Re: [PATCH][V2] selftests/seccomp: fix test failure on s390x because of

[Kees Cook]

On Sat, Jan 26, 2019 at 3:55 AM Colin King <colin.king@xxxxxxxxxxxxx> wrote:
>
> From: Colin Ian King <colin.king@xxxxxxxxxxxxx>
>
> The error return being placed in regs.SYSCALL_RET is currently positive and
> this is causing test failures on s390x. The return value should be -EPERM
> rather than EPERM otherwise a failure is not detected and errno is not set
> accordingly on s390x.

Ah, hm, interesting. Yes, the selftest was accidentally encoding the
wrong details. However, I think the test for the result is a bit more
broken. EPERM is 1, so there isn't even a sanity check that the result
is getting passed down (the -1 from syscall won't tell us if the errno
actually landed too, so we need to check that with something that
isn't EPERM).

Let me send a patch to test on s390...

-Kees


>
> Fixes: a33b2d0359a0 ("selftests/seccomp: Add tests for basic ptrace actions")
> Signed-off-by: Colin Ian King <colin.king@xxxxxxxxxxxxx>
> ---
>
> V2: remove misplaced Content-Type and Content-Transfer-Encoding fields
>
> ---
>  tools/testing/selftests/seccomp/seccomp_bpf.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/tools/testing/selftests/seccomp/seccomp_bpf.c b/tools/testing/selftests/seccomp/seccomp_bpf.c
> index 496a9a8c773a..957344884360 100644
> --- a/tools/testing/selftests/seccomp/seccomp_bpf.c
> +++ b/tools/testing/selftests/seccomp/seccomp_bpf.c
> @@ -1706,7 +1706,7 @@ void change_syscall(struct __test_metadata *_metadata,
>  #ifdef SYSCALL_NUM_RET_SHARE_REG
>                 TH_LOG("Can't modify syscall return on this architecture");
>  #else
> -               regs.SYSCALL_RET = EPERM;
> +               regs.SYSCALL_RET = -EPERM;
>  #endif
>
>  #ifdef HAVE_GETREGS
> @@ -1850,7 +1850,7 @@ TEST_F(TRACE_syscall, ptrace_syscall_dropped)
>                                            true);
>
>         /* Tracer should skip the open syscall, resulting in EPERM. */
> -       EXPECT_SYSCALL_RETURN(EPERM, syscall(__NR_openat));
> +       EXPECT_SYSCALL_RETURN(-EPERM, syscall(__NR_openat));
>  }
>
>  TEST_F(TRACE_syscall, syscall_allowed)
> @@ -1894,7 +1894,7 @@ TEST_F(TRACE_syscall, syscall_dropped)
>         ASSERT_EQ(0, ret);
>
>         /* gettid has been skipped and an altered return value stored. */
> -       EXPECT_SYSCALL_RETURN(EPERM, syscall(__NR_gettid));
> +       EXPECT_SYSCALL_RETURN(-EPERM, syscall(__NR_gettid));
>         EXPECT_NE(self->mytid, syscall(__NR_gettid));
>  }
>
> --
> 2.19.1
>


-- 
Kees Cook