Re: [PATCH RESEND 1/1] selftests/mm: skip soft-dirty tests when CONFIG_MEM_SOFT_DIRTY is disabled

[Lance Yang]

On 2025/9/17 18:51, David Hildenbrand wrote:

On 17.09.25 07:59, Lance Yang wrote:

From: Lance Yang <lance.y...@linux.dev>

The madv_populate and soft-dirty kselftests currently fail on systems where

CONFIG_MEM_SOFT_DIRTY is disabled.

Introduce a new helper softdirty_is_supported() into vm_util.c/h to ensure

tests are properly skipped when the feature is not enabled.

I'll note that tools/testing/selftests/mm/config contains

     CONFIG_MEM_SOFT_DIRTY=y

But yes, the current arm64 handling is nasty, because some other archs (e.g., riscv) also don't support it yet.

Yep.

LGTM, some nits below:

Thanks for taking time to review!

Suggested-by: David Hildenbrand <da...@redhat.com>
Signed-off-by: Lance Yang <lance.y...@linux.dev>
---
  tools/testing/selftests/mm/madv_populate.c | 21 ++--------------
  tools/testing/selftests/mm/soft-dirty.c    |  5 +++-
  tools/testing/selftests/mm/vm_util.c       | 28 ++++++++++++++++++++++
  tools/testing/selftests/mm/vm_util.h       |  1 +
  4 files changed, 35 insertions(+), 20 deletions(-)

diff --git a/tools/testing/selftests/mm/madv_populate.c b/tools/ testing/selftests/mm/madv_populate.c

index b6fabd5c27ed..43dac7783004 100644
--- a/tools/testing/selftests/mm/madv_populate.c
+++ b/tools/testing/selftests/mm/madv_populate.c
@@ -264,23 +264,6 @@ static void test_softdirty(void)
      munmap(addr, SIZE);
  }
-static int system_has_softdirty(void)
-{
-    /*
-     * There is no way to check if the kernel supports soft-dirty, other
-     * than by writing to a page and seeing if the bit was set. But the

-     * tests are intended to check that the bit gets set when it should, so

-     * doing that check would turn a potentially legitimate fail into a
-     * skip. Fortunately, we know for sure that arm64 does not support
-     * soft-dirty. So for now, let's just use the arch as a corse guide.
-     */
-#if defined(__aarch64__)
-    return 0;
-#else
-    return 1;
-#endif
-}
-
  int main(int argc, char **argv)
  {
      int nr_tests = 16;
@@ -288,7 +271,7 @@ int main(int argc, char **argv)
      pagesize = getpagesize();
-    if (system_has_softdirty())
+    if (softdirty_is_supported())
          nr_tests += 5;
      ksft_print_header();
@@ -300,7 +283,7 @@ int main(int argc, char **argv)
      test_holes();
      test_populate_read();
      test_populate_write();
-    if (system_has_softdirty())
+    if (softdirty_is_supported())
          test_softdirty();
      err = ksft_get_fail_cnt();

diff --git a/tools/testing/selftests/mm/soft-dirty.c b/tools/testing/ selftests/mm/soft-dirty.c

index 8a3f2b4b2186..98e42d2ac32a 100644
--- a/tools/testing/selftests/mm/soft-dirty.c
+++ b/tools/testing/selftests/mm/soft-dirty.c
@@ -200,8 +200,11 @@ int main(int argc, char **argv)
      int pagesize;
      ksft_print_header();
-    ksft_set_plan(15);
+    if (!softdirty_is_supported())
+        ksft_exit_skip("soft-dirty is not support\n");
+
+    ksft_set_plan(15);
      pagemap_fd = open(PAGEMAP_FILE_PATH, O_RDONLY);
      if (pagemap_fd < 0)
          ksft_exit_fail_msg("Failed to open %s\n", PAGEMAP_FILE_PATH);

diff --git a/tools/testing/selftests/mm/vm_util.c b/tools/testing/ selftests/mm/vm_util.c

index 56e9bd541edd..3173335df775 100644
--- a/tools/testing/selftests/mm/vm_util.c
+++ b/tools/testing/selftests/mm/vm_util.c
@@ -449,6 +449,34 @@ bool check_vmflag_pfnmap(void *addr)
      return check_vmflag(addr, "pf");
  }
+bool softdirty_is_supported(void)

I'd just call it "softdirty_supported" similar to "pagemap_scan_supported()".

Got it.

+{
+    char *addr;
+    int ret = 0;

bool supported = false;

+    size_t pagesize;
+
+    /* We know for sure that arm64 does not support soft-dirty. */
+#if defined(__aarch64__)
+    return ret;
+#endif

Just drop this arm special case now.

OK

+    pagesize = getpagesize();

const size_t pagesize = getpagesize();

+    /*
+     * __mmap_complete() always sets VM_SOFTDIRTY for new VMAs, so we

+     * just mmap a small region and check its VmFlags in /proc/self/ smaps

+     * for the "sd" flag.
+     */

/* New mappings are expected to be marked with VM_SOFTDIRTY (sd). */

Cool. Much better!

+    addr = mmap(0, pagesize, PROT_READ | PROT_WRITE,
+            MAP_ANONYMOUS | MAP_PRIVATE, 0, 0);
+    if (!addr)
+        ksft_exit_fail_msg("mmap failed\n");
+
+    if (check_vmflag(addr, "sd"))
+        ret = 1;

supported = true;

I'll adjust as you suggested ;)

+
+    munmap(addr, pagesize);
+    return ret;
+}
+

Cheers,
Lance