Re: [PATCH 05/11] bpf: Remove btf_id helpers resolving

From: Jiri Olsa <jolsa@redhat.com>
To: Andrii Nakryiko <andrii.nakryiko@gmail.com>
Cc: Jiri Olsa <jolsa@kernel.org>, Alexei Starovoitov <ast@kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Networking <netdev@vger.kernel.org>, bpf <bpf@vger.kernel.org>,
	Song Liu <songliubraving@fb.com>, Yonghong Song <yhs@fb.com>,
	Martin KaFai Lau <kafai@fb.com>, David Miller <davem@redhat.com>,
	John Fastabend <john.fastabend@gmail.com>,
	Wenbo Zhang <ethercflow@gmail.com>,
	KP Singh <kpsingh@chromium.org>, Andrii Nakryiko <andriin@fb.com>,
	Brendan Gregg <bgregg@netflix.com>,
	Florent Revest <revest@chromium.org>,
	Al Viro <viro@zeniv.linux.org.uk>
Subject: Re: [PATCH 05/11] bpf: Remove btf_id helpers resolving
Date: Fri, 19 Jun 2020 15:18:00 +0200	[thread overview]
Message-ID: <20200619131800.GF2465907@krava> (raw)
In-Reply-To: <CAEf4BzYw0VciF-7CS164Nk8LLnZ4odtdYQyX1MS4eWDN5WbcSg@mail.gmail.com>

On Thu, Jun 18, 2020 at 06:10:29PM -0700, Andrii Nakryiko wrote:
> On Tue, Jun 16, 2020 at 3:06 AM Jiri Olsa <jolsa@kernel.org> wrote:
> >
> > Now when we moved the helpers btf_id into .BTF_ids section,
> > we can remove the code that resolve those IDs in runtime.
> >
> > Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> > ---
> 
> Nice! :)
> 
> BTW, have you looked at bpf_ctx_convert stuff? Would we be able to
> replace it with your btfids thing as well?

good, another usage ;-) I'll check

> 
> 
> >  kernel/bpf/btf.c | 88 +++---------------------------------------------
> >  1 file changed, 4 insertions(+), 84 deletions(-)
> >
> > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
> > index 58c9af1d4808..aea7b2cc8d26 100644
> > --- a/kernel/bpf/btf.c
> > +++ b/kernel/bpf/btf.c
> > @@ -4049,96 +4049,16 @@ int btf_struct_access(struct bpf_verifier_log *log,
> >         return -EINVAL;
> >  }
> >
> 
> [...]
> 
> >  int btf_resolve_helper_id(struct bpf_verifier_log *log,
> >                           const struct bpf_func_proto *fn, int arg)
> >  {
> > -       int *btf_id = &fn->btf_id[arg];
> > -       int ret;
> > -
> >         if (fn->arg_type[arg] != ARG_PTR_TO_BTF_ID)
> >                 return -EINVAL;
> >
> > -       ret = READ_ONCE(*btf_id);
> > -       if (ret)
> > -               return ret;
> > -       /* ok to race the search. The result is the same */
> > -       ret = __btf_resolve_helper_id(log, fn->func, arg);
> > -       if (!ret) {
> > -               /* Function argument cannot be type 'void' */
> > -               bpf_log(log, "BTF resolution bug\n");
> > -               return -EFAULT;
> > -       }
> > -       WRITE_ONCE(*btf_id, ret);
> > -       return ret;
> > +       if (WARN_ON_ONCE(!fn->btf_id))
> > +               return -EINVAL;
> > +
> > +       return fn->btf_id[arg];
> 
> It probably would be a good idea to add some sanity checking here,
> making sure that btf_id is >0 (void is never a right type) and <=
> nr_types in vmlinux_btf?

yep, will add it ;-)

jirka