[PATCH] selftests/mm: pagemap_ioctl: use the correct page size for transact_test()

[PATCH] selftests/mm: pagemap_ioctl: use the correct page size for transact_test()

[Zenghui Yu]

There are several places in transact_test() where we use the hardcoded
0x1000 (4k) as page size, which is not always correct for architectures
supporting multiple page sizes.

Switch to use the correct page size. Otherwise ./ksft_pagemap.sh on a
16k-page-size arm64 box fails with

 $ ./ksft_pagemap.sh
 [...]
 # ok 96 mprotect_tests Both pages written after remap and mprotect
 # ok 97 mprotect_tests Clear and make the pages written
 # Bail out! ioctl failed
 # # Planned tests != run tests (117 != 97)
 # # Totals: pass:97 fail:0 xfail:0 xpass:0 skip:0 error:0
 # [FAIL]
 not ok 1 pagemap_ioctl # exit=1
 # SUMMARY: PASS=0 SKIP=0 FAIL=1
 1..1

Fixes: 46fd75d4a3c9 ("selftests: mm: add pagemap ioctl tests")
Signed-off-by: Zenghui Yu <zenghui.yu@linux.dev>
---
 tools/testing/selftests/mm/pagemap_ioctl.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/tools/testing/selftests/mm/pagemap_ioctl.c b/tools/testing/selftests/mm/pagemap_ioctl.c
index 762306177ad8..6f8971d5b3ce 100644
--- a/tools/testing/selftests/mm/pagemap_ioctl.c
+++ b/tools/testing/selftests/mm/pagemap_ioctl.c
@@ -1368,7 +1368,7 @@ void *thread_proc(void *mem)
 			ksft_exit_fail_msg("pthread_barrier_wait\n");
 
 		for (i = 0; i < access_per_thread; ++i)
-			__atomic_add_fetch(m + i * (0x1000 / sizeof(*m)), 1, __ATOMIC_SEQ_CST);
+			__atomic_add_fetch(m + i * (page_size / sizeof(*m)), 1, __ATOMIC_SEQ_CST);
 
 		ret = pthread_barrier_wait(&end_barrier);
 		if (ret && ret != PTHREAD_BARRIER_SERIAL_THREAD)
@@ -1403,15 +1403,15 @@ static void transact_test(int page_size)
 	if (pthread_barrier_init(&end_barrier, NULL, nthreads + 1))
 		ksft_exit_fail_msg("pthread_barrier_init\n");
 
-	mem = mmap(NULL, 0x1000 * nthreads * pages_per_thread, PROT_READ | PROT_WRITE,
+	mem = mmap(NULL, page_size * nthreads * pages_per_thread, PROT_READ | PROT_WRITE,
 		   MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
 	if (mem == MAP_FAILED)
 		ksft_exit_fail_msg("Error mmap %s.\n", strerror(errno));
 
-	wp_init(mem, 0x1000 * nthreads * pages_per_thread);
-	wp_addr_range(mem, 0x1000 * nthreads * pages_per_thread);
+	wp_init(mem, page_size * nthreads * pages_per_thread);
+	wp_addr_range(mem, page_size * nthreads * pages_per_thread);
 
-	memset(mem, 0, 0x1000 * nthreads * pages_per_thread);
+	memset(mem, 0, page_size * nthreads * pages_per_thread);
 
 	count = get_dirty_pages_reset(mem, nthreads * pages_per_thread, 1, page_size);
 	ksft_test_result(count > 0, "%s count %u\n", __func__, count);
@@ -1420,7 +1420,7 @@ static void transact_test(int page_size)
 
 	finish = 0;
 	for (i = 0; i < nthreads; ++i)
-		pthread_create(&th, NULL, thread_proc, mem + 0x1000 * i * pages_per_thread);
+		pthread_create(&th, NULL, thread_proc, mem + page_size * i * pages_per_thread);
 
 	extra_pages = 0;
 	for (i = 0; i < iter_count; ++i) {
-- 
2.53.0

Re: [PATCH] selftests/mm: pagemap_ioctl: use the correct page size for transact_test()

[Andrew Morton]

On Sun, 28 Jun 2026 18:11:18 +0800 Zenghui Yu <zenghui.yu@linux.dev> wrote:

> There are several places in transact_test() where we use the hardcoded
> 0x1000 (4k) as page size, which is not always correct for architectures
> supporting multiple page sizes.

doh.

> Switch to use the correct page size. Otherwise ./ksft_pagemap.sh on a
> 16k-page-size arm64 box fails with
> 
>  $ ./ksft_pagemap.sh
>  [...]
>  # ok 96 mprotect_tests Both pages written after remap and mprotect
>  # ok 97 mprotect_tests Clear and make the pages written
>  # Bail out! ioctl failed
>  # # Planned tests != run tests (117 != 97)
>  # # Totals: pass:97 fail:0 xfail:0 xpass:0 skip:0 error:0
>  # [FAIL]
>  not ok 1 pagemap_ioctl # exit=1
>  # SUMMARY: PASS=0 SKIP=0 FAIL=1
>  1..1

Thanks.  I'll add a cc:stable to this.

>  tools/testing/selftests/mm/pagemap_ioctl.c | 12 ++++++------
>  1 file changed, 6 insertions(+), 6 deletions(-)

I'd like to see this code initialize page_size and hpage_size before
calling init_uffd(), hugetlb_setup_default(), etc.  That won't fix
anything but it's safer and saner to get these globals set up before
doing other things.