[PATCH v3 0/3] selftests/mm: avoid false failures in hugetlb and KSM tests

[PATCH v3 0/3] selftests/mm: avoid false failures in hugetlb and KSM tests

[Sayali Patil]

Hi all,

This series fixes issues in the hugetlb and KSM MM selftest categories
that can report failures when the prerequisites for the tests are not
satisfied.

Patch 1 updates the hugetlb selftest helpers to handle -EINVAL when
attempting to configure gigantic HugeTLB pages via nr_hugepages.
PowerPC hash MMU pSeries systems expose gigantic hugepage sizes
but do not allow runtime allocation of such pages,
causing the sysfs write to fail. Handle this case gracefully and
continue running the test instead of aborting.

Patch 2 fixes the KSM NUMA merge test on systems with memoryless NUMA
nodes. The test currently relies on the number of configured NUMA nodes
and may attempt allocations on nodes that have no memory, resulting in
spurious failures. Use the existing helpers to identify
NUMA nodes that contain memory and skip the test when fewer than two
such nodes are available.

Patch 3 fixes a pre-existing operator precedence issue in ksm_tests,
where a ternary expression combined with logical OR operators could be
evaluated differently than intended. Added parentheses to ensure the
correct evaluation order.

These changes improve handling of unsupported test configurations and
unmet test prerequisites, avoiding spurious failures.

Thanks,
Sayali
---

V2 -> V3:
- For "selftests/mm: handle EINVAL when configuring gigantic hugepages":
  Reworked the implementation to use internal helper functions with an 
  ignore_einval flag, as suggested by David Hildenbrand.
  Added __write_file() and __write_num() helpers to optionally ignore
  -EINVAL from sysfs writes.
  Added write_num_ignore_einval() wrapper for callers that need to tolerate
  unsupported hugepage configurations.

V2: https://lore.kernel.org/all/cover.1782811071.git.sayalip@linux.ibm.com/

---

V1 -> V2:
- For "selftests/mm: handle EINVAL when configuring gigantic hugepages":
  Added a dedicated hugetlb_write_num() helper to handle the expected EINVAL
  returned by gigantic hugepage configuration, in the
  hugepage setup code rather than modifying the generic write_file() helper.
- For "selftests/mm: fix ksm NUMA merge test for systems with memoryless NUMA nodes":
  Updated implementation as per David's review comment to build upon
  existing helpers get_next_mem_node().
- Added new patch "selftests/mm: fix ternary operator precedence in ksm_tests"
  to address a pre-existing issue identified during Sahiko's review.

V1: https://lore.kernel.org/all/cover.1782365671.git.sayalip@linux.ibm.com/

---

Sayali Patil (3):
  selftests/mm: handle EINVAL when configuring gigantic hugepages
  selftests/mm: fix ksm NUMA merge test for systems with memoryless NUMA
    nodes
  selftests/mm: fix ternary operator precedence in ksm_tests

 .../testing/selftests/mm/hugepage_settings.c  |  2 +-
 tools/testing/selftests/mm/ksm_tests.c        | 28 ++++++++++---------
 tools/testing/selftests/mm/vm_util.c          | 26 ++++++++++++++---
 tools/testing/selftests/mm/vm_util.h          |  1 +
 4 files changed, 39 insertions(+), 18 deletions(-)

-- 
2.52.0

[PATCH v3 1/3] selftests/mm: handle EINVAL when configuring gigantic hugepages

[Sayali Patil]

Some MM selftests attempt to configure the amount of
HugeTLB pages of different sizes by writing to nr_hugepages.

PowerPC hash MMU pSeries systems advertise gigantic hugepage sizes
but do not support runtime allocation of such pages, writes
to the corresponding nr_hugepages file fail with -EINVAL.
This causes the test to bail out even though the failure is due
to a platform limitation rather than the
functionality being tested.

Ignore -EINVAL when configuring nr_hugepages so that tests continue to
run on systems where gigantic hugepage allocation is unsupported.

Before patch:
   -------------------------
   running ./hugetlb-madvise
   -------------------------
   TAP version 13
   1..1
     [INFO] detected hugetlb page size: 16777216 KiB
     [INFO] detected hugetlb page size: 16384 KiB
    ok 1 MADV_DONTNEED and MADV_REMOVE on hugetlb
    Totals: pass:1 fail:0 xfail:0 xpass:0 skip:0 error:0
    Bail out! /sys/kernel/mm/hugepages/hugepages-16777216kB/nr_hugepages
    write(0) failed: Invalid argument
    Totals: pass:0 fail:0 xfail:0 xpass:0 skip:0 error:0
    [FAIL]

After patch:
   -------------------------
   running ./hugetlb-madvise
   -------------------------
   TAP version 13
   1..1
    [INFO] detected hugetlb page size: 16777216 KiB
    [INFO] detected hugetlb page size: 16384 KiB
   ok 1 MADV_DONTNEED and MADV_REMOVE on hugetlb
   Totals: pass:1 fail:0 xfail:0 xpass:0 skip:0 error:0
   [PASS]

Fixes: 27477b28b74f ("selftests/mm: hugepage_settings: add APIs to get and set nr_hugepages")
Co-developed-by: David Hildenbrand (Arm) <david@kernel.org>
Signed-off-by: David Hildenbrand (Arm) <david@kernel.org>
Signed-off-by: Sayali Patil <sayalip@linux.ibm.com>
---
 .../testing/selftests/mm/hugepage_settings.c  |  2 +-
 tools/testing/selftests/mm/vm_util.c          | 26 ++++++++++++++++---
 tools/testing/selftests/mm/vm_util.h          |  1 +
 3 files changed, 24 insertions(+), 5 deletions(-)

diff --git a/tools/testing/selftests/mm/hugepage_settings.c b/tools/testing/selftests/mm/hugepage_settings.c
index 2eab2110ac6a..d7917dce3aba 100644
--- a/tools/testing/selftests/mm/hugepage_settings.c
+++ b/tools/testing/selftests/mm/hugepage_settings.c
@@ -437,7 +437,7 @@ void hugetlb_set_nr_pages(unsigned long size, unsigned long nr)
 
 	hugetlb_sysfs_path(path, sizeof(path), size, "nr_hugepages");
 
-	write_num(path, nr);
+	write_num_ignore_einval(path, nr);
 }
 
 unsigned long hugetlb_free_pages(unsigned long size)
diff --git a/tools/testing/selftests/mm/vm_util.c b/tools/testing/selftests/mm/vm_util.c
index 311fc5b4513e..ef1ea11981a7 100644
--- a/tools/testing/selftests/mm/vm_util.c
+++ b/tools/testing/selftests/mm/vm_util.c
@@ -719,7 +719,7 @@ int read_file(const char *path, char *buf, size_t buflen)
 	return (unsigned int) numread;
 }
 
-void write_file(const char *path, const char *buf, size_t buflen)
+static void __write_file(const char *path, const char *buf, size_t buflen, bool ignore_einval)
 {
 	int fd, saved_errno;
 	ssize_t numwritten;
@@ -735,14 +735,22 @@ void write_file(const char *path, const char *buf, size_t buflen)
 	saved_errno = errno;
 	close(fd);
 	errno = saved_errno;
-	if (numwritten < 0)
+	if (numwritten < 0) {
+		if (ignore_einval && errno == EINVAL)
+			return;
 		ksft_exit_fail_msg("%s write(%.*s) failed: %s\n", path, (int)(buflen - 1),
 				buf, strerror(errno));
+	}
 	if (numwritten != buflen - 1)
 		ksft_exit_fail_msg("%s write(%.*s) is truncated, expected %zu bytes, got %zd bytes\n",
 				path, (int)(buflen - 1), buf, buflen - 1, numwritten);
 }
 
+void write_file(const char *path, const char *buf, size_t buflen)
+{
+	__write_file(path, buf, buflen, /* ignore_einval = */ false);
+}
+
 unsigned long read_num(const char *path)
 {
 	char buf[21];
@@ -753,12 +761,22 @@ unsigned long read_num(const char *path)
 	return strtoul(buf, NULL, 10);
 }
 
-void write_num(const char *path, unsigned long num)
+static void __write_num(const char *path, unsigned long num, bool ignore_einval)
 {
 	char buf[21];
 
 	sprintf(buf, "%lu", num);
-	write_file(path, buf, strlen(buf) + 1);
+	__write_file(path, buf, strlen(buf) + 1, ignore_einval);
+}
+
+void write_num(const char *path, unsigned long num)
+{
+	return __write_num(path, num, /* ignore_einval = */ false);
+}
+
+void write_num_ignore_einval(const char *path, unsigned long num)
+{
+	return __write_num(path, num, /* ignore_einval = */ true);
 }
 
 static unsigned long shmall, shmmax;
diff --git a/tools/testing/selftests/mm/vm_util.h b/tools/testing/selftests/mm/vm_util.h
index ea8fc8fdf0eb..7799154b67ee 100644
--- a/tools/testing/selftests/mm/vm_util.h
+++ b/tools/testing/selftests/mm/vm_util.h
@@ -168,6 +168,7 @@ void write_file(const char *path, const char *buf, size_t buflen);
 int read_file(const char *path, char *buf, size_t buflen);
 unsigned long read_num(const char *path);
 void write_num(const char *path, unsigned long num);
+void write_num_ignore_einval(const char *path, unsigned long num);
 
 void shm_limits_prepare(unsigned long length);
 void __shm_limits_restore(void);
-- 
2.52.0

[PATCH v3 2/3] selftests/mm: fix ksm NUMA merge test for systems with memoryless NUMA nodes

[Sayali Patil]

The KSM NUMA merge test allocates identical pages on different NUMA
nodes and verifies KSM behavior with merge_across_nodes enabled and
disabled.

On systems with memoryless NUMA nodes, for example:
 #numactl  -H
      available: 2 nodes (0,4)
      .....
      node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
      node 0 size: 14825 MB
      node 0 free: 1382 MB
      node 4 cpus:
      node 4 size: 0 MB
      node 4 free: 0 MB

the test may attempt to allocate memory on a node without memory,
causing numa_alloc_onnode() to fail and resulting in a spurious test
failure.

The test currently checks numa_num_configured_nodes() to determine
whether sufficient NUMA nodes are available. However, configured nodes
do not necessarily have memory.

Reuse the existing get_first_mem_node() and get_next_mem_node()
helpers to locate NUMA nodes that actually contain memory, and skip
the test when fewer than two such nodes are available.

Before patch:
       ---------------------------
	running ./ksm_tests -N -m 1
       ---------------------------
        mbind: Invalid argument
        ok 1 KSM NUMA merging
	Totals: pass:1 fail:0 xfail:0 xpass:0 skip:0 error:0
        [PASS]
       ok 1 ksm_tests -N -m 1
       ---------------------------
        running ./ksm_tests -N -m 0
       ---------------------------
        mbind: Invalid argument
        not ok 1 KSM NUMA merging
	Totals: pass:0 fail:1 xfail:0 xpass:0 skip:0 error:0
        [FAIL]
       not ok 2 ksm_tests -N -m 0 # exit=1

After patch:
       ---------------------------
        running ./ksm_tests -N -m 1
       ---------------------------
        At least 2 NUMA nodes with memory must be available
	ok 1
	SKIP KSM NUMA merging
	Totals: pass:0 fail:0 xfail:0 xpass:0 skip:1 error:0
        [PASS]
        ok 1 ksm_tests -N -m 1
       ---------------------------
        running ./ksm_tests -N -m 0
       ---------------------------
        At least 2 NUMA nodes with memory must be available
	ok 1
	SKIP KSM NUMA merging
	Totals: pass:0 fail:0 xfail:0 xpass:0 skip:1 error:0
        [PASS]
        ok 2 ksm_tests -N -m 0

Fixes: e3820ab252dd ("selftest/vm: fix ksm selftest to run with different NUMA topologies")
Co-developed-by: David Hildenbrand (Arm) <david@kernel.org>
Signed-off-by: David Hildenbrand (Arm) <david@kernel.org>
Signed-off-by: Sayali Patil <sayalip@linux.ibm.com>
---
 tools/testing/selftests/mm/ksm_tests.c | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

diff --git a/tools/testing/selftests/mm/ksm_tests.c b/tools/testing/selftests/mm/ksm_tests.c
index a050f4840cfa..2ebbb544c671 100644
--- a/tools/testing/selftests/mm/ksm_tests.c
+++ b/tools/testing/selftests/mm/ksm_tests.c
@@ -440,9 +440,9 @@ static int get_next_mem_node(int node)
 		mem_node = i % (max_node + 1);
 		node_size = numa_node_size(mem_node, NULL);
 		if (node_size > 0)
-			break;
+			return mem_node;
 	}
-	return mem_node;
+	return -ENODEV;
 }
 
 static int get_first_mem_node(void)
@@ -455,8 +455,8 @@ static int check_ksm_numa_merge(int merge_type, int mapping, int prot, int timeo
 {
 	void *numa1_map_ptr, *numa2_map_ptr;
 	struct timespec start_time;
+	int first_node, second_node;
 	int page_count = 2;
-	int first_node;
 
 	if (clock_gettime(CLOCK_MONOTONIC_RAW, &start_time)) {
 		ksft_perror("clock_gettime");
@@ -467,17 +467,19 @@ static int check_ksm_numa_merge(int merge_type, int mapping, int prot, int timeo
 		ksft_print_msg("NUMA support not enabled\n");
 		return KSFT_SKIP;
 	}
-	if (numa_num_configured_nodes() <= 1) {
-		ksft_print_msg("At least 2 NUMA nodes must be available\n");
+	first_node = get_first_mem_node();
+	second_node = get_next_mem_node(first_node);
+
+	if (second_node < 0) {
+		ksft_print_msg("At least 2 NUMA nodes with memory must be available\n");
 		return KSFT_SKIP;
 	}
 	if (ksm_write_sysfs(KSM_FP("merge_across_nodes"), merge_across_nodes))
 		return KSFT_FAIL;
 
 	/* allocate 2 pages in 2 different NUMA nodes and fill them with the same data */
-	first_node = get_first_mem_node();
 	numa1_map_ptr = numa_alloc_onnode(page_size, first_node);
-	numa2_map_ptr = numa_alloc_onnode(page_size, get_next_mem_node(first_node));
+	numa2_map_ptr = numa_alloc_onnode(page_size, second_node);
 	if (!numa1_map_ptr || !numa2_map_ptr) {
 		ksft_perror("numa_alloc_onnode");
 		return KSFT_FAIL;
-- 
2.52.0

[PATCH v3 3/3] selftests/mm: fix ternary operator precedence in ksm_tests

[Sayali Patil]

The KSM selftest uses conditional expressions to skip accesses to
merge_across_nodes on systems without NUMA support. However, the
ternary operator is combined with logical OR without parentheses:

a || numa_available() ? 0 : b || c

Due to operator precedence rules, this is parsed as:

(a || numa_available()) ? 0 : (b || c)

instead of the intended:

a || (numa_available() ? 0 : b) || c

Add parentheses around the conditional expressions to ensure the
correct evaluation order.

Fixes: 9aa1af954db0 ("selftests: vm: check numa_available() before operating "merge_across_nodes" in ksm_tests")
Acked-by: David Hildenbrand (Arm) <david@kernel.org>
Signed-off-by: Sayali Patil <sayalip@linux.ibm.com>
---
 tools/testing/selftests/mm/ksm_tests.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/tools/testing/selftests/mm/ksm_tests.c b/tools/testing/selftests/mm/ksm_tests.c
index 2ebbb544c671..5fd7792a0d47 100644
--- a/tools/testing/selftests/mm/ksm_tests.c
+++ b/tools/testing/selftests/mm/ksm_tests.c
@@ -288,8 +288,8 @@ static bool assert_ksm_pages_count(long dupl_page_count)
 static int ksm_save_def(struct ksm_sysfs *ksm_sysfs)
 {
 	if (ksm_read_sysfs(KSM_FP("max_page_sharing"), &ksm_sysfs->max_page_sharing) ||
-	    numa_available() ? 0 :
-		ksm_read_sysfs(KSM_FP("merge_across_nodes"), &ksm_sysfs->merge_across_nodes) ||
+	    (numa_available() ? 0 :
+		ksm_read_sysfs(KSM_FP("merge_across_nodes"), &ksm_sysfs->merge_across_nodes)) ||
 	    ksm_read_sysfs(KSM_FP("sleep_millisecs"), &ksm_sysfs->sleep_millisecs) ||
 	    ksm_read_sysfs(KSM_FP("pages_to_scan"), &ksm_sysfs->pages_to_scan) ||
 	    ksm_read_sysfs(KSM_FP("run"), &ksm_sysfs->run) ||
@@ -304,8 +304,8 @@ static int ksm_save_def(struct ksm_sysfs *ksm_sysfs)
 static int ksm_restore(struct ksm_sysfs *ksm_sysfs)
 {
 	if (ksm_write_sysfs(KSM_FP("max_page_sharing"), ksm_sysfs->max_page_sharing) ||
-	    numa_available() ? 0 :
-		ksm_write_sysfs(KSM_FP("merge_across_nodes"), ksm_sysfs->merge_across_nodes) ||
+	    (numa_available() ? 0 :
+		ksm_write_sysfs(KSM_FP("merge_across_nodes"), ksm_sysfs->merge_across_nodes)) ||
 	    ksm_write_sysfs(KSM_FP("pages_to_scan"), ksm_sysfs->pages_to_scan) ||
 	    ksm_write_sysfs(KSM_FP("run"), ksm_sysfs->run) ||
 	    ksm_write_sysfs(KSM_FP("sleep_millisecs"), ksm_sysfs->sleep_millisecs) ||
@@ -846,8 +846,8 @@ int main(int argc, char *argv[])
 
 	if (ksm_write_sysfs(KSM_FP("run"), 2) ||
 	    ksm_write_sysfs(KSM_FP("sleep_millisecs"), 0) ||
-	    numa_available() ? 0 :
-		ksm_write_sysfs(KSM_FP("merge_across_nodes"), 1) ||
+	    (numa_available() ? 0 :
+		ksm_write_sysfs(KSM_FP("merge_across_nodes"), 1)) ||
 	    ksm_write_sysfs(KSM_FP("pages_to_scan"), page_count))
 		ksft_exit_fail_msg("Cannot set up KSM tunables\n");
 
-- 
2.52.0

Re: [PATCH v3 1/3] selftests/mm: handle EINVAL when configuring gigantic hugepages

[David Hildenbrand (Arm)]

On 7/8/26 08:59, Sayali Patil wrote:
> Some MM selftests attempt to configure the amount of
> HugeTLB pages of different sizes by writing to nr_hugepages.
> 
> PowerPC hash MMU pSeries systems advertise gigantic hugepage sizes
> but do not support runtime allocation of such pages, writes
> to the corresponding nr_hugepages file fail with -EINVAL.
> This causes the test to bail out even though the failure is due
> to a platform limitation rather than the
> functionality being tested.
> 
> Ignore -EINVAL when configuring nr_hugepages so that tests continue to
> run on systems where gigantic hugepage allocation is unsupported.
> 
> Before patch:
>    -------------------------
>    running ./hugetlb-madvise
>    -------------------------
>    TAP version 13
>    1..1
>      [INFO] detected hugetlb page size: 16777216 KiB
>      [INFO] detected hugetlb page size: 16384 KiB
>     ok 1 MADV_DONTNEED and MADV_REMOVE on hugetlb
>     Totals: pass:1 fail:0 xfail:0 xpass:0 skip:0 error:0
>     Bail out! /sys/kernel/mm/hugepages/hugepages-16777216kB/nr_hugepages
>     write(0) failed: Invalid argument
>     Totals: pass:0 fail:0 xfail:0 xpass:0 skip:0 error:0
>     [FAIL]
> 
> After patch:
>    -------------------------
>    running ./hugetlb-madvise
>    -------------------------
>    TAP version 13
>    1..1
>     [INFO] detected hugetlb page size: 16777216 KiB
>     [INFO] detected hugetlb page size: 16384 KiB
>    ok 1 MADV_DONTNEED and MADV_REMOVE on hugetlb
>    Totals: pass:1 fail:0 xfail:0 xpass:0 skip:0 error:0
>    [PASS]
> 
> Fixes: 27477b28b74f ("selftests/mm: hugepage_settings: add APIs to get and set nr_hugepages")
> Co-developed-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: Sayali Patil <sayalip@linux.ibm.com>

LGTM. Using proper flags would maybe be nicer, but no need for that for this
corner case for now.

-- 
Cheers,

David

Re: [PATCH v3 1/3] selftests/mm: handle EINVAL when configuring gigantic hugepages

[Usama Arif]

On Wed,  8 Jul 2026 12:29:05 +0530 Sayali Patil <sayalip@linux.ibm.com> wrote:

> Some MM selftests attempt to configure the amount of
> HugeTLB pages of different sizes by writing to nr_hugepages.
> 
> PowerPC hash MMU pSeries systems advertise gigantic hugepage sizes
> but do not support runtime allocation of such pages, writes
> to the corresponding nr_hugepages file fail with -EINVAL.
> This causes the test to bail out even though the failure is due
> to a platform limitation rather than the
> functionality being tested.
> 
> Ignore -EINVAL when configuring nr_hugepages so that tests continue to
> run on systems where gigantic hugepage allocation is unsupported.
> 
> Before patch:
>    -------------------------
>    running ./hugetlb-madvise
>    -------------------------
>    TAP version 13
>    1..1
>      [INFO] detected hugetlb page size: 16777216 KiB
>      [INFO] detected hugetlb page size: 16384 KiB
>     ok 1 MADV_DONTNEED and MADV_REMOVE on hugetlb
>     Totals: pass:1 fail:0 xfail:0 xpass:0 skip:0 error:0
>     Bail out! /sys/kernel/mm/hugepages/hugepages-16777216kB/nr_hugepages
>     write(0) failed: Invalid argument
>     Totals: pass:0 fail:0 xfail:0 xpass:0 skip:0 error:0
>     [FAIL]
> 
> After patch:
>    -------------------------
>    running ./hugetlb-madvise
>    -------------------------
>    TAP version 13
>    1..1
>     [INFO] detected hugetlb page size: 16777216 KiB
>     [INFO] detected hugetlb page size: 16384 KiB
>    ok 1 MADV_DONTNEED and MADV_REMOVE on hugetlb
>    Totals: pass:1 fail:0 xfail:0 xpass:0 skip:0 error:0
>    [PASS]
> 
> Fixes: 27477b28b74f ("selftests/mm: hugepage_settings: add APIs to get and set nr_hugepages")
> Co-developed-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: Sayali Patil <sayalip@linux.ibm.com>
> ---
>  .../testing/selftests/mm/hugepage_settings.c  |  2 +-
>  tools/testing/selftests/mm/vm_util.c          | 26 ++++++++++++++++---
>  tools/testing/selftests/mm/vm_util.h          |  1 +
>  3 files changed, 24 insertions(+), 5 deletions(-)
> 

Acked-by: Usama Arif <usama.arif@linux.dev>
 

Re: [PATCH v3 2/3] selftests/mm: fix ksm NUMA merge test for systems with memoryless NUMA nodes

[Usama Arif]

On Wed,  8 Jul 2026 12:29:06 +0530 Sayali Patil <sayalip@linux.ibm.com> wrote:

> The KSM NUMA merge test allocates identical pages on different NUMA
> nodes and verifies KSM behavior with merge_across_nodes enabled and
> disabled.
> 
> On systems with memoryless NUMA nodes, for example:
>  #numactl  -H
>       available: 2 nodes (0,4)
>       .....
>       node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
>       node 0 size: 14825 MB
>       node 0 free: 1382 MB
>       node 4 cpus:
>       node 4 size: 0 MB
>       node 4 free: 0 MB
> 
> the test may attempt to allocate memory on a node without memory,
> causing numa_alloc_onnode() to fail and resulting in a spurious test
> failure.
> 
> The test currently checks numa_num_configured_nodes() to determine
> whether sufficient NUMA nodes are available. However, configured nodes
> do not necessarily have memory.
> 
> Reuse the existing get_first_mem_node() and get_next_mem_node()
> helpers to locate NUMA nodes that actually contain memory, and skip
> the test when fewer than two such nodes are available.
> 
> Before patch:
>        ---------------------------
> 	running ./ksm_tests -N -m 1
>        ---------------------------
>         mbind: Invalid argument
>         ok 1 KSM NUMA merging
> 	Totals: pass:1 fail:0 xfail:0 xpass:0 skip:0 error:0
>         [PASS]
>        ok 1 ksm_tests -N -m 1
>        ---------------------------
>         running ./ksm_tests -N -m 0
>        ---------------------------
>         mbind: Invalid argument
>         not ok 1 KSM NUMA merging
> 	Totals: pass:0 fail:1 xfail:0 xpass:0 skip:0 error:0
>         [FAIL]
>        not ok 2 ksm_tests -N -m 0 # exit=1
> 
> After patch:
>        ---------------------------
>         running ./ksm_tests -N -m 1
>        ---------------------------
>         At least 2 NUMA nodes with memory must be available
> 	ok 1
> 	SKIP KSM NUMA merging
> 	Totals: pass:0 fail:0 xfail:0 xpass:0 skip:1 error:0
>         [PASS]
>         ok 1 ksm_tests -N -m 1
>        ---------------------------
>         running ./ksm_tests -N -m 0
>        ---------------------------
>         At least 2 NUMA nodes with memory must be available
> 	ok 1
> 	SKIP KSM NUMA merging
> 	Totals: pass:0 fail:0 xfail:0 xpass:0 skip:1 error:0
>         [PASS]
>         ok 2 ksm_tests -N -m 0
> 
> Fixes: e3820ab252dd ("selftest/vm: fix ksm selftest to run with different NUMA topologies")
> Co-developed-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: David Hildenbrand (Arm) <david@kernel.org>
> Signed-off-by: Sayali Patil <sayalip@linux.ibm.com>
> ---
>  tools/testing/selftests/mm/ksm_tests.c | 16 +++++++++-------
>  1 file changed, 9 insertions(+), 7 deletions(-)
> 

Acked-by: Usama Arif <usama.arif@linux.dev> 

Re: [PATCH v3 0/3] selftests/mm: avoid false failures in hugetlb and KSM tests

[Andrew Morton]

On Wed,  8 Jul 2026 12:29:04 +0530 Sayali Patil <sayalip@linux.ibm.com> wrote:

> This series fixes issues in the hugetlb and KSM MM selftest categories
> that can report failures when the prerequisites for the tests are not
> satisfied.

Thanks.  AI review asked one question:
	https://sashiko.dev/#/patchset/cover.1783446924.git.sayalip@linux.ibm.com