun 07, 2026 at 12:18:59AM -0000, Kevin B wrote:

Interesting

https://lwn.net/SubscriberLink/1076018/16f01bbbb8e0d1f0/ _______________________________________________ Lug-nuts mailing list -- lug-nuts@bigbrie.com To unsubscribe send an email to lug-nuts-leave@bigbrie.com