[PATCH v3 2/4] KVM: selftests: Use an allowed memory node in test_mbind()

[Shivank Garg <shivankg@amd.com>]

test_mbind() hardcodes node 0 and skips the test unless sysfs reports
more than one NUMA node.  This can fail if node 0 has no memory or is
excluded by the task's cpuset.

Pick a node from MPOL_F_MEMS_ALLOWED instead.  The test only needs one
node, so allow it to run on single-node systems too.

Signed-off-by: Shivank Garg <shivankg@amd.com>
---
 tools/testing/selftests/kvm/guest_memfd_test.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/kvm/guest_memfd_test.c b/tools/testing/selftests/kvm/guest_memfd_test.c
index b3bd03167ba8..ea8e38064e83 100644
--- a/tools/testing/selftests/kvm/guest_memfd_test.c
+++ b/tools/testing/selftests/kvm/guest_memfd_test.c
@@ -81,15 +81,16 @@ static void test_mbind(int fd, size_t total_size)
 	unsigned long nodemask = 0;
 	unsigned long maxnode = MAXNODE_FOR_MASK(nodemask);
 	unsigned long bind_nodemask;
+	unsigned long mems_allowed;
 	int policy;
 	char *mem;
 	int nid;
 	int ret;
 
-	if (!is_multi_numa_node_system())
+	if (!kvm_get_numa_memory_nodes(&mems_allowed))
 		return;
 
-	nid = 0;
+	nid = kvm_get_next_numa_node(mems_allowed, -1);
 	bind_nodemask = BIT(nid);
 
 	mem = kvm_mmap(total_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd);

-- 
2.43.0