On 24.06.25 06:24, Li Wang wrote:
> The current implementation of test_unmerge_uffd_wp() explicitly sets
> `uffdio_api.features = UFFD_FEATURE_PAGEFAULT_FLAG_WP` before calling
> UFFDIO_API. This can cause the ioctl() call to fail with EINVAL on kernels
> that do not support UFFD-WP, leading the test to fail unnecessarily:
>
>    # ------------------------------
>    # running ./ksm_functional_tests
>    # ------------------------------
>    # TAP version 13
>    # 1..9
>    # # [RUN] test_unmerge
>    # ok 1 Pages were unmerged
>    # # [RUN] test_unmerge_zero_pages
>    # ok 2 KSM zero pages were unmerged
>    # # [RUN] test_unmerge_discarded
>    # ok 3 Pages were unmerged
>    # # [RUN] test_unmerge_uffd_wp
>    # not ok 4 UFFDIO_API failed     <-----
>    # # [RUN] test_prot_none
>    # ok 5 Pages were unmerged
>    # # [RUN] test_prctl
>    # ok 6 Setting/clearing PR_SET_MEMORY_MERGE works
>    # # [RUN] test_prctl_fork
>    # # No pages got merged
>    # # [RUN] test_prctl_fork_exec
>    # ok 7 PR_SET_MEMORY_MERGE value is inherited
>    # # [RUN] test_prctl_unmerge
>    # ok 8 Pages were unmerged
>    # Bail out! 1 out of 8 tests failed
>    # # Planned tests != run tests (9 != 8)
>    # # Totals: pass:7 fail:1 xfail:0 xpass:0 skip:0 error:0
>    # [FAIL]
>
> This patch improves compatibility and robustness of the UFFD-WP test
> (test_unmerge_uffd_wp) by correctly implementing the UFFDIO_API
> two-step handshake as recommended by the userfaultfd(2) man page.
>
> Key changes:
>
> 1. Use features=0 in the initial UFFDIO_API call to query supported
>     feature bits, rather than immediately requesting WP support.
>
> 2. Skip the test gracefully if:
>     - UFFDIO_API fails with EINVAL (e.g. unsupported API version), or
>     - UFFD_FEATURE_PAGEFAULT_FLAG_WP is not advertised by the kernel.
>
> 3. Close the initial userfaultfd and create a new one before enabling
>     the required feature, since UFFDIO_API can only be called once per fd.
>
> 4. Improve diagnostics by distinguishing between expected and unexpected
>     failures, using strerror() to report errors.
>
> This ensures the test behaves correctly across a wider range of kernel
> versions and configurations, while preserving the intended behavior on
> kernels that support UFFD-WP.
>
> Suggestted-by: David Hildenbrand <david@redhat.com>
> Signed-off-by: Li Wang <liwang@redhat.com>
> Cc: Aruna Ramakrishna <aruna.ramakrishna@oracle.com>
> Cc: Bagas Sanjaya <bagasdotme@gmail.com>
> Cc: Catalin Marinas <catalin.marinas@arm.com>
> Cc: Dave Hansen <dave.hansen@linux.intel.com>
> Cc: Joey Gouly <joey.gouly@arm.com>
> Cc: Johannes Weiner <hannes@cmpxchg.org>
> Cc: Keith Lucas <keith.lucas@oracle.com>
> Cc: Ryan Roberts <ryan.roberts@arm.com>
> Cc: Shuah Khan <shuah@kernel.org>
> ---
>
> Notes:
>      v1 --> v2:
>       * Close the original userfaultfd and open a new one before enabling features
>       * Reworked UFFDIO_API negotiation to follow the official two-step handshake
>
>   .../selftests/mm/ksm_functional_tests.c       | 28 +++++++++++++++++--
>   1 file changed, 26 insertions(+), 2 deletions(-)
>
> diff --git a/tools/testing/selftests/mm/ksm_functional_tests.c b/tools/testing/selftests/mm/ksm_functional_tests.c
> index b61803e36d1c..19e5b741893a 100644
> --- a/tools/testing/selftests/mm/ksm_functional_tests.c
> +++ b/tools/testing/selftests/mm/ksm_functional_tests.c
> @@ -393,9 +393,13 @@ static void test_unmerge_uffd_wp(void)
>   
>       /* See if UFFD-WP is around. */
>       uffdio_api.api = UFFD_API;
> -     uffdio_api.features = UFFD_FEATURE_PAGEFAULT_FLAG_WP;
> +     uffdio_api.features = 0;
>       if (ioctl(uffd, UFFDIO_API, &uffdio_api) < 0) {
> -             ksft_test_result_fail("UFFDIO_API failed\n");
> +             if (errno == EINVAL)
> +                     ksft_test_result_skip("The API version requested is not supported\n");
> +             else
> +                     ksft_test_result_fail("UFFDIO_API failed: %s\n", strerror(errno));
> +
>               goto close_uffd;
>       }
>       if (!(uffdio_api.features & UFFD_FEATURE_PAGEFAULT_FLAG_WP)) {
> @@ -403,6 +407,26 @@ static void test_unmerge_uffd_wp(void)
>               goto close_uffd;
>       }
>   
> +     /*
> +      * UFFDIO_API must only be called once to enable features.
> +      * So we close the old userfaultfd and create a new one to
> +      * actually enable UFFD_FEATURE_PAGEFAULT_FLAG_WP.
> +      */
> +     close(uffd);
> +     uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK);
> +     if (uffd < 0) {
> +             ksft_test_result_skip("__NR_userfaultfd failed\n");

If it now suddenly fails (after it working above), this sure is a fail,
right?

Apart from that

Acked-by: David Hildenbrand <david@redhat.com>