> On Mon, Dec 11, 2023 at 01:30:33PM +0100, Jiri Olsa wrote:
> > + /*
> > + * The first prog in the chain is going to be attached to the target
> > + * fentry program, the second one to the previous in the chain.
> > + */
> > + if (i == 0) {
> > + prog = tracing_chain[0]->progs.recursive_attach;
> > + prev_fd = bpf_program__fd(target_skel->progs.test1);
> > + err = bpf_program__set_attach_target(prog, prev_fd, "test1");
> > + } else {
> > + prog = tracing_chain[i]->progs.recursive_attach;
> perhaps also the bpf_program__set_attach_target call does not need to be
> in the if path, I think it should work with NULL for attach_func_name as
> long as we provide attach_prog_fd
Just have tried that, but it didn't work -- when loading, libbpf tried
to search for XXX in BTF.
> I wonder now with just 2 skels the test might be easier to read
> without the loop, but I dont have strong opinion about that
Yeah, I was pondering that too. But eventually I came to the conclusion
that the difference for readability would be marginal, and went with
keeping the loop.
