cgroup: Fixes for v6.3-rc6

* Fix several cpuset bugs including one where it wasn't applying the target
   cgroup when tasks are created with CLONE_INTO_CGROUP.
 
 * Fix inversed locking order in cgroup1 freezer implementation.
 
 * Fix garbage cpu.stat::core_sched.forceidle_usec reporting in the root
   cgroup.
 
 This is a relatively big pull request this late in the cycle but the major
 contributor is the above mentioned cpuset bug which is rather significant.
 -----BEGIN PGP SIGNATURE-----
 
 iIQEABYIACwWIQTfIjM1kS57o3GsC/uxYfJx3gVYGQUCZDiJuw4cdGpAa2VybmVs
 Lm9yZwAKCRCxYfJx3gVYGbUfAQCLYhxijWvCpRYlQ3mfd1pgyvWNB90o4lnFkltz
 D0iSpwD/SOL5zwkR7WBeejJDKVIsezPpz3SZvxzzKMSk1VODkgo=
 =eOCG
 -----END PGP SIGNATURE-----

Merge tag 'cgroup-for-6.3-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup

Pull cgroup fixes from Tejun Heo:
 "This is a relatively big pull request this late in the cycle but the
  major contributor is the cpuset bug which is rather significant:

   - Fix several cpuset bugs including one where it wasn't applying the
     target cgroup when tasks are created with CLONE_INTO_CGROUP

  With a few smaller fixes:

   - Fix inversed locking order in cgroup1 freezer implementation

   - Fix garbage cpu.stat::core_sched.forceidle_usec reporting in the
     root cgroup"

* tag 'cgroup-for-6.3-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
  cgroup/cpuset: Make cpuset_attach_task() skip subpartitions CPUs for top_cpuset
  cgroup/cpuset: Add cpuset_can_fork() and cpuset_cancel_fork() methods
  cgroup/cpuset: Make cpuset_fork() handle CLONE_INTO_CGROUP properly
  cgroup/cpuset: Wake up cpuset_attach_wq tasks in cpuset_cancel_attach()
  cgroup,freezer: hold cpu_hotplug_lock before freezer_mutex
  cgroup/cpuset: Fix partition root's cpuset.cpus update bug
  cgroup: fix display of forceidle time at root

kernel/cgroup/cpuset.c

@@ -1513,7 +1513,7 @@ static int update_parent_subparts_cpumask(struct cpuset *cs, int cmd,
 	spin_unlock_irq(&callback_lock);
 
 	if (adding || deleting)
-		update_tasks_cpumask(parent, tmp->new_cpus);
+		update_tasks_cpumask(parent, tmp->addmask);
 
 	/*
 	 * Set or clear CS_SCHED_LOAD_BALANCE when partcmd_update, if necessary.
@@ -1770,10 +1770,13 @@ static int update_cpumask(struct cpuset *cs, struct cpuset *trialcs,
 	/*
 	 * Use the cpumasks in trialcs for tmpmasks when they are pointers
 	 * to allocated cpumasks.
+	 *
+	 * Note that update_parent_subparts_cpumask() uses only addmask &
+	 * delmask, but not new_cpus.
 	 */
 	tmp.addmask  = trialcs->subparts_cpus;
 	tmp.delmask  = trialcs->effective_cpus;
-	tmp.new_cpus = trialcs->cpus_allowed;
+	tmp.new_cpus = NULL;
 #endif
 
 	retval = validate_change(cs, trialcs);
@@ -1838,6 +1841,11 @@ static int update_cpumask(struct cpuset *cs, struct cpuset *trialcs,
 	}
 	spin_unlock_irq(&callback_lock);
 
+#ifdef CONFIG_CPUMASK_OFFSTACK
+	/* Now trialcs->cpus_allowed is available */
+	tmp.new_cpus = trialcs->cpus_allowed;
+#endif
+
 	/* effective_cpus will be updated here */
 	update_cpumasks_hier(cs, &tmp, false);
 
@@ -2445,6 +2453,20 @@ static int fmeter_getrate(struct fmeter *fmp)
 
 static struct cpuset *cpuset_attach_old_cs;
 
+/*
+ * Check to see if a cpuset can accept a new task
+ * For v1, cpus_allowed and mems_allowed can't be empty.
+ * For v2, effective_cpus can't be empty.
+ * Note that in v1, effective_cpus = cpus_allowed.
+ */
+static int cpuset_can_attach_check(struct cpuset *cs)
+{
+	if (cpumask_empty(cs->effective_cpus) ||
+	   (!is_in_v2_mode() && nodes_empty(cs->mems_allowed)))
+		return -ENOSPC;
+	return 0;
+}
+
 /* Called by cgroups to determine if a cpuset is usable; cpuset_rwsem held */
 static int cpuset_can_attach(struct cgroup_taskset *tset)
 {
@@ -2459,16 +2481,9 @@ static int cpuset_can_attach(struct cgroup_taskset *tset)
 
 	percpu_down_write(&cpuset_rwsem);
 
-	/* allow moving tasks into an empty cpuset if on default hierarchy */
-	ret = -ENOSPC;
-	if (!is_in_v2_mode() &&
-	    (cpumask_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed)))
-		goto out_unlock;
-
-	/*
-	 * Task cannot be moved to a cpuset with empty effective cpus.
-	 */
-	if (cpumask_empty(cs->effective_cpus))
+	/* Check to see if task is allowed in the cpuset */
+	ret = cpuset_can_attach_check(cs);
+	if (ret)
 		goto out_unlock;
 
 	cgroup_taskset_for_each(task, css, tset) {
@@ -2485,7 +2500,6 @@ static int cpuset_can_attach(struct cgroup_taskset *tset)
 	 * changes which zero cpus/mems_allowed.
 	 */
 	cs->attach_in_progress++;
-	ret = 0;
 out_unlock:
 	percpu_up_write(&cpuset_rwsem);
 	return ret;
@@ -2494,25 +2508,47 @@ out_unlock:
 static void cpuset_cancel_attach(struct cgroup_taskset *tset)
 {
 	struct cgroup_subsys_state *css;
+	struct cpuset *cs;
 
 	cgroup_taskset_first(tset, &css);
+	cs = css_cs(css);
 
 	percpu_down_write(&cpuset_rwsem);
-	css_cs(css)->attach_in_progress--;
+	cs->attach_in_progress--;
+	if (!cs->attach_in_progress)
+		wake_up(&cpuset_attach_wq);
 	percpu_up_write(&cpuset_rwsem);
 }
 
 /*
- * Protected by cpuset_rwsem.  cpus_attach is used only by cpuset_attach()
+ * Protected by cpuset_rwsem. cpus_attach is used only by cpuset_attach_task()
  * but we can't allocate it dynamically there.  Define it global and
  * allocate from cpuset_init().
  */
 static cpumask_var_t cpus_attach;
+static nodemask_t cpuset_attach_nodemask_to;
+
+static void cpuset_attach_task(struct cpuset *cs, struct task_struct *task)
+{
+	percpu_rwsem_assert_held(&cpuset_rwsem);
+
+	if (cs != &top_cpuset)
+		guarantee_online_cpus(task, cpus_attach);
+	else
+		cpumask_andnot(cpus_attach, task_cpu_possible_mask(task),
+			       cs->subparts_cpus);
+	/*
+	 * can_attach beforehand should guarantee that this doesn't
+	 * fail.  TODO: have a better way to handle failure here
+	 */
+	WARN_ON_ONCE(set_cpus_allowed_ptr(task, cpus_attach));
+
+	cpuset_change_task_nodemask(task, &cpuset_attach_nodemask_to);
+	cpuset_update_task_spread_flags(cs, task);
+}
 
 static void cpuset_attach(struct cgroup_taskset *tset)
 {
-	/* static buf protected by cpuset_rwsem */
-	static nodemask_t cpuset_attach_nodemask_to;
 	struct task_struct *task;
 	struct task_struct *leader;
 	struct cgroup_subsys_state *css;
@@ -2543,20 +2579,8 @@ static void cpuset_attach(struct cgroup_taskset *tset)
 
 	guarantee_online_mems(cs, &cpuset_attach_nodemask_to);
 
-	cgroup_taskset_for_each(task, css, tset) {
-		if (cs != &top_cpuset)
-			guarantee_online_cpus(task, cpus_attach);
-		else
-			cpumask_copy(cpus_attach, task_cpu_possible_mask(task));
-		/*
-		 * can_attach beforehand should guarantee that this doesn't
-		 * fail.  TODO: have a better way to handle failure here
-		 */
-		WARN_ON_ONCE(set_cpus_allowed_ptr(task, cpus_attach));
-
-		cpuset_change_task_nodemask(task, &cpuset_attach_nodemask_to);
-		cpuset_update_task_spread_flags(cs, task);
-	}
+	cgroup_taskset_for_each(task, css, tset)
+		cpuset_attach_task(cs, task);
 
 	/*
 	 * Change mm for all threadgroup leaders. This is expensive and may
@@ -3247,6 +3271,68 @@ static void cpuset_bind(struct cgroup_subsys_state *root_css)
 	percpu_up_write(&cpuset_rwsem);
 }
 
+/*
+ * In case the child is cloned into a cpuset different from its parent,
+ * additional checks are done to see if the move is allowed.
+ */
+static int cpuset_can_fork(struct task_struct *task, struct css_set *cset)
+{
+	struct cpuset *cs = css_cs(cset->subsys[cpuset_cgrp_id]);
+	bool same_cs;
+	int ret;
+
+	rcu_read_lock();
+	same_cs = (cs == task_cs(current));
+	rcu_read_unlock();
+
+	if (same_cs)
+		return 0;
+
+	lockdep_assert_held(&cgroup_mutex);
+	percpu_down_write(&cpuset_rwsem);
+
+	/* Check to see if task is allowed in the cpuset */
+	ret = cpuset_can_attach_check(cs);
+	if (ret)
+		goto out_unlock;
+
+	ret = task_can_attach(task, cs->effective_cpus);
+	if (ret)
+		goto out_unlock;
+
+	ret = security_task_setscheduler(task);
+	if (ret)
+		goto out_unlock;
+
+	/*
+	 * Mark attach is in progress.  This makes validate_change() fail
+	 * changes which zero cpus/mems_allowed.
+	 */
+	cs->attach_in_progress++;
+out_unlock:
+	percpu_up_write(&cpuset_rwsem);
+	return ret;
+}
+
+static void cpuset_cancel_fork(struct task_struct *task, struct css_set *cset)
+{
+	struct cpuset *cs = css_cs(cset->subsys[cpuset_cgrp_id]);
+	bool same_cs;
+
+	rcu_read_lock();
+	same_cs = (cs == task_cs(current));
+	rcu_read_unlock();
+
+	if (same_cs)
+		return;
+
+	percpu_down_write(&cpuset_rwsem);
+	cs->attach_in_progress--;
+	if (!cs->attach_in_progress)
+		wake_up(&cpuset_attach_wq);
+	percpu_up_write(&cpuset_rwsem);
+}
+
 /*
  * Make sure the new task conform to the current state of its parent,
  * which could have been changed by cpuset just after it inherits the
@@ -3254,11 +3340,33 @@ static void cpuset_bind(struct cgroup_subsys_state *root_css)
  */
 static void cpuset_fork(struct task_struct *task)
 {
-	if (task_css_is_root(task, cpuset_cgrp_id))
+	struct cpuset *cs;
+	bool same_cs;
+
+	rcu_read_lock();
+	cs = task_cs(task);
+	same_cs = (cs == task_cs(current));
+	rcu_read_unlock();
+
+	if (same_cs) {
+		if (cs == &top_cpuset)
 			return;
 
 		set_cpus_allowed_ptr(task, current->cpus_ptr);
 		task->mems_allowed = current->mems_allowed;
+		return;
+	}
+
+	/* CLONE_INTO_CGROUP */
+	percpu_down_write(&cpuset_rwsem);
+	guarantee_online_mems(cs, &cpuset_attach_nodemask_to);
+	cpuset_attach_task(cs, task);
+
+	cs->attach_in_progress--;
+	if (!cs->attach_in_progress)
+		wake_up(&cpuset_attach_wq);
+
+	percpu_up_write(&cpuset_rwsem);
 }
 
 struct cgroup_subsys cpuset_cgrp_subsys = {
@@ -3271,6 +3379,8 @@ struct cgroup_subsys cpuset_cgrp_subsys = {
 	.attach		= cpuset_attach,
 	.post_attach	= cpuset_post_attach,
 	.bind		= cpuset_bind,
+	.can_fork	= cpuset_can_fork,
+	.cancel_fork	= cpuset_cancel_fork,
 	.fork		= cpuset_fork,
 	.legacy_cftypes	= legacy_files,
 	.dfl_cftypes	= dfl_files,

kernel/cgroup/legacy_freezer.c

@@ -22,6 +22,7 @@
 #include <linux/freezer.h>
 #include <linux/seq_file.h>
 #include <linux/mutex.h>
+#include <linux/cpu.h>
 
 /*
  * A cgroup is freezing if any FREEZING flags are set.  FREEZING_SELF is
@@ -350,7 +351,7 @@ static void freezer_apply_state(struct freezer *freezer, bool freeze,
 
 	if (freeze) {
 		if (!(freezer->state & CGROUP_FREEZING))
-			static_branch_inc(&freezer_active);
+			static_branch_inc_cpuslocked(&freezer_active);
 		freezer->state |= state;
 		freeze_cgroup(freezer);
 	} else {
@@ -361,7 +362,7 @@ static void freezer_apply_state(struct freezer *freezer, bool freeze,
 		if (!(freezer->state & CGROUP_FREEZING)) {
 			freezer->state &= ~CGROUP_FROZEN;
 			if (was_freezing)
-				static_branch_dec(&freezer_active);
+				static_branch_dec_cpuslocked(&freezer_active);
 			unfreeze_cgroup(freezer);
 		}
 	}
@@ -379,6 +380,7 @@ static void freezer_change_state(struct freezer *freezer, bool freeze)
 {
 	struct cgroup_subsys_state *pos;
 
+	cpus_read_lock();
 	/*
 	 * Update all its descendants in pre-order traversal.  Each
 	 * descendant will try to inherit its parent's FREEZING state as
@@ -407,6 +409,7 @@ static void freezer_change_state(struct freezer *freezer, bool freeze)
 	}
 	rcu_read_unlock();
 	mutex_unlock(&freezer_mutex);
+	cpus_read_unlock();
 }
 
 static ssize_t freezer_write(struct kernfs_open_file *of,

kernel/cgroup/rstat.c

@@ -457,9 +457,7 @@ static void root_cgroup_cputime(struct cgroup_base_stat *bstat)
 	struct task_cputime *cputime = &bstat->cputime;
 	int i;
 
-	cputime->stime = 0;
-	cputime->utime = 0;
-	cputime->sum_exec_runtime = 0;
+	memset(bstat, 0, sizeof(*bstat));
 	for_each_possible_cpu(i) {
 		struct kernel_cpustat kcpustat;
 		u64 *cpustat = kcpustat.cpustat;