ACK: [SRU Bionic, Focal 0/1] LP: #1928522 seccomp_bpf:syscall_faked from kselftests fail on s390x
Tim Gardner
tim.gardner at canonical.com
Mon May 24 15:46:23 UTC 2021
Acked-by: Tim Gardner <tim.gardner at canonical.com>
On 5/21/21 5:36 PM, Thadeu Lima de Souza Cascardo wrote:
> [Impact]
> kselftests seccomp_bpf will start failing on s390x, though due to a test
> problem, not a kernel regression, thus wasting people's time doing test
> review.
>
> [Regression potential]
> Though this is only changing the test, we might be missing or miss failures
> in the future because some tests are not executed anymore.
> ptrace_syscall_faked, ptrace_syscall_errno and syscall_errno should all
> work just fine on s390x, but will be skipped as well with this change.
>
> [Fix]
> The fix is to consider that s390x cannot change syscall and errno at once
> by using ptrace. It can, however, change the errno at syscall exit time,
> which could even be used with seccomp SECCOMP_RET_TRACE as long as ptrace
> is used to track the exit transitions. But the test is not designed to deal
> with such situations.
>
> [Test case]
> seccomp_bpf was run, ptrace_faked passed and no new subtests have failed.
>
> Sven Schnelle (1):
> selftests/seccomp: s390 shares the syscall and return value register
>
> tools/testing/selftests/seccomp/seccomp_bpf.c | 1 +
> 1 file changed, 1 insertion(+)
>
--
-----------
Tim Gardner
Canonical, Inc
More information about the kernel-team
mailing list