cgroup/cpuset: Ensure domain isolated CPUs stay in root or isolated partition

Commit 4a74e418881f ("cgroup/cpuset: Check partition conflict with
housekeeping setup") is supposed to ensure that domain isolated CPUs
designated by the "isolcpus" boot command line option stay either in
root partition or in isolated partitions. However, the required check
wasn't implemented when a remote partition was created or when an
existing partition changed type from "root" to "isolated".

Even though this is a relatively minor issue, we still need to add the
required prstate_housekeeping_conflict() call in the right places to
ensure that the rule is strictly followed.

The following steps can be used to reproduce the problem before this
fix.

  # fmt -1 /proc/cmdline | grep isolcpus
  isolcpus=9
  # cd /sys/fs/cgroup/
  # echo +cpuset > cgroup.subtree_control
  # mkdir test
  # echo 9 > test/cpuset.cpus
  # echo isolated > test/cpuset.cpus.partition
  # cat test/cpuset.cpus.partition
  isolated
  # cat test/cpuset.cpus.effective
  9
  # echo root > test/cpuset.cpus.partition
  # cat test/cpuset.cpus.effective
  9
  # cat test/cpuset.cpus.partition
  root

With this fix, the last few steps will become:

  # echo root > test/cpuset.cpus.partition
  # cat test/cpuset.cpus.effective
  0-8,10-95
  # cat test/cpuset.cpus.partition
  root invalid (partition config conflicts with housekeeping setup)

Reported-by: Chen Ridong <chenridong@huawei.com>
Signed-off-by: Waiman Long <longman@redhat.com>
Reviewed-by: Chen Ridong <chenridong@huawei.com>
Signed-off-by: Tejun Heo <tj@kernel.org>

diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
index cc9c3402f16bc49524b49babd2f555a0873c4d8d..2daf58bf0bbb7877f1f611a0ba86e9ef025368b8 100644 (file)
--- a/kernel/cgroup/cpuset.c
+++ b/kernel/cgroup/cpuset.c
@@ -1610,8 +1610,9 @@ static int remote_partition_enable(struct cpuset *cs, int new_prs,
        if (!cpumask_intersects(tmp->new_cpus, cpu_active_mask) ||
            cpumask_subset(top_cpuset.effective_cpus, tmp->new_cpus))
                return PERR_INVCPUS;
-       if ((new_prs == PRS_ISOLATED) &&
-           !isolated_cpus_can_update(tmp->new_cpus, NULL))
+       if (((new_prs == PRS_ISOLATED) &&
+            !isolated_cpus_can_update(tmp->new_cpus, NULL)) ||
+           prstate_housekeeping_conflict(new_prs, tmp->new_cpus))
                return PERR_HKEEPING;
 
        spin_lock_irq(&callback_lock);
@@ -3062,8 +3063,9 @@ static int update_prstate(struct cpuset *cs, int new_prs)
                 * A change in load balance state only, no change in cpumasks.
                 * Need to update isolated_cpus.
                 */
-               if ((new_prs == PRS_ISOLATED) &&
-                   !isolated_cpus_can_update(cs->effective_xcpus, NULL))
+               if (((new_prs == PRS_ISOLATED) &&
+                    !isolated_cpus_can_update(cs->effective_xcpus, NULL)) ||
+                   prstate_housekeeping_conflict(new_prs, cs->effective_xcpus))
                        err = PERR_HKEEPING;
                else
                        isolcpus_updated = true;