Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit ae8086ce authored by Tejun Heo's avatar Tejun Heo
Browse files

cpuset: introduce cpuset_for_each_child() 



Instead of iterating cgroup->children directly, introduce and use
cpuset_for_each_child() which wraps cgroup_for_each_child() and
performs online check.  As it uses the generic iterator, it requires
RCU read locking too.

As cpuset is currently protected by cgroup_mutex, non-online cpusets
aren't visible to all the iterations and this patch currently doesn't
make any functional difference.  This will be used to de-couple cpuset
locking from cgroup core.

Signed-off-by: default avatarTejun Heo <tj@kernel.org>
Acked-by: default avatarLi Zefan <lizefan@huawei.com>
parent efeb77b2

kernel/cpuset.c

+54 −31
Original line number Original line Diff line number Diff line
@@ -200,6 +200,19 @@ static struct cpuset top_cpuset = { 
		  (1 << CS_MEM_EXCLUSIVE)),

		  (1 << CS_MEM_EXCLUSIVE)),

};

};





/**

 * cpuset_for_each_child - traverse online children of a cpuset

 * @child_cs: loop cursor pointing to the current child

 * @pos_cgrp: used for iteration

 * @parent_cs: target cpuset to walk children of

 *

 * Walk @child_cs through the online children of @parent_cs.  Must be used

 * with RCU read locked.

 */

#define cpuset_for_each_child(child_cs, pos_cgrp, parent_cs)		\

	cgroup_for_each_child((pos_cgrp), (parent_cs)->css.cgroup)	\

		if (is_cpuset_online(((child_cs) = cgroup_cs((pos_cgrp)))))



/*

/*

 * There are two global mutexes guarding cpuset structures.  The first

 * There are two global mutexes guarding cpuset structures.  The first

 * is the main control groups cgroup_mutex, accessed via

 * is the main control groups cgroup_mutex, accessed via
@@ -419,48 +432,55 @@ static int validate_change(const struct cpuset *cur, const struct cpuset *trial) 
{

{

	struct cgroup *cont;

	struct cgroup *cont;

	struct cpuset *c, *par;

	struct cpuset *c, *par;

	int ret;



	rcu_read_lock();





	/* Each of our child cpusets must be a subset of us */

	/* Each of our child cpusets must be a subset of us */

	list_for_each_entry(cont, &cur->css.cgroup->children, sibling) {

	ret = -EBUSY;

		if (!is_cpuset_subset(cgroup_cs(cont), trial))

	cpuset_for_each_child(c, cont, cur)

			return -EBUSY;

		if (!is_cpuset_subset(c, trial))

	}

			goto out;





	/* Remaining checks don't apply to root cpuset */

	/* Remaining checks don't apply to root cpuset */

	ret = 0;

	if (cur == &top_cpuset)

	if (cur == &top_cpuset)

		return 0;

		goto out;





	par = cur->parent;

	par = cur->parent;





	/* We must be a subset of our parent cpuset */

	/* We must be a subset of our parent cpuset */

	ret = -EACCES;

	if (!is_cpuset_subset(trial, par))

	if (!is_cpuset_subset(trial, par))

		return -EACCES;

		goto out;





	/*

	/*

	 * If either I or some sibling (!= me) is exclusive, we can't

	 * If either I or some sibling (!= me) is exclusive, we can't

	 * overlap

	 * overlap

	 */

	 */

	list_for_each_entry(cont, &par->css.cgroup->children, sibling) {

	ret = -EINVAL;

		c = cgroup_cs(cont);

	cpuset_for_each_child(c, cont, par) {

		if ((is_cpu_exclusive(trial) || is_cpu_exclusive(c)) &&

		if ((is_cpu_exclusive(trial) || is_cpu_exclusive(c)) &&

		    c != cur &&

		    c != cur &&

		    cpumask_intersects(trial->cpus_allowed, c->cpus_allowed))

		    cpumask_intersects(trial->cpus_allowed, c->cpus_allowed))

			return -EINVAL;

			goto out;

		if ((is_mem_exclusive(trial) || is_mem_exclusive(c)) &&

		if ((is_mem_exclusive(trial) || is_mem_exclusive(c)) &&

		    c != cur &&

		    c != cur &&

		    nodes_intersects(trial->mems_allowed, c->mems_allowed))

		    nodes_intersects(trial->mems_allowed, c->mems_allowed))

			return -EINVAL;

			goto out;

	}

	}





	/* Cpusets with tasks can't have empty cpus_allowed or mems_allowed */

	/* Cpusets with tasks can't have empty cpus_allowed or mems_allowed */

	if (cgroup_task_count(cur->css.cgroup)) {

	ret = -ENOSPC;

		if (cpumask_empty(trial->cpus_allowed) ||

	if (cgroup_task_count(cur->css.cgroup) &&

		    nodes_empty(trial->mems_allowed)) {

	    (cpumask_empty(trial->cpus_allowed) ||

			return -ENOSPC;

	     nodes_empty(trial->mems_allowed)))

		}

		goto out;

	}





	return 0;

	ret = 0;

out:

	rcu_read_unlock();

	return ret;

}

}





#ifdef CONFIG_SMP

#ifdef CONFIG_SMP
@@ -501,10 +521,10 @@ update_domain_attr_tree(struct sched_domain_attr *dattr, struct cpuset *c) 
		if (is_sched_load_balance(cp))

		if (is_sched_load_balance(cp))

			update_domain_attr(dattr, cp);

			update_domain_attr(dattr, cp);





		list_for_each_entry(cont, &cp->css.cgroup->children, sibling) {

		rcu_read_lock();

			child = cgroup_cs(cont);

		cpuset_for_each_child(child, cont, cp)

			list_add_tail(&child->stack_list, &q);

			list_add_tail(&child->stack_list, &q);

		}

		rcu_read_unlock();

	}

	}

}

}
@@ -623,10 +643,10 @@ static int generate_sched_domains(cpumask_var_t **domains, 
			continue;

			continue;

		}

		}





		list_for_each_entry(cont, &cp->css.cgroup->children, sibling) {

		rcu_read_lock();

			child = cgroup_cs(cont);

		cpuset_for_each_child(child, cont, cp)

			list_add_tail(&child->stack_list, &q);

			list_add_tail(&child->stack_list, &q);

		}

		rcu_read_unlock();

  	}

  	}





	for (i = 0; i < csn; i++)

	for (i = 0; i < csn; i++)
@@ -1824,7 +1844,8 @@ static int cpuset_css_online(struct cgroup *cgrp) 
{

{

	struct cpuset *cs = cgroup_cs(cgrp);

	struct cpuset *cs = cgroup_cs(cgrp);

	struct cpuset *parent = cs->parent;

	struct cpuset *parent = cs->parent;

	struct cgroup *tmp_cg;

	struct cpuset *tmp_cs;

	struct cgroup *pos_cg;





	if (!parent)

	if (!parent)

		return 0;

		return 0;
@@ -1853,12 +1874,14 @@ static int cpuset_css_online(struct cgroup *cgrp) 
	 * changed to grant parent->cpus_allowed-sibling_cpus_exclusive

	 * changed to grant parent->cpus_allowed-sibling_cpus_exclusive

	 * (and likewise for mems) to the new cgroup.

	 * (and likewise for mems) to the new cgroup.

	 */

	 */

	list_for_each_entry(tmp_cg, &cgrp->parent->children, sibling) {

	rcu_read_lock();

		struct cpuset *tmp_cs = cgroup_cs(tmp_cg);

	cpuset_for_each_child(tmp_cs, pos_cg, parent) {



		if (is_mem_exclusive(tmp_cs) || is_cpu_exclusive(tmp_cs)) {

		if (is_mem_exclusive(tmp_cs) || is_cpu_exclusive(tmp_cs))

			rcu_read_unlock();

			return 0;

			return 0;

		}

		}

	}

	rcu_read_unlock();





	mutex_lock(&callback_mutex);

	mutex_lock(&callback_mutex);

	cs->mems_allowed = parent->mems_allowed;

	cs->mems_allowed = parent->mems_allowed;
@@ -2027,10 +2050,10 @@ static struct cpuset *cpuset_next(struct list_head *queue) 




	cp = list_first_entry(queue, struct cpuset, stack_list);

	cp = list_first_entry(queue, struct cpuset, stack_list);

	list_del(queue->next);

	list_del(queue->next);

	list_for_each_entry(cont, &cp->css.cgroup->children, sibling) {

	rcu_read_lock();

		child = cgroup_cs(cont);

	cpuset_for_each_child(child, cont, cp)

		list_add_tail(&child->stack_list, queue);

		list_add_tail(&child->stack_list, queue);

	}

	rcu_read_unlock();





	return cp;

	return cp;

}

}