[PATCH bpf-next 0/4] bpf: tailcall: Eliminate max_entries and bpf_func access at runtime

[Alexei Starovoitov]

On Wed, Jan 14, 2026 at 1:00 PM Jiri Olsa <olsajiri at gmail.com> wrote:
>
> >
> > > fyi I briefly discussed that with Andrii indicating that it might not
> > > be worth the effort at this stage.
> >
> > depending on complexity of course.
>
> for my tests I just had to allow BPF_MAP_TYPE_PROG_ARRAY map
> for sleepable programs
>
> jirka
>
>
> ---
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index faa1ecc1fe9d..1f6fc74c7ea1 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -20969,6 +20969,7 @@ static int check_map_prog_compatibility(struct bpf_verifier_env *env,
>                 case BPF_MAP_TYPE_STACK:
>                 case BPF_MAP_TYPE_ARENA:
>                 case BPF_MAP_TYPE_INSN_ARRAY:
> +               case BPF_MAP_TYPE_PROG_ARRAY:
>                         break;
>                 default:
>                         verbose(env,

Think it through, add selftests, ship it.
On the surface the easy part is to make
__bpf_prog_map_compatible() reject sleepable/non-sleepable combo.
Maybe there are other things.