[PATCH-next v3 09/11] cgroup/cpuset: Release invalid isolated CPUs with a shrunk root partition parent

[Waiman Long]

It is also reported by sashiko that there may be leakages of isolated
CPUs when a root partition parent shrinks its effective_xcpus with
isolated partition children. Leakage here means that isolated_cpus may
contain extra CPUs that are not in any isolated partitions and not boot
time isolated. Those CPUs cannot be used in any isolated partitions
anymore.

There are 2 different cases here. The effective_xcpus becomes empty and
the isolated partition should be invalidated in this case. Alternatively
some of its effective_xcpus are not in parent's effective xcpus but
still have some valid effective_xcpus left. The isolated partition can
remain valid. In both cases, the invalid isolated CPUs will have to be
released from isolated_cpus if appropriate.

Fix that by calling isolated_cpus_release() to release those isolated
CPUs from isolated_cpus.

New test cases are added to test_cpuset_prs.sh to test this.

Fixes: 11e5f407b64a ("cgroup/cpuset: Keep track of CPUs in isolated partitions")
Signed-off-by: Waiman Long <longman@xxxxxxxxxx>
---
 kernel/cgroup/cpuset.c                           | 16 +++++++++++++++-
 .../testing/selftests/cgroup/test_cpuset_prs.sh  |  6 ++++++
 2 files changed, 21 insertions(+), 1 deletion(-)

diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
index 6d224e614e74..3daab2381f04 100644
--- a/kernel/cgroup/cpuset.c
+++ b/kernel/cgroup/cpuset.c
@@ -2012,8 +2012,22 @@ static int update_parent_effective_cpumask(struct cpuset *cs, int cmd,
 		 *
 		 * A partition error happens when parent has tasks and all
 		 * its effective CPUs will have to be distributed out.
+		 *
+		 * For an isolated partition with a shrunk root partition parent
+		 * we may need to return some or all of its isolated CPUs.
 		 */
-		if (nocpu || cpumask_empty(tmp->delmask)) {
+		nocpu |= cpumask_empty(tmp->delmask);
+		if ((parent_prs == PRS_ROOT) && (old_prs == PRS_ISOLATED)) {
+			cpumask_clear(tmp->addmask);
+			if (nocpu)
+				cpumask_copy(tmp->addmask, cs->effective_xcpus);
+			else if (!cpumask_subset(cs->effective_xcpus, parent->effective_xcpus))
+				cpumask_andnot(tmp->addmask, cs->effective_cpus,
+							     parent->effective_xcpus);
+			if (!cpumask_empty(tmp->addmask))
+				isolated_cpus_release(cs, tmp->addmask);
+		}
+		if (nocpu) {
 			part_error = PERR_NOCPUS;
 			if (is_partition_valid(cs))
 				adding = cpumask_and(tmp->addmask,
diff --git a/tools/testing/selftests/cgroup/test_cpuset_prs.sh b/tools/testing/selftests/cgroup/test_cpuset_prs.sh
index d7d0bef0a17a..ceb1a3b4f94e 100755
--- a/tools/testing/selftests/cgroup/test_cpuset_prs.sh
+++ b/tools/testing/selftests/cgroup/test_cpuset_prs.sh
@@ -522,6 +522,12 @@ TEST_MATRIX=(
 	# its effective xcpus will be removed from isolated_cpus.
 	" C1-4:P1 C3-4:P2   .      .    T:C3-4   .      .      .     0 A1:3-4|A2:3-4 A1:P1|A2:P-2 ."
 
+	# When a root partition parent with an isolated partition child is shrunk
+	# so that its effective xcpus is no longer a superset of its child effective
+	# xcpus, the invalid isolated CPUs should be removed from cpuset.cpus.isolated.
+	" C1-4:P1 C3-4:P2   .      .    T:C4-6   .      .      .     0 A1:5-6|A2:4 A1:P1|A2:P2 4"
+	" C1-4:P1 C1-2:P2   .      .    T:C4-6   .      .      .     0 A1:4-6|A2:4-6 A1:P1|A2:P-2 ."
+
 	#
 	#  old-A1 old-A2 old-A3 old-B1 new-A1 new-A2 new-A3 new-B1 fail ECPUs Pstate ISOLCPUS
 	#  ------ ------ ------ ------ ------ ------ ------ ------ ---- ----- ------ --------
-- 
2.56.0