[PATCH bpf-next] libbpf: Fix struct and union zero detection in BTF dumper

[PATCH bpf-next] libbpf: Fix struct and union zero detection in BTF dumper

[Luis Vieira]

btf_dump_type_data_check_zero() uses -ENODATA to indicate that a value
is zero. The array path handles this sentinel correctly, but the struct
and union path compares the return value against positive ENODATA.

As a result, the member loop returns after examining its first member.
A nested struct or union whose first member is zero can therefore be
treated as entirely zero even when a later member is non-zero.

Compare against -ENODATA so zero members are skipped while looking for
a non-zero member. Add a regression test with a nested struct whose
first member is zero and second member is non-zero.

Signed-off-by: Luis Vieira <luisflavieira@gmail.com>
---
 tools/lib/bpf/btf_dump.c                          | 2 +-
 tools/testing/selftests/bpf/prog_tests/btf_dump.c | 5 +++++
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/tools/lib/bpf/btf_dump.c b/tools/lib/bpf/btf_dump.c
index fa995c02a170..dbf9f17e4b37 100644
--- a/tools/lib/bpf/btf_dump.c
+++ b/tools/lib/bpf/btf_dump.c
@@ -2474,7 +2474,7 @@ static int btf_dump_type_data_check_zero(struct btf_dump *d,
 			bit_sz = btf_member_bitfield_size(t, i);
 			err = btf_dump_type_data_check_zero(d, mtype, m->type, data + moffset / 8,
 							    moffset % 8, bit_sz);
-			if (err != ENODATA)
+			if (err != -ENODATA)
 				return err;
 		}
 		return -ENODATA;
diff --git a/tools/testing/selftests/bpf/prog_tests/btf_dump.c b/tools/testing/selftests/bpf/prog_tests/btf_dump.c
index fe04a955d46c..dd1241fcdb01 100644
--- a/tools/testing/selftests/bpf/prog_tests/btf_dump.c
+++ b/tools/testing/selftests/bpf/prog_tests/btf_dump.c
@@ -886,6 +886,11 @@ static void test_btf_dump_struct_data(struct btf *btf, struct btf_dump *d,
 			   "(union bpf_iter_link_info){.map = (struct){.map_fd = (__u32)1,},.cgroup = (struct){.order = (enum bpf_cgroup_iter_order)BPF_CGROUP_ITER_SELF_ONLY,.cgroup_fd = (__u32)1,},.task = (struct){.tid = (__u32)1,.pid = (__u32)1,},}",
 			   { .cgroup = { .order = 1, .cgroup_fd = 1, }});
 
+	/* nested structs with zero first member and non-zero second member */
+	TEST_BTF_DUMP_DATA(btf, d, "union", str, union bpf_iter_link_info, BTF_F_COMPACT,
+			   "(union bpf_iter_link_info){.cgroup = (struct){.cgroup_fd = (__u32)1,},.task = (struct){.pid = (__u32)1,},}",
+			   { .cgroup = { .cgroup_fd = 1 }});
+
 	/* struct skb with nested structs/unions; because type output is so
 	 * complex, we don't do a string comparison, just verify we return
 	 * the type size as the amount of data displayed.

---
base-commit: ea9358e1270ab2c3ba6f36bd9bdda68617665516
change-id: 20260928-libbpf-btf-enodata-fix-ac6381645ab8

Best regards,
--  
Luis Vieira <luisflavieira@gmail.com>

Re: [PATCH bpf-next] libbpf: Fix struct and union zero detection in BTF dumper

[Andrii Nakryiko]

On Mon, Sep 28, 2026 at 4:47 AM Luis Vieira <luisflavieira@gmail.com> wrote:
>
> btf_dump_type_data_check_zero() uses -ENODATA to indicate that a value
> is zero. The array path handles this sentinel correctly, but the struct
> and union path compares the return value against positive ENODATA.
>
> As a result, the member loop returns after examining its first member.
> A nested struct or union whose first member is zero can therefore be
> treated as entirely zero even when a later member is non-zero.
>
> Compare against -ENODATA so zero members are skipped while looking for
> a non-zero member. Add a regression test with a nested struct whose
> first member is zero and second member is non-zero.
>
> Signed-off-by: Luis Vieira <luisflavieira@gmail.com>
> ---
>  tools/lib/bpf/btf_dump.c                          | 2 +-
>  tools/testing/selftests/bpf/prog_tests/btf_dump.c | 5 +++++
>  2 files changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/tools/lib/bpf/btf_dump.c b/tools/lib/bpf/btf_dump.c
> index fa995c02a170..dbf9f17e4b37 100644
> --- a/tools/lib/bpf/btf_dump.c
> +++ b/tools/lib/bpf/btf_dump.c
> @@ -2474,7 +2474,7 @@ static int btf_dump_type_data_check_zero(struct btf_dump *d,
>                         bit_sz = btf_member_bitfield_size(t, i);
>                         err = btf_dump_type_data_check_zero(d, mtype, m->type, data + moffset / 8,
>                                                             moffset % 8, bit_sz);
> -                       if (err != ENODATA)
> +                       if (err != -ENODATA)
>                                 return err;
>                 }
>                 return -ENODATA;
> diff --git a/tools/testing/selftests/bpf/prog_tests/btf_dump.c b/tools/testing/selftests/bpf/prog_tests/btf_dump.c
> index fe04a955d46c..dd1241fcdb01 100644
> --- a/tools/testing/selftests/bpf/prog_tests/btf_dump.c
> +++ b/tools/testing/selftests/bpf/prog_tests/btf_dump.c
> @@ -886,6 +886,11 @@ static void test_btf_dump_struct_data(struct btf *btf, struct btf_dump *d,
>                            "(union bpf_iter_link_info){.map = (struct){.map_fd = (__u32)1,},.cgroup = (struct){.order = (enum bpf_cgroup_iter_order)BPF_CGROUP_ITER_SELF_ONLY,.cgroup_fd = (__u32)1,},.task = (struct){.tid = (__u32)1,.pid = (__u32)1,},}",
>                            { .cgroup = { .order = 1, .cgroup_fd = 1, }});
>
> +       /* nested structs with zero first member and non-zero second member */
> +       TEST_BTF_DUMP_DATA(btf, d, "union", str, union bpf_iter_link_info, BTF_F_COMPACT,
> +                          "(union bpf_iter_link_info){.cgroup = (struct){.cgroup_fd = (__u32)1,},.task = (struct){.pid = (__u32)1,},}",
> +                          { .cgroup = { .cgroup_fd = 1 }});
> +

can you please split selftest into a separate patch and send as v2?
the fix makes sense, that missing - is definitely a typo we missed

pw-bot: cr

>         /* struct skb with nested structs/unions; because type output is so
>          * complex, we don't do a string comparison, just verify we return
>          * the type size as the amount of data displayed.
>
> ---
> base-commit: ea9358e1270ab2c3ba6f36bd9bdda68617665516
> change-id: 20260928-libbpf-btf-enodata-fix-ac6381645ab8
>
> Best regards,
> --
> Luis Vieira <luisflavieira@gmail.com>
>