> libbpf: Resolve kfunc variants by prototype TL;DR: A module BTF load error in find_kfunc_btf_id() may be returned as -EINVAL and mistaken for "no compatible prototype", so a weak kfunc in vmlinux could be treated as missing and strong ones get a misleading error. > diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c > --- a/tools/lib/bpf/libbpf.c > +++ b/tools/lib/bpf/libbpf.c [ ... ] > +/* > + * Search vmlinux and module BTFs. Return: > + * - a BTF ID for a unique match; > + * - -ESRCH if the family is absent; > + * - -EINVAL if no candidate has a compatible prototype; > + * - -ENOTUNIQ if more than one candidate has a compatible prototype; > + * - another negative error on failure. > + */ > +static int find_kfunc_btf_id(struct bpf_object *obj, const struct extern_desc *ext, > + struct btf **res_btf, struct module_btf **res_mod_btf) > +{ > + size_t name_len = bpf_core_essential_name_len(ext->name); > + int kfunc_id = 0, nr_cands = 0, err; > + size_t i; > + > + err = load_module_btfs(obj); > + if (err) > + return err; Can this return -EINVAL for a reason other than a prototype mismatch? load_module_btfs() calls btf_get_from_fd(), which goes through btf_new() and can return -EINVAL from btf_parse_hdr(), btf_parse_type_sec() or btf_sanity_check() when libbpf rejects a module's BTF. That error is returned here unchanged. [ ... ] > + if (kfunc_id == -EINVAL) { > if (ext->is_weak) > return 0; > + pr_warn("extern (func ksym) '%s': no compatible prototype in kernel or module BTFs\n", > + ext->name); > + return kfunc_id; > } bpf_object__resolve_ksym_func_btf_id() treats that -EINVAL as "no compatible prototype". A __weak kfunc is then left unresolved even if it exists in vmlinux, so bpf_ksym_exists() reports it as missing. A strong kfunc fails with the misleading "no compatible prototype" message. Also, load_module_btfs() has already set obj->btf_modules_loaded = true by then, so later kfuncs and the CO-RE lookups in bpf_core_find_cands() would continue with a partial module list instead of failing the load. Before this patch, a kfunc found in vmlinux never called load_module_btfs(). When the lookup did reach it, find_ksym_btf_id() returned the error as is, and only -ESRCH was treated as absence. Other errors from load_module_btfs() (-ENOMEM, errors from the BTF syscalls) also return here without a message naming the extern, because the generic "if (kfunc_id < 0) return kfunc_id;" path prints nothing. Would it make sense to call load_module_btfs() in the caller, or to keep its error apart from the -ESRCH/-EINVAL/-ENOTUNIQ results? --- AI reviewed your patch. Please fix the bug or email reply why it's not a bug. See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md CI run summary: https://github.com/kernel-patches/bpf/actions/runs/44845964556