On Mon, Oct 28, 2019 at 1:09 PM Ilya Leoshkevich <iii@xxxxxxxxxxxxx> wrote:
>
> "ctx:file_pos sysctl:read read ok narrow" works on s390 by accident: it
> reads the wrong byte, which happens to have the expected value of 0.
> Improve the test by seeking to the 4th byte and expecting 4 instead of
> 0.
>
> This makes the latent problem apparent: the test attempts to read the
> first byte of bpf_sysctl.file_pos, assuming this is the least-significant
> byte, which is not the case on big-endian machines: a non-zero offset is
> needed.
>
> The point of the test is to verify narrow loads, so we cannot cheat our
> way out by simply using BPF_W. The existence of the test means that such
> loads have to be supported, most likely because llvm can generate them.
> Fix the test by adding a big-endian variant, which uses an offset to
> access the least-significant byte of bpf_sysctl.file_pos.
>
> This reveals the final problem: verifier rejects accesses to bpf_sysctl
> fields with offset > 0. Such accesses are already allowed for a wide
> range of structs: __sk_buff, bpf_sock_addr and sk_msg_md to name a few.
> Extend this support to bpf_sysctl by using bpf_ctx_range instead of
> offsetof when matching field offsets.
>
> Fixes: 7b146cebe30c ("bpf: Sysctl hook")
> Fixes: e1550bfe0de4 ("bpf: Add file_pos field to bpf_sysctl ctx")
> Fixes: 9a1027e52535 ("selftests/bpf: Test file_pos field in bpf_sysctl ctx")
> Signed-off-by: Ilya Leoshkevich <iii@xxxxxxxxxxxxx>
> ---
> kernel/bpf/cgroup.c | 4 ++--
> tools/testing/selftests/bpf/test_sysctl.c | 8 +++++++-
> 2 files changed, 9 insertions(+), 3 deletions(-)
>
> diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
> index ddd8addcdb5c..a3eaf08e7dd3 100644
> --- a/kernel/bpf/cgroup.c
> +++ b/kernel/bpf/cgroup.c
> @@ -1311,12 +1311,12 @@ static bool sysctl_is_valid_access(int off, int size, enum bpf_access_type type,
> return false;
>
> switch (off) {
> - case offsetof(struct bpf_sysctl, write):
> + case bpf_ctx_range(struct bpf_sysctl, write):
this will actually allow reads pas t write field (e.g., offset = 2, size = 4).
> if (type != BPF_READ)
> return false;
> bpf_ctx_record_field_size(info, size_default);
> return bpf_ctx_narrow_access_ok(off, size, size_default);
> - case offsetof(struct bpf_sysctl, file_pos):
> + case bpf_ctx_range(struct bpf_sysctl, file_pos)
this will allow read past context struct altogether. When we allow
ranges, we will have to adjust allowed read size.
> if (type == BPF_READ) {
> bpf_ctx_record_field_size(info, size_default);
> return bpf_ctx_narrow_access_ok(off, size, size_default);
> diff --git a/tools/testing/selftests/bpf/test_sysctl.c b/tools/testing/selftests/bpf/test_sysctl.c
> index a320e3844b17..7c6e5b173f33 100644
> --- a/tools/testing/selftests/bpf/test_sysctl.c
> +++ b/tools/testing/selftests/bpf/test_sysctl.c
> @@ -161,9 +161,14 @@ static struct sysctl_test tests[] = {
> .descr = "ctx:file_pos sysctl:read read ok narrow",
> .insns = {
> /* If (file_pos == X) */
> +#if __BYTE_ORDER == __LITTLE_ENDIAN
> BPF_LDX_MEM(BPF_B, BPF_REG_7, BPF_REG_1,
> offsetof(struct bpf_sysctl, file_pos)),
> - BPF_JMP_IMM(BPF_JNE, BPF_REG_7, 0, 2),
> +#else
> + BPF_LDX_MEM(BPF_B, BPF_REG_7, BPF_REG_1,
> + offsetof(struct bpf_sysctl, file_pos) + 3),
> +#endif
> + BPF_JMP_IMM(BPF_JNE, BPF_REG_7, 4, 2),
>
> /* return ALLOW; */
> BPF_MOV64_IMM(BPF_REG_0, 1),
> @@ -176,6 +181,7 @@ static struct sysctl_test tests[] = {
> .attach_type = BPF_CGROUP_SYSCTL,
> .sysctl = "kernel/ostype",
> .open_flags = O_RDONLY,
> + .seek = 4,
> .result = SUCCESS,
> },
> {
> --
> 2.23.0
>
