Hi SJ, Andrew, The kernel test robot caught this issue in my earlier addr_unit KUnit patch. Feel free to fold this fix into that patch. Thanks, Aveline Noir -- >8 -- When phys_addr_t is 32-bit, Sparse warns about truncating the u64 address passed to damon_pa_core_addr() in damon_test_pa_addr_over_4g(). The runtime kunit_skip() does not prevent Sparse from checking that conversion. Compile the high-address assertions only with CONFIG_PHYS_ADDR_T_64BIT, and keep the test skipped otherwise. Fixes: 7a983cc1b573 ("mm/damon/paddr: add KUnit tests for addr_unit conversion") Reported-by: kernel test robot Closes: https://lore.kernel.org/oe-kbuild-all/20261117170155.fEdeuuhm-lkp@intel.com/ Signed-off-by: Aveline Noir --- The reported warning is gone with the original i386 config, Clang 22.1.8 and Sparse v0.6.5-rc1-21-g84ec0658. All executed DAMON KUnit tests passed on x86_64 and i386 with and without PAE; the high-address case skips without PAE. --- mm/damon/tests/paddr-kunit.h | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/mm/damon/tests/paddr-kunit.h b/mm/damon/tests/paddr-kunit.h index 9839cc233071..2b11f0e67a12 100644 --- a/mm/damon/tests/paddr-kunit.h +++ b/mm/damon/tests/paddr-kunit.h @@ -26,15 +26,16 @@ static void damon_test_pa_core_addr(struct kunit *test) static void damon_test_pa_addr_over_4g(struct kunit *test) { +#ifdef CONFIG_PHYS_ADDR_T_64BIT unsigned long addr = 0xffffffffUL; unsigned long addr_unit = 0x1000; u64 pa = 0xffffffff000ULL; - if (!IS_ENABLED(CONFIG_PHYS_ADDR_T_64BIT)) - kunit_skip(test, "phys_addr_t is 32-bit"); - KUNIT_EXPECT_EQ(test, damon_pa_phys_addr(addr, addr_unit), pa); KUNIT_EXPECT_EQ(test, damon_pa_core_addr(pa, addr_unit), addr); +#else + kunit_skip(test, "phys_addr_t is 32-bit"); +#endif } static struct kunit_case damon_paddr_test_cases[] = { --- base-commit: 7a983cc1b57309a875a9e07a934c9944bf3d088d change-id: 20261118-fix-damon-paddr-sparse-20261118-fb08902768d7