> So, first of all, I'm not sure whether the BPF verifier being so > strict about BTF names is necessary, but I haven't thought a lot about > that. Just pointing out I myself wouldn't object lifting any kind of > restriction there, in principle. > > But then for the Zig naming thing. Doesn't Zig have some sort of > mangling schema, like C++ or Rust do? Would it make sense to store > mangled names in BTF then? I'm afraid not. Zig is still in its infancy. So I finally decided to sort of work around this issue on the zbpf side by introducing a BTF sanitizer [1]. Thanks Andrii, Eduard and Alan for the help along this thread. [1] https://github.com/tw4452852/zbpf/commit/866d83cdf57316f63f6c20ee5fbc0880575a0213
