[PATCH 0/2] selftests/mm: avoid false failures in hugetlb and KSM tests

[PATCH 0/2] 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. Count only NUMA nodes that contain memory and skip
the test when fewer than two such nodes are available.

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

Thanks,
Sayali

Sayali Patil (2):
  selftests/mm: handle EINVAL when configuring gigantic hugepages
  selftests/mm: fix ksm NUMA merge test for systems with memoryless NUMA
    nodes

 tools/testing/selftests/mm/ksm_tests.c | 26 ++++++++++++++++++--------
 tools/testing/selftests/mm/vm_util.c   |  7 +++++++
 2 files changed, 25 insertions(+), 8 deletions(-)

-- 
2.52.0

[PATCH 1/2] 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.

Treat -EINVAL from the sysfs write as a skipped configuration request
and continue running the test instead of failing.

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
   /sys/kernel/mm/hugepages/hugepages-16777216kB/nr_hugepages
   write(0) failed: Invalid argument
   [PASS]

Fixes: 9d07250ea1eb ("selftests/mm: hugepage_settings: add APIs to get and set nr_hugepages")
Signed-off-by: Sayali Patil <sayalip@linux.ibm.com>
---
 tools/testing/selftests/mm/vm_util.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/tools/testing/selftests/mm/vm_util.c b/tools/testing/selftests/mm/vm_util.c
index 311fc5b4513e..a8f16eef5c7c 100644
--- a/tools/testing/selftests/mm/vm_util.c
+++ b/tools/testing/selftests/mm/vm_util.c
@@ -735,6 +735,13 @@ void write_file(const char *path, const char *buf, size_t buflen)
 	saved_errno = errno;
 	close(fd);
 	errno = saved_errno;
+
+	if (numwritten < 0 && errno == EINVAL) {
+		ksft_print_msg("%s write(%.*s) failed: %s\n", path,
+			       (int)(buflen - 1), buf, strerror(errno));
+		return;
+	}
+
 	if (numwritten < 0)
 		ksft_exit_fail_msg("%s write(%.*s) failed: %s\n", path, (int)(buflen - 1),
 				buf, strerror(errno));
-- 
2.52.0

[PATCH 2/2] 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.

Add count_mem_nodes() helper to count only nodes with memory and use it
instead of numa_num_configured_nodes() to ensure the test runs only on
systems with at least two NUMA nodes that have memory. Skip the test
otherwise.

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

Fixes: e3820ab252dd ("selftest/vm: fix ksm selftest to run with different NUMA topologies")
Signed-off-by: Sayali Patil <sayalip@linux.ibm.com>
---
 tools/testing/selftests/mm/ksm_tests.c | 26 ++++++++++++++++++--------
 1 file changed, 18 insertions(+), 8 deletions(-)

diff --git a/tools/testing/selftests/mm/ksm_tests.c b/tools/testing/selftests/mm/ksm_tests.c
index a050f4840cfa..d1790b4347d3 100644
--- a/tools/testing/selftests/mm/ksm_tests.c
+++ b/tools/testing/selftests/mm/ksm_tests.c
@@ -450,6 +450,18 @@ static int get_first_mem_node(void)
 	return get_next_mem_node(numa_max_node());
 }
 
+static int count_mem_nodes(void)
+{
+	int node, count = 0;
+
+	for (node = 0; node <= numa_max_node(); node++) {
+		if (numa_node_size(node, NULL) > 0)
+			count++;
+	}
+
+	return count;
+}
+
 static int check_ksm_numa_merge(int merge_type, int mapping, int prot, int timeout,
 				bool merge_across_nodes, size_t page_size)
 {
@@ -463,14 +475,12 @@ static int check_ksm_numa_merge(int merge_type, int mapping, int prot, int timeo
 		return KSFT_FAIL;
 	}
 
-	if (numa_available() < 0) {
-		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");
-		return KSFT_SKIP;
-	}
+	if (numa_available() < 0)
+		ksft_exit_skip("NUMA support not enabled\n");
+
+	if (count_mem_nodes() <= 1)
+		ksft_exit_skip("At least 2 NUMA nodes with memory must be available\n");
+
 	if (ksm_write_sysfs(KSM_FP("merge_across_nodes"), merge_across_nodes))
 		return KSFT_FAIL;
 
-- 
2.52.0