Re: [PATCH bpf-next v2] selftests/bpf: Fix build errors if CONFIG_NF_CONNTRACK=m

[Andrii Nakryiko]

On Sat, Jan 28, 2023 at 10:25 PM Tiezhu Yang <yangtiezhu@xxxxxxxxxxx> wrote:
>
>
>
> On 01/24/2023 07:17 AM, Andrii Nakryiko wrote:
> > On Tue, Jan 17, 2023 at 11:57 PM Tiezhu Yang <yangtiezhu@xxxxxxxxxxx> wrote:
> >>
> >> If CONFIG_NF_CONNTRACK=m, there are no definitions of NF_NAT_MANIP_SRC
> >> and NF_NAT_MANIP_DST in vmlinux.h, build test_bpf_nf.c failed.
> >>
> >> $ make -C tools/testing/selftests/bpf/
> >>
> >>   CLNG-BPF [test_maps] test_bpf_nf.bpf.o
> >> progs/test_bpf_nf.c:160:42: error: use of undeclared identifier 'NF_NAT_MANIP_SRC'
> >>                 bpf_ct_set_nat_info(ct, &saddr, sport, NF_NAT_MANIP_SRC);
> >>                                                        ^
> >> progs/test_bpf_nf.c:163:42: error: use of undeclared identifier 'NF_NAT_MANIP_DST'
> >>                 bpf_ct_set_nat_info(ct, &daddr, dport, NF_NAT_MANIP_DST);
> >>                                                        ^
> >> 2 errors generated.
> >>
> >> Copy the definitions in include/net/netfilter/nf_nat.h to test_bpf_nf.c,
> >> in order to avoid redefinitions if CONFIG_NF_CONNTRACK=y, rename them with
> >> ___local suffix. This is similar with commit 1058b6a78db2 ("selftests/bpf:
> >> Do not fail build if CONFIG_NF_CONNTRACK=m/n").
> >>
> >> Fixes: b06b45e82b59 ("selftests/bpf: add tests for bpf_ct_set_nat_info kfunc")
> >> Signed-off-by: Tiezhu Yang <yangtiezhu@xxxxxxxxxxx>
> >> ---
> >>  tools/testing/selftests/bpf/progs/test_bpf_nf.c | 11 ++++++++---
> >>  1 file changed, 8 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/tools/testing/selftests/bpf/progs/test_bpf_nf.c b/tools/testing/selftests/bpf/progs/test_bpf_nf.c
> >> index 227e85e..9fc603c 100644
> >> --- a/tools/testing/selftests/bpf/progs/test_bpf_nf.c
> >> +++ b/tools/testing/selftests/bpf/progs/test_bpf_nf.c
> >> @@ -34,6 +34,11 @@ __be16 dport = 0;
> >>  int test_exist_lookup = -ENOENT;
> >>  u32 test_exist_lookup_mark = 0;
> >>
> >> +enum nf_nat_manip_type___local {
> >> +       NF_NAT_MANIP_SRC___local,
> >> +       NF_NAT_MANIP_DST___local
> >> +};
> >> +
> >>  struct nf_conn;
> >>
> >>  struct bpf_ct_opts___local {
> >> @@ -58,7 +63,7 @@ int bpf_ct_change_timeout(struct nf_conn *, u32) __ksym;
> >>  int bpf_ct_set_status(struct nf_conn *, u32) __ksym;
> >>  int bpf_ct_change_status(struct nf_conn *, u32) __ksym;
> >>  int bpf_ct_set_nat_info(struct nf_conn *, union nf_inet_addr *,
> >> -                       int port, enum nf_nat_manip_type) __ksym;
> >> +                       int port, enum nf_nat_manip_type___local) __ksym;
> >>
> >>  static __always_inline void
> >>  nf_ct_test(struct nf_conn *(*lookup_fn)(void *, struct bpf_sock_tuple *, u32,
> >> @@ -157,10 +162,10 @@ nf_ct_test(struct nf_conn *(*lookup_fn)(void *, struct bpf_sock_tuple *, u32,
> >>
> >>                 /* snat */
> >>                 saddr.ip = bpf_get_prandom_u32();
> >> -               bpf_ct_set_nat_info(ct, &saddr, sport, NF_NAT_MANIP_SRC);
> >> +               bpf_ct_set_nat_info(ct, &saddr, sport, NF_NAT_MANIP_SRC___local);
> >>                 /* dnat */
> >>                 daddr.ip = bpf_get_prandom_u32();
> >> -               bpf_ct_set_nat_info(ct, &daddr, dport, NF_NAT_MANIP_DST);
> >> +               bpf_ct_set_nat_info(ct, &daddr, dport, NF_NAT_MANIP_DST___local);
> >>
> >
> > it would be a bit more reliable if you used `bpf_core_enum_value(enum
> > nf_nat_manip_type___local, NF_NAT_MANIP_DST___local)`. That would make
> > libbpf substitute correct absolute value, if actual enum
> > nf_nat_manip_type in kernel ever changed. Please consider a follow up
> > patch for this.
>
> Sorry for the late reply, I tested the code as your suggestion, but it
> failed.
>
> failed to resolve CO-RE relocation <enumval_value> [101] enum
> nf_nat_manip_type___local::NF_NAT_MANIP_SRC___local = 0


Was nf_conntrack kernel module loaded at the time when you ran the
test? If yes, what's the output of

bpftool btf dump file /sys/kernel/btf/nf_conntrack | grep nf_nat_manip_type

?

>
> Is it necessary to send a follow patch now? Thank you.
>
> Here are the test results.
>
> (1) bpf_nf passed with the current code if CONFIG_NF_CONNTRACK=m:
>
> [root@fedora bpf]# ./test_progs -a bpf_nf
> #16/1    bpf_nf/xdp-ct:OK
> #16/2    bpf_nf/tc-bpf-ct:OK
> #16/3    bpf_nf/alloc_release:OK
> #16/4    bpf_nf/insert_insert:OK
> #16/5    bpf_nf/lookup_insert:OK
> #16/6    bpf_nf/set_timeout_after_insert:OK
> #16/7    bpf_nf/set_status_after_insert:OK
> #16/8    bpf_nf/change_timeout_after_alloc:OK
> #16/9    bpf_nf/change_status_after_alloc:OK
> #16/10   bpf_nf/write_not_allowlisted_field:OK
> #16      bpf_nf:OK
> Summary: 1/10 PASSED, 0 SKIPPED, 0 FAILED
>
> (2) bpf_nf failed with the following changes if CONFIG_NF_CONNTRACK=m:
>
> [yangtiezhu@fedora bpf.git]$ git diff
> diff --git a/tools/testing/selftests/bpf/progs/test_bpf_nf.c
> b/tools/testing/selftests/bpf/progs/test_bpf_nf.c
> index 9fc603c9d673..f56ba60ab809 100644
> --- a/tools/testing/selftests/bpf/progs/test_bpf_nf.c
> +++ b/tools/testing/selftests/bpf/progs/test_bpf_nf.c
> @@ -2,6 +2,7 @@
>   #include <vmlinux.h>
>   #include <bpf/bpf_helpers.h>
>   #include <bpf/bpf_endian.h>
> +#include <bpf/bpf_core_read.h>
>
>   #define EAFNOSUPPORT 97
>   #define EPROTO 71
> @@ -162,10 +163,14 @@ nf_ct_test(struct nf_conn *(*lookup_fn)(void *,
> struct bpf_sock_tuple *, u32,
>
>                  /* snat */
>                  saddr.ip = bpf_get_prandom_u32();
> -               bpf_ct_set_nat_info(ct, &saddr, sport,
> NF_NAT_MANIP_SRC___local);
> +               bpf_ct_set_nat_info(ct, &saddr, sport,
> +                                   bpf_core_enum_value(enum
> nf_nat_manip_type___local,
> +
> NF_NAT_MANIP_SRC___local));
>                  /* dnat */
>                  daddr.ip = bpf_get_prandom_u32();
> -               bpf_ct_set_nat_info(ct, &daddr, dport,
> NF_NAT_MANIP_DST___local);
> +               bpf_ct_set_nat_info(ct, &daddr, dport,
> +                                   bpf_core_enum_value(enum
> nf_nat_manip_type___local,
> +
> NF_NAT_MANIP_DST___local));
>
>                  ct_ins = bpf_ct_insert_entry(ct);
>                  if (ct_ins) {
>
> [root@fedora bpf]# ./test_progs -a bpf_nf
> ...
> All error logs:
> libbpf: prog 'nf_xdp_ct_test': BPF program load failed: Invalid argument
> libbpf: prog 'nf_xdp_ct_test': -- BEGIN PROG LOAD LOG --
> ...
> libbpf: failed to load object 'test_bpf_nf'
> libbpf: failed to load BPF skeleton 'test_bpf_nf': -22
> test_bpf_nf_ct:FAIL:test_bpf_nf__open_and_load unexpected error: -22
> #16/1    bpf_nf/xdp-ct:FAIL
> libbpf: prog 'nf_xdp_ct_test': BPF program load failed: Invalid argument
> libbpf: prog 'nf_xdp_ct_test': -- BEGIN PROG LOAD LOG --
> ...
> 217: (bf) r1 = r7                     ;
> R1_w=ptr_nf_conn___init(ref_obj_id=18,off=0,imm=0)
> R7=ptr_nf_conn___init(ref_obj_id=18,off=0,imm=0) refs=18
> 218: <invalid CO-RE relocation>
> failed to resolve CO-RE relocation <enumval_value> [101] enum
> nf_nat_manip_type___local::NF_NAT_MANIP_SRC___local = 0
> processed 170 insns (limit 1000000) max_states_per_insn 0 total_states
> 12 peak_states 12 mark_read 2
> -- END PROG LOAD LOG --
> libbpf: prog 'nf_xdp_ct_test': failed to load: -22
> libbpf: failed to load object 'test_bpf_nf'
> libbpf: failed to load BPF skeleton 'test_bpf_nf': -22
> test_bpf_nf_ct:FAIL:test_bpf_nf__open_and_load unexpected error: -22
> #16/2    bpf_nf/tc-bpf-ct:FAIL
> #16      bpf_nf:FAIL
> Summary: 0/8 PASSED, 0 SKIPPED, 1 FAILED
>
> Thanks,
> Tiezhu
>