On 1 May 2020, at 21:16, Andrii Nakryiko wrote: > On Fri, May 1, 2020 at 2:56 AM Eelco Chaudron <echaudro@xxxxxxxxxx> wrote: <SNIP> >> >> Let me know, and I sent out a v2. > > Yes, that's the split I had in mind, but I'd move > bpf_object__probe_loading() call directly into bpf_object__load() to > be the first thing to check. probe_caps() should still be non-failing > if any feature is missing. Does it make sense? I think I got it :) I’ll send out a v2 soon… //Eelco
