Merge branch 'for-4.15-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup

Pull cgroup fixes from Tejun Heo:

 - Prateek posted a couple patches to fix a deadlock involving cpuset
   and workqueue. It unfortunately caused a different deadlock and the
   recent workqueue hotplug simplification removed the original
   deadlock, so Prateek's two patches are reverted for now.

 - The new stat code was missing u64_stats initialization. Fixed.

 - Doc and other misc changes

* 'for-4.15-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
  cgroup: add warning about RT not being supported on cgroup2
  Revert "cgroup/cpuset: remove circular dependency deadlock"
  Revert "cpuset: Make cpuset hotplug synchronous"
  cgroup: properly init u64_stats
  debug cgroup: use task_css_set instead of rcu_dereference
  cpuset: Make cpuset hotplug synchronous
  cgroup/cpuset: remove circular dependency deadlock

3 files changed, 15 insertions, 4 deletions

diff --git a/Documentation/cgroup-v2.txt b/Documentation/cgroup-v2.txt
index 779211fbb69f..2cddab7efb20 100644
--- a/Documentation/cgroup-v2.txt
+++ b/Documentation/cgroup-v2.txt
@@ -898,6 +898,13 @@ controller implements weight and absolute bandwidth limit models for
 normal scheduling policy and absolute bandwidth allocation model for
 realtime scheduling policy.
 
+WARNING: cgroup2 doesn't yet support control of realtime processes and
+the cpu controller can only be enabled when all RT processes are in
+the root cgroup.  Be aware that system management software may already
+have placed RT processes into nonroot cgroups during the system boot
+process, and these processes may need to be moved to the root cgroup
+before the cpu controller can be enabled.
+
 
 CPU Interface Files
 ~~~~~~~~~~~~~~~~~~~
diff --git a/kernel/cgroup/debug.c b/kernel/cgroup/debug.c
index 5f780d8f6a9d..9caeda610249 100644
--- a/kernel/cgroup/debug.c
+++ b/kernel/cgroup/debug.c
@@ -50,7 +50,7 @@ static int current_css_set_read(struct seq_file *seq, void *v)
 
         spin_lock_irq(&css_set_lock);
         rcu_read_lock();
-        cset = rcu_dereference(current->cgroups);
+        cset = task_css_set(current);
         refcnt = refcount_read(&cset->refcount);
         seq_printf(seq, "css_set %pK %d", cset, refcnt);
         if (refcnt > cset->nr_tasks)
@@ -96,7 +96,7 @@ static int current_css_set_cg_links_read(struct seq_file *seq, void *v)
 
         spin_lock_irq(&css_set_lock);
         rcu_read_lock();
-        cset = rcu_dereference(current->cgroups);
+        cset = task_css_set(current);
         list_for_each_entry(link, &cset->cgrp_links, cgrp_link) {
                 struct cgroup *c = link->cgrp;
 
diff --git a/kernel/cgroup/stat.c b/kernel/cgroup/stat.c
index 133b465691d6..1e111dd455c4 100644
--- a/kernel/cgroup/stat.c
+++ b/kernel/cgroup/stat.c
@@ -296,8 +296,12 @@ int cgroup_stat_init(struct cgroup *cgrp)
         }
 
         /* ->updated_children list is self terminated */
-        for_each_possible_cpu(cpu)
-                cgroup_cpu_stat(cgrp, cpu)->updated_children = cgrp;
+        for_each_possible_cpu(cpu) {
+                struct cgroup_cpu_stat *cstat = cgroup_cpu_stat(cgrp, cpu);
+
+                cstat->updated_children = cgrp;
+                u64_stats_init(&cstat->sync);
+        }
 
         prev_cputime_init(&cgrp->stat.prev_cputime);