On Fri, Jun 10, 2022 at 4:53 PM KP Singh <kpsingh@xxxxxxxxxx> wrote:
> > >>> +static const struct bpf_func_proto bpf_verify_signature_proto = {
> > >>> + .func = bpf_verify_signature,
> > >>> + .gpl_only = false,
> > >>> + .ret_type = RET_INTEGER,
> > >>> + .arg1_type = ARG_PTR_TO_MEM,
> > >>> + .arg2_type = ARG_CONST_SIZE_OR_ZERO,
> > >>
> > >> Can verify_pkcs7_signature() handle null/0 len for data* args?
> > >
> > > Shouldn't ARG_PTR_TO_MEM require valid memory? 0 len should
> > > not be a problem.
> >
> > check_helper_mem_access() has:
> >
> > /* Allow zero-byte read from NULL, regardless of pointer type */
> > if (zero_size_allowed && access_size == 0 &&
> > register_is_null(reg))
> > return 0;
>
> Daniel, makes a fair point here. Alexei, what do you think?
>
> I wonder if some "future" signature verification would need even more
> / different arguments so a unified bpf_verify_signature might get more
> complex / not easy to extend.
You mean a pkcs7 specific helper for now?
Makes sense.
