netdev - Re: "resolve_btfids: unresolved" warnings while building v5.16-rc1

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 15 Nov 2021 22:25:39 -0800
From:   Andrii Nakryiko <andrii.nakryiko@...il.com>
To:     Kumar Kartikeya Dwivedi <memxor@...il.com>,
        Jiri Olsa <jolsa@...nel.org>
Cc:     Pavel Skripkin <paskripkin@...il.com>,
        Alexei Starovoitov <ast@...nel.org>,
        Jakub Kicinski <kuba@...nel.org>,
        "David S. Miller" <davem@...emloft.net>,
        Networking <netdev@...r.kernel.org>, bpf <bpf@...r.kernel.org>,
        Andrii Nakryiko <andrii@...nel.org>
Subject: Re: "resolve_btfids: unresolved" warnings while building v5.16-rc1

On Mon, Nov 15, 2021 at 9:27 AM Kumar Kartikeya Dwivedi
<memxor@...il.com> wrote:
>
> On Mon, Nov 15, 2021 at 07:47:35PM IST, Kumar Kartikeya Dwivedi wrote:
> > On Mon, Nov 15, 2021 at 07:04:51PM IST, Pavel Skripkin wrote:
> > > Hi, net/bpf developers!
> > >
> > > While building newest kernel for fuzzing I met following warnings:
> > >
> > > ```
> > >   BTFIDS  vmlinux
> > > WARN: resolve_btfids: unresolved symbol tcp_dctcp_kfunc_ids
> > > WARN: resolve_btfids: unresolved symbol tcp_cubic_kfunc_ids
> > > WARN: resolve_btfids: unresolved symbol tcp_bbr_kfunc_ids
> > >   SORTTAB vmlinux
> > >
> > > ```
>
> +Cc Andrii

+Cc Jiri ;)

>
> So the reason should be CONFIG_DYNAMIC_FTRACE=n, when that is turned off,
> all these three BTF sets should be empty. Earlier they were all part of the
> set in bpf_tcp_ca.c, which would never be empty, so there was no warning.
>
> I guess we can demote that warning to debug, but not sure, since it isn't

It's a real warning, since when hiding a warning is the right thing to do?

Jiri, does resolve_btfids emit this warning when we get an empty set?
I'd expect not (empty set is totally reasonable in some cases), but
who knows. Can you please help understand this?

> limited to BTF sets, but also other symbols (e.g. kernel functions referenced in
> .BTF_ids).
>
> The other option is to add a dummy function in the set so that set->cnt != 0.
>
> > >
> > > I haven't seen such warnings before and have no idea are they important or
> > > not. Config is attached.
> > >
> > > My host is openSUSE Tumbleweed with gcc (SUSE Linux) 10.3.1 20210707
> > > [revision 048117e16c77f82598fca9af585500572d46ad73] if it's important :)
> > >
> > >
> >
> > I'll take a look later today.
> >
> > >
> > > With regards,
> > > Pavel Skripkin
> >
> > --
> > Kartikeya
>
> --
> Kartikeya