Red-black trees for BPF programs
Red-black trees for BPF programs
Posted Feb 28, 2023 1:15 UTC (Tue) by NYKevin (subscriber, #129325)In reply to: Red-black trees for BPF programs by mathstuf
Parent article: Red-black trees for BPF programs
Looking through the patches and comments, it looks like it's doing some kind of funky non-owned ref logic to protect against that, but I don't speak kernel-C well enough to understand exactly how it all works. In particular, I couldn't figure out where it prevents you from passing a different less (but I didn't look very hard, frankly).
Posted Feb 28, 2023 2:21 UTC (Tue)
by mathstuf (subscriber, #69389)
[Link]
Red-black trees for BPF programs
The non-owned ref logic can help prevent mutating the nodes in the `less` function (I presume/hope that `const_cast` spell in the C way is verbotten as well?). But are there any BPF rules that forbid referencing some BPF map, mutable global variable, or volatile (to BPF) kernel data binding?