mbox series

[B,C,SRU,0/1] selftests/seccomp: Enhance per-arch ptracesyscall skip tests

Message ID 20190516102023.19004-1-po-hsu.lin@canonical.com
Headers show
Series selftests/seccomp: Enhance per-arch ptracesyscall skip tests | expand

Message

Po-Hsu Lin May 16, 2019, 10:20 a.m. UTC
BugLink: https://bugs.launchpad.net/bugs/1812796

== SRU Justification ==
The TRACE_syscall.ptrace_syscall_dropped test in seccomp from
ubuntu_kernel_selftests will fail on a Power8 node with B/C kernel.

[ RUN ] TRACE_syscall.ptrace_syscall_dropped
seccomp_bpf.c:1808:TRACE_syscall.ptrace_syscall_dropped:Expected 1 (1) == syscall(286) (18446744073709551615)
TRACE_syscall.ptrace_syscall_dropped: Test failed at step #13
[ FAIL ] TRACE_syscall.ptrace_syscall_dropped

== Fix ==
* ed5f1326 (selftests/seccomp: Enhance per-arch ptrace syscall skip tests)

This patch can be cherry-picked into B/C, and it's already in Disco.
The ptrace_syscall_dropped will be changed into ptrace_syscall_faked,
and a new ptrace_syscall_errno test will be added.

== Test ==
Patch tested with different arch (amd64 / power8 / arm64 / i386), the
new test will pass on them except for PowerPC.

Although the faked (bug 1829363) and errno (bug 1829359) test will fail
on PowerPC, it's actually helping us to nail down a real issue in our
kernel.

== Regression Potential ==
Low, changes limited to the testing tool in the kernel.

Kees Cook (1):
  selftests/seccomp: Enhance per-arch ptrace syscall skip tests

 tools/testing/selftests/seccomp/seccomp_bpf.c | 72 +++++++++++++++++++++------
 1 file changed, 57 insertions(+), 15 deletions(-)

Comments

Connor Kuehl June 3, 2019, 6:24 p.m. UTC | #1
On 5/16/19 3:20 AM, Po-Hsu Lin wrote:
> BugLink: https://bugs.launchpad.net/bugs/1812796
> 
> == SRU Justification ==
> The TRACE_syscall.ptrace_syscall_dropped test in seccomp from
> ubuntu_kernel_selftests will fail on a Power8 node with B/C kernel.
> 
> [ RUN ] TRACE_syscall.ptrace_syscall_dropped
> seccomp_bpf.c:1808:TRACE_syscall.ptrace_syscall_dropped:Expected 1 (1) == syscall(286) (18446744073709551615)
> TRACE_syscall.ptrace_syscall_dropped: Test failed at step #13
> [ FAIL ] TRACE_syscall.ptrace_syscall_dropped
> 
> == Fix ==
> * ed5f1326 (selftests/seccomp: Enhance per-arch ptrace syscall skip tests)
> 
> This patch can be cherry-picked into B/C, and it's already in Disco.
> The ptrace_syscall_dropped will be changed into ptrace_syscall_faked,
> and a new ptrace_syscall_errno test will be added.
> 
> == Test ==
> Patch tested with different arch (amd64 / power8 / arm64 / i386), the
> new test will pass on them except for PowerPC.
> 
> Although the faked (bug 1829363) and errno (bug 1829359) test will fail
> on PowerPC, it's actually helping us to nail down a real issue in our
> kernel.
> 
> == Regression Potential ==
> Low, changes limited to the testing tool in the kernel.
> 
> Kees Cook (1):
>   selftests/seccomp: Enhance per-arch ptrace syscall skip tests
> 
>  tools/testing/selftests/seccomp/seccomp_bpf.c | 72 +++++++++++++++++++++------
>  1 file changed, 57 insertions(+), 15 deletions(-)
> 

Acked-by: Connor Kuehl <connor.kuehl@canonical.com>
Kleber Sacilotto de Souza June 4, 2019, 12:51 p.m. UTC | #2
On 5/16/19 12:20 PM, Po-Hsu Lin wrote:
> BugLink: https://bugs.launchpad.net/bugs/1812796
> 
> == SRU Justification ==
> The TRACE_syscall.ptrace_syscall_dropped test in seccomp from
> ubuntu_kernel_selftests will fail on a Power8 node with B/C kernel.
> 
> [ RUN ] TRACE_syscall.ptrace_syscall_dropped
> seccomp_bpf.c:1808:TRACE_syscall.ptrace_syscall_dropped:Expected 1 (1) == syscall(286) (18446744073709551615)
> TRACE_syscall.ptrace_syscall_dropped: Test failed at step #13
> [ FAIL ] TRACE_syscall.ptrace_syscall_dropped
> 
> == Fix ==
> * ed5f1326 (selftests/seccomp: Enhance per-arch ptrace syscall skip tests)
> 
> This patch can be cherry-picked into B/C, and it's already in Disco.
> The ptrace_syscall_dropped will be changed into ptrace_syscall_faked,
> and a new ptrace_syscall_errno test will be added.
> 
> == Test ==
> Patch tested with different arch (amd64 / power8 / arm64 / i386), the
> new test will pass on them except for PowerPC.
> 
> Although the faked (bug 1829363) and errno (bug 1829359) test will fail
> on PowerPC, it's actually helping us to nail down a real issue in our
> kernel.
> 
> == Regression Potential ==
> Low, changes limited to the testing tool in the kernel.
> 
> Kees Cook (1):
>   selftests/seccomp: Enhance per-arch ptrace syscall skip tests
> 
>  tools/testing/selftests/seccomp/seccomp_bpf.c | 72 +++++++++++++++++++++------
>  1 file changed, 57 insertions(+), 15 deletions(-)
> 

Applied to {bionic,cosmic]/master-next branches.

Thanks,
Kleber