Skip to contentSkip to conversation

selftests/mm: skip several tests when thp is not available

PATCH Superseded Chunyu Hu 5 participants 15 messages Mar 19 – Mar 23

Subsystems (4)

v4 of 8 Superseded Latest is v8 →

lore
Share & tools
lore

Chunyu Hu

author
#0 lore raw

v4 cover selftests/mm: skip several tests when thp is not available

There are several tests requires transprarent hugepages, when run on thp disabled kernel such as realtime kernel, there will be false negative. Mark those tests as skip when thp is not available.

Changes in v4:
- patch 1 update to use thp_available instead of thp_is_enabled, suggested by ai review
- removed reviewed-by and acked-by

Changes in v3:
- patch 1 update commit message to show the log snippet with where the fail happens, change the 'false positive' to 'false negative'. Update reviwed by
- patch 2 update reviewed-by
- patch 3 make write_file to void return
- patch 4 and patch 5 updated reviewed-by/acked-by

Changes in v2:
- replace 'false postive' with 'false negative' in cover letter
- patch 1 and patch 2 add reviewed-by/acked-by
- new patch 3 to add write_file() in vm_util as a common helper
- patch 4 removed the {} in if block, removed the write_file helper rename chunk in v1. Add reviewed-by
- patch 5 move the exit chunk to the front of ksft_print_headers() as suggested by David. Add review by.

Chunyu Hu (5):
selftests/mm/guard-regions: skip collapse test when thp not enabled selftests/mm: soft-dirty: skip two tests when thp is not available selftests/mm: move write_file helper to vm_util selftests/mm: split_huge_page_test: skip the test when thp is not
available
selftests/mm: transhuge_stress: skip the test when thp not available

 tools/testing/selftests/mm/guard-regions.c    |  4 ++++
 tools/testing/selftests/mm/soft-dirty.c       |  4 +++-
 .../selftests/mm/split_huge_page_test.c       | 19 ++++-----------
 tools/testing/selftests/mm/thp_settings.c     | 24 +------------------
 tools/testing/selftests/mm/thp_settings.h     |  1 -
 tools/testing/selftests/mm/transhuge-stress.c |  4 ++++
 tools/testing/selftests/mm/vm_util.c          | 16 +++++++++++++
 tools/testing/selftests/mm/vm_util.h          |  2 ++
 8 files changed, 34 insertions(+), 40 deletions(-)

base-commit: 8a30aeb0d1b4e4aaf7f7bae72f20f2ae75385ccb

Signature · 1 line
2.53.0
author
#1 lore raw

PATCH v4 1/5 selftests/mm/guard-regions: skip collapse test when thp not enabled

When thp is not available, just skip the collape tests to avoid the false negative.

Without the change, run with a thp disabled kernel:
./run_vmtests.sh -t madv_guard -n 1
<snip/>
# RUN guard_regions.anon.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.anon.collapse
not ok 2 guard_regions.anon.collapse
<snip/>
# RUN guard_regions.shmem.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.shmem.collapse
not ok 32 guard_regions.shmem.collapse
<snip/>
# RUN guard_regions.file.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.file.collapse
not ok 62 guard_regions.file.collapse
<snip/>
# FAILED: 87 / 90 tests passed.
# 17 skipped test(s) detected. Consider enabling relevant config options to improve coverage. # Totals: pass:70 fail:3 xfail:0 xpass:0 skip:17 error:0

With this change, run with thp disabled kernel:
./run_vmtests.sh -t madv_guard -n 1
<snip/>

  #  RUN           guard_regions.anon.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.anon.collapse
  ok 2 guard_regions.anon.collapse # SKIP Transparent Hugepages not available
  <snip/>
  #  RUN           guard_regions.file.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.file.collapse
  ok 62 guard_regions.file.collapse # SKIP Transparent Hugepages not available
  <snip/>
  #  RUN           guard_regions.shmem.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.shmem.collapse

ok 32 guard_regions.shmem.collapse # SKIP Transparent Hugepages not available <snip/>
# PASSED: 90 / 90 tests passed.
# 20 skipped test(s) detected. Consider enabling relevant config options to improve coverage. # Totals: pass:70 fail:0 xfail:0 xpass:0 skip:20 error:0

---
Changes in v4:
- use thp_available instead of thp_is_enabled() as when thp is set to never, madvise(MADV_COLLAPSE) will still succeed by design. So a failure in madvise(MADV_COLLAPSE) on guard region will verify guard region denies it. This is suggested from AI.
- removed the 'Reviewed-by' and 'Acked-by' as the code changes. Sorry for that but it seems the changes is a little huge? (1 out of 2 lines)..
Changes in v3:
- commit message: update the log snippet with where the fail happens and add the '-n1' to the command.
- fix the 'false positive' to 'false negative'
- add reviwed by from Mike
Changes in v2:
- add reviewed by from Zi and Lorenzo
- add acked-by from David

---
 tools/testing/selftests/mm/guard-regions.c | 4 ++++
 1 file changed, 4 insertions(+)
tools/testing/selftests/mm/guard-regions.c +4 -0
index dbd21d66d383..48e8b1539be3 100644
@@ -21,6 +21,7 @@
 #include <sys/uio.h>
 #include <unistd.h>
 #include "vm_util.h"
+#include "thp_settings.h"
 
 #include "../pidfd/pidfd.h"
 
@@ -2195,6 +2196,9 @@ TEST_F(guard_regions, collapse)
 	char *ptr;
 	int i;
 
+	if (!thp_available())
+		SKIP(return, "Transparent Hugepages not available\n");
+
 	/* Need file to be correct size for tests for non-anon. */
 	if (variant->backing != ANON_BACKED)
 		ASSERT_EQ(ftruncate(self->fd, size), 0);

base-commit: 8a30aeb0d1b4e4aaf7f7bae72f20f2ae75385ccb

Signature · 1 line
2.53.0
reviewer Acked-by on v4 1/5
#2 lore raw

Re: [PATCH v4 1/5] selftests/mm/guard-regions: skip collapse test when thp not enabled

Quoted 44 lines · Chunyu Hu

When thp is not available, just skip the collape tests to avoid the false negative.

Without the change, run with a thp disabled kernel:
./run_vmtests.sh -t madv_guard -n 1
<snip/>
# RUN guard_regions.anon.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.anon.collapse
not ok 2 guard_regions.anon.collapse
<snip/>
# RUN guard_regions.shmem.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.shmem.collapse
not ok 32 guard_regions.shmem.collapse
<snip/>
# RUN guard_regions.file.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.file.collapse
not ok 62 guard_regions.file.collapse
<snip/>
# FAILED: 87 / 90 tests passed.
# 17 skipped test(s) detected. Consider enabling relevant config options to improve coverage. # Totals: pass:70 fail:3 xfail:0 xpass:0 skip:17 error:0

With this change, run with thp disabled kernel:
./run_vmtests.sh -t madv_guard -n 1
<snip/>

  #  RUN           guard_regions.anon.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.anon.collapse
  ok 2 guard_regions.anon.collapse # SKIP Transparent Hugepages not available
  <snip/>
  #  RUN           guard_regions.file.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.file.collapse
  ok 62 guard_regions.file.collapse # SKIP Transparent Hugepages not available
  <snip/>
  #  RUN           guard_regions.shmem.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.shmem.collapse
Quoted 8 lines

ok 32 guard_regions.shmem.collapse # SKIP Transparent Hugepages not available <snip/>
# PASSED: 90 / 90 tests passed.
# 20 skipped test(s) detected. Consider enabling relevant config options to improve coverage. # Totals: pass:70 fail:0 xfail:0 xpass:0 skip:20 error:0

Quoted 47 lines

---
Changes in v4:
- use thp_available instead of thp_is_enabled() as when thp is set to never, madvise(MADV_COLLAPSE) will still succeed by design. So a failure in madvise(MADV_COLLAPSE) on guard region will verify guard region denies it. This is suggested from AI.
- removed the 'Reviewed-by' and 'Acked-by' as the code changes. Sorry for that but it seems the changes is a little huge? (1 out of 2 lines)..
Changes in v3:
- commit message: update the log snippet with where the fail happens and add the '-n1' to the command.
- fix the 'false positive' to 'false negative'
- add reviwed by from Mike
Changes in v2:
- add reviewed by from Zi and Lorenzo
- add acked-by from David

---
 tools/testing/selftests/mm/guard-regions.c | 4 ++++
 1 file changed, 4 insertions(+)
tools/testing/selftests/mm/guard-regions.c +4 -0
index dbd21d66d383..48e8b1539be3 100644
@@ -21,6 +21,7 @@
 #include <sys/uio.h>
 #include <unistd.h>
 #include "vm_util.h"
+#include "thp_settings.h"
 
 #include "../pidfd/pidfd.h"
 
@@ -2195,6 +2196,9 @@ TEST_F(guard_regions, collapse)
 	char *ptr;
 	int i;
 
+	if (!thp_available())
+		SKIP(return, "Transparent Hugepages not available\n");
+
 	/* Need file to be correct size for tests for non-anon. */
 	if (variant->backing != ANON_BACKED)
 		ASSERT_EQ(ftruncate(self->fd, size), 0);

base-commit: 8a30aeb0d1b4e4aaf7f7bae72f20f2ae75385ccb

Signature · 1 line
2.53.0
Signature · 2 lines
Sincerely yours,
Mike.
reviewer Reviewed-by on v4 1/5
#3 lore raw

Re: [PATCH v4 1/5] selftests/mm/guard-regions: skip collapse test when thp not enabled

In reply to Chunyu Hu
Quoted 68 lines · Chunyu Hu

When thp is not available, just skip the collape tests to avoid the false negative.

Without the change, run with a thp disabled kernel:
./run_vmtests.sh -t madv_guard -n 1
<snip/>
# RUN guard_regions.anon.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.anon.collapse
not ok 2 guard_regions.anon.collapse
<snip/>
# RUN guard_regions.shmem.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.shmem.collapse
not ok 32 guard_regions.shmem.collapse
<snip/>
# RUN guard_regions.file.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.file.collapse
not ok 62 guard_regions.file.collapse
<snip/>
# FAILED: 87 / 90 tests passed.
# 17 skipped test(s) detected. Consider enabling relevant config options to improve coverage. # Totals: pass:70 fail:3 xfail:0 xpass:0 skip:17 error:0

With this change, run with thp disabled kernel:
./run_vmtests.sh -t madv_guard -n 1
<snip/>

  #  RUN           guard_regions.anon.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.anon.collapse
  ok 2 guard_regions.anon.collapse # SKIP Transparent Hugepages not available
  <snip/>
  #  RUN           guard_regions.file.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.file.collapse
  ok 62 guard_regions.file.collapse # SKIP Transparent Hugepages not available
  <snip/>
  #  RUN           guard_regions.shmem.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.shmem.collapse

ok 32 guard_regions.shmem.collapse # SKIP Transparent Hugepages not available <snip/>
# PASSED: 90 / 90 tests passed.
# 20 skipped test(s) detected. Consider enabling relevant config options to improve coverage. # Totals: pass:70 fail:0 xfail:0 xpass:0 skip:20 error:0

---
Changes in v4:
- use thp_available instead of thp_is_enabled() as when thp is set to never, madvise(MADV_COLLAPSE) will still succeed by design. So a failure in madvise(MADV_COLLAPSE) on guard region will verify guard region denies it. This is suggested from AI.
- removed the 'Reviewed-by' and 'Acked-by' as the code changes. Sorry for that but it seems the changes is a little huge? (1 out of 2 lines)..
Changes in v3:
- commit message: update the log snippet with where the fail happens and add the '-n1' to the command.
- fix the 'false positive' to 'false negative'
- add reviwed by from Mike
Changes in v2:
- add reviewed by from Zi and Lorenzo
- add acked-by from David

Quoted 4 lines
---
 tools/testing/selftests/mm/guard-regions.c | 4 ++++
 1 file changed, 4 insertions(+)

Best Regards,
Yan, Zi

maintainer Acked-by on v4 1/5
#4 lore raw

Re: [PATCH v4 1/5] selftests/mm/guard-regions: skip collapse test when thp not enabled

In reply to Chunyu Hu
Quoted 44 lines · Chunyu Hu

When thp is not available, just skip the collape tests to avoid the false negative.

Without the change, run with a thp disabled kernel:
./run_vmtests.sh -t madv_guard -n 1
<snip/>
# RUN guard_regions.anon.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.anon.collapse
not ok 2 guard_regions.anon.collapse
<snip/>
# RUN guard_regions.shmem.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.shmem.collapse
not ok 32 guard_regions.shmem.collapse
<snip/>
# RUN guard_regions.file.collapse ...
# guard-regions.c:2217:collapse:Expected madvise(ptr, size, MADV_NOHUGEPAGE) (-1) == 0 (0) # collapse: Test terminated by assertion
# FAIL guard_regions.file.collapse
not ok 62 guard_regions.file.collapse
<snip/>
# FAILED: 87 / 90 tests passed.
# 17 skipped test(s) detected. Consider enabling relevant config options to improve coverage. # Totals: pass:70 fail:3 xfail:0 xpass:0 skip:17 error:0

With this change, run with thp disabled kernel:
./run_vmtests.sh -t madv_guard -n 1
<snip/>

  #  RUN           guard_regions.anon.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.anon.collapse
  ok 2 guard_regions.anon.collapse # SKIP Transparent Hugepages not available
  <snip/>
  #  RUN           guard_regions.file.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.file.collapse
  ok 62 guard_regions.file.collapse # SKIP Transparent Hugepages not available
  <snip/>
  #  RUN           guard_regions.shmem.collapse ...
  #      SKIP      Transparent Hugepages not available
  #            OK  guard_regions.shmem.collapse
Quoted 8 lines

ok 32 guard_regions.shmem.collapse # SKIP Transparent Hugepages not available <snip/>
# PASSED: 90 / 90 tests passed.
# 20 skipped test(s) detected. Consider enabling relevant config options to improve coverage. # Totals: pass:70 fail:0 xfail:0 xpass:0 skip:20 error:0

Signature · 3 lines
Cheers,

David
author
#5 lore raw

PATCH v4 2/5 selftests/mm: soft-dirty: skip two tests when thp is not available

In reply to Chunyu Hu

The test_hugepage test contain two sub tests. If just reporting one skip when thp not available, there will be error in the log because the test count don't match the test plan. Change to skip two tests by running the ksft_test_result_skip twice in this case.

Without the fix (run test on thp disabled kernel):
./run_vmtests.sh -t soft_dirty
# --------------------

  # running ./soft-dirty
  # --------------------
  # TAP version 13
  # 1..19
  # ok 1 Test test_simple
  # ok 2 Test test_vma_reuse dirty bit of allocated page
  # ok 3 Test test_vma_reuse dirty bit of reused address page
  # ok 4 # SKIP Transparent Hugepages not available
  # ok 5 Test test_mprotect-anon dirty bit of new written page
  # ok 6 Test test_mprotect-anon soft-dirty clear after clear_refs
  # ok 7 Test test_mprotect-anon soft-dirty clear after marking RO
  # ok 8 Test test_mprotect-anon soft-dirty clear after marking RW
  # ok 9 Test test_mprotect-anon soft-dirty after rewritten
  # ok 10 Test test_mprotect-file dirty bit of new written page
  # ok 11 Test test_mprotect-file soft-dirty clear after clear_refs
  # ok 12 Test test_mprotect-file soft-dirty clear after marking RO
  # ok 13 Test test_mprotect-file soft-dirty clear after marking RW
  # ok 14 Test test_mprotect-file soft-dirty after rewritten
  # ok 15 Test test_merge-anon soft-dirty after remap merge 1st pg
  # ok 16 Test test_merge-anon soft-dirty after remap merge 2nd pg
  # ok 17 Test test_merge-anon soft-dirty after mprotect merge 1st pg
  # ok 18 Test test_merge-anon soft-dirty after mprotect merge 2nd pg
  # # 1 skipped test(s) detected. Consider enabling relevant config options to improve coverage.
  # # Planned tests != run tests (19 != 18)
  # # Totals: pass:17 fail:0 xfail:0 xpass:0 skip:1 error:0
  # [FAIL]
  not ok 52 soft-dirty # exit=1

With the fix (run test on thp disabled kernel):
./run_vmtests.sh -t soft_dirty
# --------------------

  # running ./soft-dirty
  # TAP version 13
  # --------------------
  # running ./soft-dirty
  # --------------------
  # TAP version 13
  # 1..19
  # ok 1 Test test_simple
  # ok 2 Test test_vma_reuse dirty bit of allocated page
  # ok 3 Test test_vma_reuse dirty bit of reused address page
  # # Transparent Hugepages not available
  # ok 4 # SKIP Test test_hugepage huge page allocation
  # ok 5 # SKIP Test test_hugepage huge page dirty bit
  # ok 6 Test test_mprotect-anon dirty bit of new written page
  # ok 7 Test test_mprotect-anon soft-dirty clear after clear_refs
  # ok 8 Test test_mprotect-anon soft-dirty clear after marking RO
  # ok 9 Test test_mprotect-anon soft-dirty clear after marking RW
  # ok 10 Test test_mprotect-anon soft-dirty after rewritten
  # ok 11 Test test_mprotect-file dirty bit of new written page
  # ok 12 Test test_mprotect-file soft-dirty clear after clear_refs
  # ok 13 Test test_mprotect-file soft-dirty clear after marking RO
  # ok 14 Test test_mprotect-file soft-dirty clear after marking RW
  # ok 15 Test test_mprotect-file soft-dirty after rewritten
  # ok 16 Test test_merge-anon soft-dirty after remap merge 1st pg
  # ok 17 Test test_merge-anon soft-dirty after remap merge 2nd pg
  # ok 18 Test test_merge-anon soft-dirty after mprotect merge 1st pg
  # ok 19 Test test_merge-anon soft-dirty after mprotect merge 2nd pg
  # # 2 skipped test(s) detected. Consider enabling relevant config options to improve coverage.
  # # Totals: pass:17 fail:0 xfail:0 xpass:0 skip:2 error:0
  # [PASS]
  ok 1 soft-dirty
  hwpoison_inject
  # SUMMARY: PASS=1 SKIP=0 FAIL=0
  1..1

---
Changes in v3:
- Add reviewed by from Mike
Changes in v2:
- add reviewed by from Lorenzo and Zi
- add acked-by from David

---
 tools/testing/selftests/mm/soft-dirty.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
tools/testing/selftests/mm/soft-dirty.c +3 -1
index 59c0dbe99a9b..bcfcac99b436 100644
@@ -82,7 +82,9 @@ static void test_hugepage(int pagemap_fd, int pagesize)
 	int i, ret;
 
 	if (!thp_is_enabled()) {
-		ksft_test_result_skip("Transparent Hugepages not available\n");
+		ksft_print_msg("Transparent Hugepages not available\n");
+		ksft_test_result_skip("Test %s huge page allocation\n", __func__);
+		ksft_test_result_skip("Test %s huge page dirty bit\n", __func__);
 		return;
 	}
 
Signature · 1 line
2.53.0
author
#6 lore raw

PATCH v4 3/5 selftests/mm: move write_file helper to vm_util

In reply to Chunyu Hu

thp_settings provides write_file() helper for safely writing to a file and exit when write failure happens. It's a very low level helper and many sub tests need such a helper, not only thp tests.

split_huge_page_test also defines a write_file locally. The two have minior differences in return type and used exit api. And there would be conflicts if split_huge_page_test wanted to include thp_settings.h because of different prototype, making it less convenient.

It's possisble to merge the two, although some tests don't use the kselftest infrastrucutre for testing. It would also work when using the ksft_exit_msg() to exit in my test, as the counters are all zero. Output will be like:

TAP version 13
1..62
Bail out! /proc/sys/vm/drop_caches1 open failed: No such file or directory # Totals: pass:0 fail:0 xfail:0 xpass:0 skip:0 error:0

So here we just keep the version in split_huge_page_test, and move it into the vm_util. This makes it easy to maitain and user could just include one vm_util.h when they don't need thp setting helpers. Keep the prototype of void return as the function will exit on any error, return value is not necessary, and will simply the callers like write_num() and write_string().

---
Changes in v3:
- make the write_file void return
Changes in v2:
new patch from v2

---
 .../selftests/mm/split_huge_page_test.c       | 15 --------
 tools/testing/selftests/mm/thp_settings.c     | 35 ++-----------------
 tools/testing/selftests/mm/thp_settings.h     |  1 -
 tools/testing/selftests/mm/vm_util.c          | 15 ++++++++
 tools/testing/selftests/mm/vm_util.h          |  2 ++
 5 files changed, 20 insertions(+), 48 deletions(-)
tools/testing/selftests/mm/split_huge_page_test.c +0 -15
index e0167111bdd1..93f205327b84 100644
@@ -255,21 +255,6 @@ static int check_after_split_folio_orders(char *vaddr_start, size_t len,
 	return status;
 }
 
-static void write_file(const char *path, const char *buf, size_t buflen)
-{
-	int fd;
-	ssize_t numwritten;
-
-	fd = open(path, O_WRONLY);
-	if (fd == -1)
-		ksft_exit_fail_msg("%s open failed: %s\n", path, strerror(errno));
-
-	numwritten = write(fd, buf, buflen - 1);
-	close(fd);
-	if (numwritten < 1)
-		ksft_exit_fail_msg("Write failed\n");
-}
-
 static void write_debugfs(const char *fmt, ...)
 {
 	char input[INPUT_MAX];
tools/testing/selftests/mm/thp_settings.c +3 -32
index 574bd0f8ae48..e748ebfb3d4e 100644
@@ -6,6 +6,7 @@
 #include <string.h>
 #include <unistd.h>
 
+#include "vm_util.h"
 #include "thp_settings.h"
 
 #define THP_SYSFS "/sys/kernel/mm/transparent_hugepage/"
@@ -64,29 +65,6 @@ int read_file(const char *path, char *buf, size_t buflen)
 	return (unsigned int) numread;
 }
 
-int write_file(const char *path, const char *buf, size_t buflen)
-{
-	int fd;
-	ssize_t numwritten;
-
-	fd = open(path, O_WRONLY);
-	if (fd == -1) {
-		printf("open(%s)\n", path);
-		exit(EXIT_FAILURE);
-		return 0;
-	}
-
-	numwritten = write(fd, buf, buflen - 1);
-	close(fd);
-	if (numwritten < 1) {
-		printf("write(%s)\n", buf);
-		exit(EXIT_FAILURE);
-		return 0;
-	}
-
-	return (unsigned int) numwritten;
-}
-
 unsigned long read_num(const char *path)
 {
 	char buf[21];
@@ -104,10 +82,7 @@ void write_num(const char *path, unsigned long num)
 	char buf[21];
 
 	sprintf(buf, "%ld", num);
-	if (!write_file(path, buf, strlen(buf) + 1)) {
-		perror(path);
-		exit(EXIT_FAILURE);
-	}
+	write_file(path, buf, strlen(buf) + 1);
 }
 
 int thp_read_string(const char *name, const char * const strings[])
@@ -165,11 +140,7 @@ void thp_write_string(const char *name, const char *val)
 		printf("%s: Pathname is too long\n", __func__);
 		exit(EXIT_FAILURE);
 	}
-
-	if (!write_file(path, val, strlen(val) + 1)) {
-		perror(path);
-		exit(EXIT_FAILURE);
-	}
+	write_file(path, val, strlen(val) + 1);
 }
 
 unsigned long thp_read_num(const char *name)
tools/testing/selftests/mm/thp_settings.h +0 -1
index 76eeb712e5f1..7748a9009191 100644
@@ -63,7 +63,6 @@ struct thp_settings {
 };
 
 int read_file(const char *path, char *buf, size_t buflen);
-int write_file(const char *path, const char *buf, size_t buflen);
 unsigned long read_num(const char *path);
 void write_num(const char *path, unsigned long num);
 
tools/testing/selftests/mm/vm_util.c +15 -0
index a6d4ff7dfdc0..ad96d19d1b85 100644
@@ -764,3 +764,18 @@ int unpoison_memory(unsigned long pfn)
 
 	return ret > 0 ? 0 : -errno;
 }
+
+void write_file(const char *path, const char *buf, size_t buflen)
+{
+	int fd;
+	ssize_t numwritten;
+
+	fd = open(path, O_WRONLY);
+	if (fd == -1)
+		ksft_exit_fail_msg("%s open failed: %s\n", path, strerror(errno));
+
+	numwritten = write(fd, buf, buflen - 1);
+	close(fd);
+	if (numwritten < 1)
+		ksft_exit_fail_msg("Write failed\n");
+}
tools/testing/selftests/mm/vm_util.h +2 -0
index e9c4e24769c1..1a07305ceff4 100644
@@ -166,3 +166,5 @@ int unpoison_memory(unsigned long pfn);
 
 #define PAGEMAP_PRESENT(ent)	(((ent) & (1ull << 63)) != 0)
 #define PAGEMAP_PFN(ent)	((ent) & ((1ull << 55) - 1))
+
+void write_file(const char *path, const char *buf, size_t buflen);
Signature · 1 line
2.53.0
reviewer Acked-by on v4 3/5
#7 lore raw

Re: [PATCH v4 3/5] selftests/mm: move write_file helper to vm_util

Quoted 19 lines · Chunyu Hu

thp_settings provides write_file() helper for safely writing to a file and exit when write failure happens. It's a very low level helper and many sub tests need such a helper, not only thp tests.

split_huge_page_test also defines a write_file locally. The two have minior differences in return type and used exit api. And there would be conflicts if split_huge_page_test wanted to include thp_settings.h because of different prototype, making it less convenient.

It's possisble to merge the two, although some tests don't use the kselftest infrastrucutre for testing. It would also work when using the ksft_exit_msg() to exit in my test, as the counters are all zero. Output will be like:

TAP version 13
1..62
Bail out! /proc/sys/vm/drop_caches1 open failed: No such file or directory # Totals: pass:0 fail:0 xfail:0 xpass:0 skip:0 error:0

Quoted 8 lines

So here we just keep the version in split_huge_page_test, and move it into the vm_util. This makes it easy to maitain and user could just include one vm_util.h when they don't need thp setting helpers. Keep the prototype of void return as the function will exit on any error, return value is not necessary, and will simply the callers like write_num() and write_string().

Quoted 154 lines

---
Changes in v3:
- make the write_file void return
Changes in v2:
new patch from v2

---
 .../selftests/mm/split_huge_page_test.c       | 15 --------
 tools/testing/selftests/mm/thp_settings.c     | 35 ++-----------------
 tools/testing/selftests/mm/thp_settings.h     |  1 -
 tools/testing/selftests/mm/vm_util.c          | 15 ++++++++
 tools/testing/selftests/mm/vm_util.h          |  2 ++
 5 files changed, 20 insertions(+), 48 deletions(-)
tools/testing/selftests/mm/split_huge_page_test.c +0 -15
index e0167111bdd1..93f205327b84 100644
@@ -255,21 +255,6 @@ static int check_after_split_folio_orders(char *vaddr_start, size_t len,
 	return status;
 }
 
-static void write_file(const char *path, const char *buf, size_t buflen)
-{
-	int fd;
-	ssize_t numwritten;
-
-	fd = open(path, O_WRONLY);
-	if (fd == -1)
-		ksft_exit_fail_msg("%s open failed: %s\n", path, strerror(errno));
-
-	numwritten = write(fd, buf, buflen - 1);
-	close(fd);
-	if (numwritten < 1)
-		ksft_exit_fail_msg("Write failed\n");
-}
-
 static void write_debugfs(const char *fmt, ...)
 {
 	char input[INPUT_MAX];
tools/testing/selftests/mm/thp_settings.c +3 -32
index 574bd0f8ae48..e748ebfb3d4e 100644
@@ -6,6 +6,7 @@
 #include <string.h>
 #include <unistd.h>
 
+#include "vm_util.h"
 #include "thp_settings.h"
 
 #define THP_SYSFS "/sys/kernel/mm/transparent_hugepage/"
@@ -64,29 +65,6 @@ int read_file(const char *path, char *buf, size_t buflen)
 	return (unsigned int) numread;
 }
 
-int write_file(const char *path, const char *buf, size_t buflen)
-{
-	int fd;
-	ssize_t numwritten;
-
-	fd = open(path, O_WRONLY);
-	if (fd == -1) {
-		printf("open(%s)\n", path);
-		exit(EXIT_FAILURE);
-		return 0;
-	}
-
-	numwritten = write(fd, buf, buflen - 1);
-	close(fd);
-	if (numwritten < 1) {
-		printf("write(%s)\n", buf);
-		exit(EXIT_FAILURE);
-		return 0;
-	}
-
-	return (unsigned int) numwritten;
-}
-
 unsigned long read_num(const char *path)
 {
 	char buf[21];
@@ -104,10 +82,7 @@ void write_num(const char *path, unsigned long num)
 	char buf[21];
 
 	sprintf(buf, "%ld", num);
-	if (!write_file(path, buf, strlen(buf) + 1)) {
-		perror(path);
-		exit(EXIT_FAILURE);
-	}
+	write_file(path, buf, strlen(buf) + 1);
 }
 
 int thp_read_string(const char *name, const char * const strings[])
@@ -165,11 +140,7 @@ void thp_write_string(const char *name, const char *val)
 		printf("%s: Pathname is too long\n", __func__);
 		exit(EXIT_FAILURE);
 	}
-
-	if (!write_file(path, val, strlen(val) + 1)) {
-		perror(path);
-		exit(EXIT_FAILURE);
-	}
+	write_file(path, val, strlen(val) + 1);
 }
 
 unsigned long thp_read_num(const char *name)
tools/testing/selftests/mm/thp_settings.h +0 -1
index 76eeb712e5f1..7748a9009191 100644
@@ -63,7 +63,6 @@ struct thp_settings {
 };
 
 int read_file(const char *path, char *buf, size_t buflen);
-int write_file(const char *path, const char *buf, size_t buflen);
 unsigned long read_num(const char *path);
 void write_num(const char *path, unsigned long num);
 
tools/testing/selftests/mm/vm_util.c +15 -0
index a6d4ff7dfdc0..ad96d19d1b85 100644
@@ -764,3 +764,18 @@ int unpoison_memory(unsigned long pfn)
 
 	return ret > 0 ? 0 : -errno;
 }
+
+void write_file(const char *path, const char *buf, size_t buflen)
+{
+	int fd;
+	ssize_t numwritten;
+
+	fd = open(path, O_WRONLY);
+	if (fd == -1)
+		ksft_exit_fail_msg("%s open failed: %s\n", path, strerror(errno));
+
+	numwritten = write(fd, buf, buflen - 1);
+	close(fd);
+	if (numwritten < 1)
+		ksft_exit_fail_msg("Write failed\n");
+}
tools/testing/selftests/mm/vm_util.h +2 -0
index e9c4e24769c1..1a07305ceff4 100644
@@ -166,3 +166,5 @@ int unpoison_memory(unsigned long pfn);
 
 #define PAGEMAP_PRESENT(ent)	(((ent) & (1ull << 63)) != 0)
 #define PAGEMAP_PFN(ent)	((ent) & ((1ull << 55) - 1))
+
+void write_file(const char *path, const char *buf, size_t buflen);
Signature · 1 line
2.53.0
Signature · 2 lines
Sincerely yours,
Mike.
reviewer Reviewed-by on v4 3/5
#8 lore raw

Re: [PATCH v4 3/5] selftests/mm: move write_file helper to vm_util

In reply to Chunyu Hu
Quoted 32 lines · Chunyu Hu

thp_settings provides write_file() helper for safely writing to a file and exit when write failure happens. It's a very low level helper and many sub tests need such a helper, not only thp tests.

split_huge_page_test also defines a write_file locally. The two have minior differences in return type and used exit api. And there would be conflicts if split_huge_page_test wanted to include thp_settings.h because of different prototype, making it less convenient.

It's possisble to merge the two, although some tests don't use the kselftest infrastrucutre for testing. It would also work when using the ksft_exit_msg() to exit in my test, as the counters are all zero. Output will be like:

TAP version 13
1..62
Bail out! /proc/sys/vm/drop_caches1 open failed: No such file or directory # Totals: pass:0 fail:0 xfail:0 xpass:0 skip:0 error:0

So here we just keep the version in split_huge_page_test, and move it into the vm_util. This makes it easy to maitain and user could just include one vm_util.h when they don't need thp setting helpers. Keep the prototype of void return as the function will exit on any error, return value is not necessary, and will simply the callers like write_num() and write_string().

---
Changes in v3:
- make the write_file void return
Changes in v2:
new patch from v2

Quoted 8 lines
---
 .../selftests/mm/split_huge_page_test.c       | 15 --------
 tools/testing/selftests/mm/thp_settings.c     | 35 ++-----------------
 tools/testing/selftests/mm/thp_settings.h     |  1 -
 tools/testing/selftests/mm/vm_util.c          | 15 ++++++++
 tools/testing/selftests/mm/vm_util.h          |  2 ++
 5 files changed, 20 insertions(+), 48 deletions(-)

LGTM.

Best Regards,
Yan, Zi

maintainer Acked-by on v4 3/5
#9 lore raw

Re: [PATCH v4 3/5] selftests/mm: move write_file helper to vm_util

In reply to Chunyu Hu
Quoted 5 lines
tools/testing/selftests/mm/vm_util.c +4 -0
+
+void write_file(const char *path, const char *buf, size_t buflen)
+{
+	int fd;

tools/testing/selftests/mm/vm_util.c int unpoison_memory(unsigned long pfn) · line 780

Quoted 10 lines
tools/testing/selftests/mm/vm_util.c +10 -0
+	ssize_t numwritten;
+
+	fd = open(path, O_WRONLY);
+	if (fd == -1)
+		ksft_exit_fail_msg("%s open failed: %s\n", path, strerror(errno));
+
+	numwritten = write(fd, buf, buflen - 1);
+	close(fd);
+	if (numwritten < 1)
+		ksft_exit_fail_msg("Write failed\n");

I guess we could indeed test here easily for

	if (numwritten != buflen - 1)

But then, we don't really expect only partial writes to succeed here.

Signature · 3 lines
Cheers,

David
author
#10 lore raw

Re: [PATCH v4 3/5] selftests/mm: move write_file helper to vm_util

Quoted 17 lines · David Hildenbrand (Arm)
Quoted 5 lines
tools/testing/selftests/mm/vm_util.c +4 -0
+
+void write_file(const char *path, const char *buf, size_t buflen)
+{
+	int fd;

tools/testing/selftests/mm/vm_util.c int unpoison_memory(unsigned long pfn) · line 780

Quoted 10 lines
tools/testing/selftests/mm/vm_util.c +10 -0
+	ssize_t numwritten;
+
+	fd = open(path, O_WRONLY);
+	if (fd == -1)
+		ksft_exit_fail_msg("%s open failed: %s\n", path, strerror(errno));
+
+	numwritten = write(fd, buf, buflen - 1);
+	close(fd);
+	if (numwritten < 1)
+		ksft_exit_fail_msg("Write failed\n");

I guess we could indeed test here easily for

Quoted 1 line
	if (numwritten != buflen - 1)

Looks like you are right. AI provided other questions on this helper, maybe I can append a patch in the series to do such robust improvement.

Quoted 10 lines

But then, we don't really expect only partial writes to succeed here.

Signature · 3 lines
Cheers,

David
author
#11 lore raw

PATCH v4 4/5 selftests/mm: split_huge_page_test: skip the test when thp is not available

In reply to Chunyu Hu

When thp is not enabled on some kernel config such as realtime kernel, the test will report failure. Fix the false positive by skipping the test directly when thp is not enabled.

Tested with thp disabled kernel:
Before The fix:
# --------------------------------------------------

  # running ./split_huge_page_test /tmp/xfs_dir_Ywup9p
  # --------------------------------------------------
  # TAP version 13
  # Bail out! Reading PMD pagesize failed
  # # Totals: pass:0 fail:0 xfail:0 xpass:0 skip:0 error:0
  # [FAIL]
  not ok 61 split_huge_page_test /tmp/xfs_dir_Ywup9p # exit=1

After the fix:
# --------------------------------------------------

  # running ./split_huge_page_test /tmp/xfs_dir_YHPUPl
  # --------------------------------------------------
  # TAP version 13
  # 1..0 # SKIP Transparent Hugepages not available
  # [SKIP]
  ok 6 split_huge_page_test /tmp/xfs_dir_YHPUPl # SKIP

---
V3 chagnes:
- add acked-by from David
- add reviewed by from Mike
V2 changes:
- removed the {} in if block
- removed the write_file helper rename chunk, as there's a new patch 3 to resolve the conflict with thp_settings.h by moving it to vm_util.
- Add reviewed by from Zi and Lorenzo

---
 tools/testing/selftests/mm/split_huge_page_test.c | 4 ++++
 1 file changed, 4 insertions(+)
tools/testing/selftests/mm/split_huge_page_test.c +4 -0
index 93f205327b84..500d07c4938b 100644
@@ -21,6 +21,7 @@
 #include <time.h>
 #include "vm_util.h"
 #include "kselftest.h"
+#include "thp_settings.h"
 
 uint64_t pagesize;
 unsigned int pageshift;
@@ -757,6 +758,9 @@ int main(int argc, char **argv)
 		ksft_finished();
 	}
 
+	if (!thp_is_enabled())
+		ksft_exit_skip("Transparent Hugepages not available\n");
+
 	if (argc > 1)
 		optional_xfs_path = argv[1];
 
Signature · 1 line
2.53.0
author
#12 lore raw

PATCH v4 5/5 selftests/mm: transhuge_stress: skip the test when thp not available

In reply to Chunyu Hu

The test requires thp, skip the test when thp is not available to avoid false positive.

Tested with thp disabled kernel.
Before the fix:
# --------------------------------

  # running ./transhuge-stress -d 20
  # --------------------------------
  # TAP version 13
  # 1..1
  # transhuge-stress: allocate 1453 transhuge pages, using 2907 MiB virtual memory and 11 MiB of ram
  # Bail out! MADV_HUGEPAGE# Planned tests != run tests (1 != 0)
  # # Totals: pass:0 fail:0 xfail:0 xpass:0 skip:0 error:0
  # [FAIL]
  not ok 60 transhuge-stress -d 20 # exit=1

After the fix:
# --------------------------------

  # running ./transhuge-stress -d 20
  # --------------------------------
  # TAP version 13
  # 1..0 # SKIP Transparent Hugepages not available
  # [SKIP]
  ok 5 transhuge-stress -d 20 # SKIP

---
V3 changes:
- Add review by from Mike
- Add acked-by from David
V2 changes:
- Move the exit chunk to the front of ksft_print_headers() as suggested by David.
- Add reviewed by from Zi and Lorenzo

---
 tools/testing/selftests/mm/transhuge-stress.c | 4 ++++
 1 file changed, 4 insertions(+)
tools/testing/selftests/mm/transhuge-stress.c +4 -0
index bcad47c09518..7a9f1035099b 100644
@@ -17,6 +17,7 @@
 #include <sys/mman.h>
 #include "vm_util.h"
 #include "kselftest.h"
+#include "thp_settings.h"
 
 int backing_fd = -1;
 int mmap_flags = MAP_ANONYMOUS | MAP_NORESERVE | MAP_PRIVATE;
@@ -37,6 +38,9 @@ int main(int argc, char **argv)
 
 	ksft_print_header();
 
+	if (!thp_is_enabled())
+		ksft_exit_skip("Transparent Hugepages not available\n");
+
 	ram = sysconf(_SC_PHYS_PAGES);
 	if (ram > SIZE_MAX / psize() / 4)
 		ram = SIZE_MAX / 4;
Signature · 1 line
2.53.0
author
#14 lore raw
Quoted 9 lines · Andrew Morton
Quoted 3 lines · On Fri, 20 Mar 2026 00:06:51 +0800 Chunyu Hu <[email protected]> wrote:

There are several tests requires transprarent hugepages, when run on thp disabled kernel such as realtime kernel, there will be false negative. Mark those tests as skip when thp is not available.

A couple of questions from Sashiko:

	https://sashiko.dev/#/patchset/20260319160656.1480279-1-chuhu%40redhat.com

AI question one:

```
> +void write_file(const char *path, const char *buf, size_t buflen)
> +{
> +	int fd;
> +	ssize_t numwritten;
> +
> +	fd = open(path, O_WRONLY);
> +	if (fd == -1)
> +		ksft_exit_fail_msg("%s open failed: %s\n", path, strerror(errno));
> +
> +	numwritten = write(fd, buf, buflen - 1);
Is it safe to keep the buflen - 1 behavior in a shared generic utility?
Since this is now a generic API in vm_util, callers passing the exact buffer
size (like sizeof(buf) or strlen(buf)) will suffer a 1-byte data truncation.
Additionally, if a caller passes buflen = 0, buflen - 1 will underflow,
causing write() to attempt writing an invalid size.
```

Here I think there's no single anwser. We can use buflen, but if user don't provide a '\0' in the end of buffer, out of boundary access could happen when such as printf(). We called it buflen, then we expect it's a '\0' ended buffer as I undersand. So maye we can just leave as it is and user should be awared of the requiremnt of the helper here.

AI question two:

```
> +	close(fd);
> +	if (numwritten < 1)
> +		ksft_exit_fail_msg("Write failed\n");

Does this drop useful debugging context compared to the previous version
in thp_settings.c?
The old implementation printed the buffer contents on failure. With this
generic message, if a test fails to write, it will only print "Write failed"
without indicating which file path failed, what data was being written, or
the underlying errno.
```

That makes sense, indeed we dropped the printing of the buffer content compared with the version in the thp_settings.c. I can append a new patch in v5 to improve this, together with other safety checks.