Re: [PATCH v2] selftests/harness: harness-selftest: cover XFAIL_ADD ctor ordering

[Thomas Weißschuh]

Hi!

On 2026-03-09 18:10:44+0800, Sun Jian wrote:
> Add a fixture-based test with a single variant and mark it with
> XFAIL_ADD(). The test intentionally fails and is expected to be
> accounted as XFAIL, covering the constructor ordering dependency
> between TEST_F() registration and XFAIL_ADD() registration.
> 
> Suggested-by: Thomas Weißschuh <linux@xxxxxxxxxxxxxx>
> Signed-off-by: Sun Jian <sun.jian.kdev@xxxxxxxxx>
> ---
> v2:
> - update Suggested-by address
> - update harness-selftest.expected
> - verify with 'make run_tests'
> 
>  .../kselftest_harness/harness-selftest.c      | 23 ++++++++++
>  .../harness-selftest.expected                 | 43 +++++++++++--------
>  2 files changed, 47 insertions(+), 19 deletions(-)

Can you rebase this on kselftest/fixes?
https://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest.git/log/?h=fixes

This doesn't apply anymore and I suspect the proposed ordering suffers
from the issue fixed by commit 465b05bae5ac ("selftests: harness:
Restore order of test functions")


Thomas