Add cgroup bcc flag for compilation

[tewaro]

Summary: This is to ensure that the reversions in #1738 are not necessary for #1638. It enables a flag for bcc compilation for detecting that the bpf_get_current_cgroup_id exists within that kernel version. That flag is GET_CGROUP_ID_ENABLED. And an example of it's usage is in src/stirling/bpf_tools/testdata/get_current_cgroup_id.c.

Type of change: /kind feature

Test Plan: Automated test added to ensure that this feature is enabled and works.

Signed-off-as: AdityaAtulTewari adityaatewari@gmail.com

[ddelnano]

@AdityaAtulTewari sorry for the late reply here. The solution I envisioned was what I've implemented in 3bab3f0. That change isn't production ready, but shows how the pl_bpf_cc_resource bazel rule can be enhanced to support delaying some c preprocessing logic. I've verified that this prevents the bcc compilation issues and that the resulting socket_trace.c file still contains the #ifdef macros.

ddelnano@pixie-dev:~/code/pixie (ddelnano/add-cgroup-short-proc-fix-behind-feature-toggle) $ bazel build  src/stirling/source_connectors/socket_tracer\/bcc_bpf:socket_trace_bpf_preprocess
Starting local Bazel server and connecting to it...
INFO: Analyzed target //src/stirling/source_connectors/socket_tracer/bcc_bpf:socket_trace_bpf_preprocess (87 packages loaded, 1828 targets configured).
INFO: Found 1 target...
INFO: Deleting stale sandbox base /home/ddelnano/.cache/bazel/_bazel_ddelnano/b2e8b093089e374aa7d44dfed4f9953b/sandbox
Target //src/stirling/source_connectors/socket_tracer/bcc_bpf:socket_trace_bpf_preprocess up-to-date:
  bazel-bin/src/stirling/source_connectors/socket_tracer/bcc_bpf/socket_trace_bpf_preprocess.c
INFO: Elapsed time: 13.225s, Critical Path: 0.18s
INFO: 2 processes: 1 internal, 1 linux-sandbox.
INFO: Build completed successfully, 2 total actions


ddelnano@pixie-dev:~/code/pixie (ddelnano/add-cgroup-short-proc-fix-behind-feature-toggle) $ head -n 11 bazel-bin/src/stirling/source_connectors/socket_tracer/bcc_bpf/socket_trace_bpf_preprocess.c
#ifdef GET_CGROUP_ID_ENABLED
uint64_t pl_bpf_get_current_cgroup_id() {
    return bpf_get_current_cgroup_id();
}
#else

uint64_t pl_bpf_get_current_cgroup_id() {
    // TODO(ddelnano): UINT64_MAX doesn't work in BPF. Find another way to represent this.
    return (uint64_t)18446744073709551615;
}
#endif

In order to move forward with this, I think we will need some feedback on this approach from @vihangm and @JamesMBartlett. If this is something we are comfortable with, then we can shape up what I've prototyped.