linux-kernel - Re: [PATCH bpf-next] selftests/bpf: fix compilation failure when CONFIG_NET

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]

Message-ID: <CAEf4BzbgeCo09sfrQVgBHJJ-=uZEEm287xXkjoLMrUkcLN6VMQ@mail.gmail.com>
Date: Fri, 19 Jul 2024 11:44:34 -0700
From: Andrii Nakryiko <andrii.nakryiko@...il.com>
To: Alexei Starovoitov <alexei.starovoitov@...il.com>
Cc: Daniel Borkmann <daniel@...earbox.net>, Artem Savkov <asavkov@...hat.com>, 
	Alexei Starovoitov <ast@...nel.org>, Andrii Nakryiko <andrii@...nel.org>, bpf <bpf@...r.kernel.org>, 
	Network Development <netdev@...r.kernel.org>, LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH bpf-next] selftests/bpf: fix compilation failure when CONFIG_NET_FOU!=y

On Fri, Jul 19, 2024 at 10:09 AM Alexei Starovoitov
<alexei.starovoitov@...il.com> wrote:
>
> On Fri, Jul 19, 2024 at 8:45 AM Daniel Borkmann <daniel@...earbox.net> wrote:
> >
> > Hi Artem,
> >
> > On 7/18/24 4:31 PM, Artem Savkov wrote:
> > > Without CONFIG_NET_FOU bpf selftests are unable to build because of
> > > missing definitions. Add ___local versions of struct bpf_fou_encap and
> > > enum bpf_fou_encap_type to fix the issue.
> > >
> > > Signed-off-by: Artem Savkov <asavkov@...hat.com>
> >
> > This breaks BPF CI, ptal:
> >
> > https://github.com/kernel-patches/bpf/actions/runs/9999691294/job/27641198557
> >
> >    [...]
> >      CLNG-BPF [test_maps] btf__core_reloc_existence___wrong_field_defs.bpf.o
> >      CLNG-BPF [test_maps] verifier_bswap.bpf.o
> >      CLNG-BPF [test_maps] test_core_reloc_existence.bpf.o
> >      CLNG-BPF [test_maps] test_global_func8.bpf.o
> >      CLNG-BPF [test_maps] verifier_bitfield_write.bpf.o
> >      CLNG-BPF [test_maps] local_storage_bench.bpf.o
> >      CLNG-BPF [test_maps] verifier_runtime_jit.bpf.o
> >      CLNG-BPF [test_maps] test_pkt_access.bpf.o
> >    progs/test_tunnel_kern.c:39:5: error: conflicting types for 'bpf_skb_set_fou_encap'
> >       39 | int bpf_skb_set_fou_encap(struct __sk_buff *skb_ctx,
> >          |     ^
> >    /tmp/work/bpf/bpf/tools/testing/selftests/bpf/tools/include/vmlinux.h:107714:12: note: previous declaration is here
> >     107714 | extern int bpf_skb_set_fou_encap(struct __sk_buff *skb_ctx, struct bpf_fou_encap *encap, int type) __weak __ksym;
> >            |            ^
> >    progs/test_tunnel_kern.c:41:5: error: conflicting types for 'bpf_skb_get_fou_encap'
> >       41 | int bpf_skb_get_fou_encap(struct __sk_buff *skb_ctx,
> >          |     ^
> >    /tmp/work/bpf/bpf/tools/testing/selftests/bpf/tools/include/vmlinux.h:107715:12: note: previous declaration is here
> >     107715 | extern int bpf_skb_get_fou_encap(struct __sk_buff *skb_ctx, struct bpf_fou_encap *encap) __weak __ksym;
> >            |            ^
> >      CLNG-BPF [test_maps] verifier_typedef.bpf.o
> >      CLNG-BPF [test_maps] user_ringbuf_fail.bpf.o
> >      CLNG-BPF [test_maps] verifier_map_in_map.bpf.o
> >    progs/test_tunnel_kern.c:782:35: error: incompatible pointer types passing 'struct bpf_fou_encap___local *' to parameter of type 'struct bpf_fou_encap *' [-Werror,-Wincompatible-pointer-types]
> >      782 |         ret = bpf_skb_set_fou_encap(skb, &encap, FOU_BPF_ENCAP_GUE___local);
> >          |                                          ^~~~~~
> >    /tmp/work/bpf/bpf/tools/testing/selftests/bpf/tools/include/vmlinux.h:107714:83: note: passing argument to parameter 'encap' here
> >     107714 | extern int bpf_skb_set_fou_encap(struct __sk_buff *skb_ctx, struct bpf_fou_encap *encap, int type) __weak __ksym;
>
> It's a good idea to introduce struct bpf_fou_encap___local
> for !FOU builds, but kfunc signature needs to stay and
> __local variable needs to be type casted to (struct bpf_fou_encap *)
> when calling kfunc.

Given we specify

CONFIG_NET_FOU=y (not =m)

in selftests/bpf/config, do we really need to work around this? I bet
we have a bunch of other missing types if we don't set all the
settings as required by selftests/bpf/config.