POC: Kernel notifier framework user space notifications - upstream Kernel changes

[maryamtahhan]

Implementing the POC made necessary the following kernel changes:

1. Allowing kernel/module access to BPF maps by path similarly to how it has been allowed to programs.
2. Exposing the following ringbuf functionality previously available only to eBPF callers via the helper interface (helper function names are being used for clarity, kernel side additions are prefixed with _ for POC purposes):
   a. bpf_ringbuf_reserve()
   b. bpf_ringbuf_commit()
   c. bpf_ringbuf_discard()
   d. bpf_ringbuf_output() - this can be implemented using a, b and c, so not essential. Added in order to not duplicate code.
   e. bpf_ringbuf_query() - not used in the POC
   f. bpf_user_ringbuf_drain() - should probably be dropped in future versions as it is very eBPF helper specific
3. Adding the following new kernel functionality
   a. bpf_ringbuf_fetch_next() - fetch next record. The functionality is similar to bpf_user_ringbuf_drain, but without the callback semantics
   b. bpf_ringbuf_has_data() - may be dropped in the future - intended for wait conditions
   c. void ringbuf_wait_for_data() - not functional yet due to lack of wake up triggers

The goal of this task is to create a patchset for the changes to the kernel and submit them to the BPF mailing list for revew