cgroup.c

	cgroup_attach_task(new_cgroup, task, false);
	mutex_unlock(&cgroup_mutex);
}

/**
 * cgroup_trasnsfer_tasks - move tasks from one cgroup to another
 * @to: cgroup to which the tasks will be moved
 * @from: cgroup in which the tasks currently reside
 */
int cgroup_transfer_tasks(struct cgroup *to, struct cgroup *from)
{
	return css_scan_tasks(&from->dummy_css, NULL, cgroup_transfer_one_task,
			      to, NULL);
}

/*
 * Stuff for reading the 'tasks'/'procs' files.
 *
 * Reading this file can return large amounts of data if a cgroup has
 * *lots* of attached tasks. So it may need several calls to read(),
 * but we cannot guarantee that the information we produce is correct
 * unless we produce it entirely atomically.
 *
 */

/* which pidlist file are we talking about? */
enum cgroup_filetype {
	CGROUP_FILE_PROCS,
	CGROUP_FILE_TASKS,
};

/*
 * A pidlist is a list of pids that virtually represents the contents of one
 * of the cgroup files ("procs" or "tasks"). We keep a list of such pidlists,
 * a pair (one each for procs, tasks) for each pid namespace that's relevant
 * to the cgroup.
 */
struct cgroup_pidlist {
	/*
	 * used to find which pidlist is wanted. doesn't change as long as
	 * this particular list stays in the list.
	*/
	struct { enum cgroup_filetype type; struct pid_namespace *ns; } key;
	/* array of xids */
	pid_t *list;
	/* how many elements the above list has */
	int length;
	/* each of these stored in a list by its cgroup */
	struct list_head links;
	/* pointer to the cgroup we belong to, for list removal purposes */
	struct cgroup *owner;
	/* for delayed destruction */
	struct delayed_work destroy_dwork;
};

/*
 * The following two functions "fix" the issue where there are more pids
 * than kmalloc will give memory for; in such cases, we use vmalloc/vfree.
 * TODO: replace with a kernel-wide solution to this problem
 */
#define PIDLIST_TOO_LARGE(c) ((c) * sizeof(pid_t) > (PAGE_SIZE * 2))
static void *pidlist_allocate(int count)
{
	if (PIDLIST_TOO_LARGE(count))
		return vmalloc(count * sizeof(pid_t));
	else
		return kmalloc(count * sizeof(pid_t), GFP_KERNEL);
}

static void pidlist_free(void *p)
{
	if (is_vmalloc_addr(p))
		vfree(p);
	else
		kfree(p);
}

/*
 * Used to destroy all pidlists lingering waiting for destroy timer.  None
 * should be left afterwards.
 */
static void cgroup_pidlist_destroy_all(struct cgroup *cgrp)
{
	struct cgroup_pidlist *l, *tmp_l;

	mutex_lock(&cgrp->pidlist_mutex);
	list_for_each_entry_safe(l, tmp_l, &cgrp->pidlists, links)
		mod_delayed_work(cgroup_pidlist_destroy_wq, &l->destroy_dwork, 0);
	mutex_unlock(&cgrp->pidlist_mutex);

	flush_workqueue(cgroup_pidlist_destroy_wq);
	BUG_ON(!list_empty(&cgrp->pidlists));
}

static void cgroup_pidlist_destroy_work_fn(struct work_struct *work)
{
	struct delayed_work *dwork = to_delayed_work(work);
	struct cgroup_pidlist *l = container_of(dwork, struct cgroup_pidlist,
						destroy_dwork);
	struct cgroup_pidlist *tofree = NULL;

	mutex_lock(&l->owner->pidlist_mutex);

	/*
	 * Destroy iff we didn't get queued again.  The state won't change
	 * as destroy_dwork can only be queued while locked.
	 */
	if (!delayed_work_pending(dwork)) {
		list_del(&l->links);
		pidlist_free(l->list);
		put_pid_ns(l->key.ns);
		tofree = l;
	}

	mutex_unlock(&l->owner->pidlist_mutex);
	kfree(tofree);
}

/*
 * pidlist_uniq - given a kmalloc()ed list, strip out all duplicate entries
 * Returns the number of unique elements.
 */
static int pidlist_uniq(pid_t *list, int length)
{
	int src, dest = 1;

	/*
	 * we presume the 0th element is unique, so i starts at 1. trivial
	 * edge cases first; no work needs to be done for either
	 */
	if (length == 0 || length == 1)
		return length;
	/* src and dest walk down the list; dest counts unique elements */
	for (src = 1; src < length; src++) {
		/* find next unique element */
		while (list[src] == list[src-1]) {
			src++;
			if (src == length)
				goto after;
		}
		/* dest always points to where the next unique element goes */
		list[dest] = list[src];
		dest++;
	}
after:
	return dest;
}

/*
 * The two pid files - task and cgroup.procs - guaranteed that the result
 * is sorted, which forced this whole pidlist fiasco.  As pid order is
 * different per namespace, each namespace needs differently sorted list,
 * making it impossible to use, for example, single rbtree of member tasks
 * sorted by task pointer.  As pidlists can be fairly large, allocating one
 * per open file is dangerous, so cgroup had to implement shared pool of
 * pidlists keyed by cgroup and namespace.
 *
 * All this extra complexity was caused by the original implementation
 * committing to an entirely unnecessary property.  In the long term, we
 * want to do away with it.  Explicitly scramble sort order if
 * sane_behavior so that no such expectation exists in the new interface.
 *
 * Scrambling is done by swapping every two consecutive bits, which is
 * non-identity one-to-one mapping which disturbs sort order sufficiently.
 */
static pid_t pid_fry(pid_t pid)
{
	unsigned a = pid & 0x55555555;
	unsigned b = pid & 0xAAAAAAAA;

	return (a << 1) | (b >> 1);
}

static pid_t cgroup_pid_fry(struct cgroup *cgrp, pid_t pid)
{
	if (cgroup_sane_behavior(cgrp))
		return pid_fry(pid);
	else
		return pid;
}

static int cmppid(const void *a, const void *b)
{
	return *(pid_t *)a - *(pid_t *)b;
}

static int fried_cmppid(const void *a, const void *b)
{
	return pid_fry(*(pid_t *)a) - pid_fry(*(pid_t *)b);
}

static struct cgroup_pidlist *cgroup_pidlist_find(struct cgroup *cgrp,
						  enum cgroup_filetype type)
{
	struct cgroup_pidlist *l;
	/* don't need task_nsproxy() if we're looking at ourself */
	struct pid_namespace *ns = task_active_pid_ns(current);

	lockdep_assert_held(&cgrp->pidlist_mutex);

	list_for_each_entry(l, &cgrp->pidlists, links)
		if (l->key.type == type && l->key.ns == ns)
			return l;
	return NULL;
}

/*
 * find the appropriate pidlist for our purpose (given procs vs tasks)
 * returns with the lock on that pidlist already held, and takes care
 * of the use count, or returns NULL with no locks held if we're out of
 * memory.
 */
static struct cgroup_pidlist *cgroup_pidlist_find_create(struct cgroup *cgrp,
						enum cgroup_filetype type)
{
	struct cgroup_pidlist *l;

	lockdep_assert_held(&cgrp->pidlist_mutex);

	l = cgroup_pidlist_find(cgrp, type);
	if (l)
		return l;

	/* entry not found; create a new one */
	l = kzalloc(sizeof(struct cgroup_pidlist), GFP_KERNEL);
	if (!l)
		return l;

	INIT_DELAYED_WORK(&l->destroy_dwork, cgroup_pidlist_destroy_work_fn);
	l->key.type = type;
	/* don't need task_nsproxy() if we're looking at ourself */
	l->key.ns = get_pid_ns(task_active_pid_ns(current));
	l->owner = cgrp;
	list_add(&l->links, &cgrp->pidlists);
	return l;
}

/*
 * Load a cgroup's pidarray with either procs' tgids or tasks' pids
 */
static int pidlist_array_load(struct cgroup *cgrp, enum cgroup_filetype type,
			      struct cgroup_pidlist **lp)
{
	pid_t *array;
	int length;
	int pid, n = 0; /* used for populating the array */
	struct css_task_iter it;
	struct task_struct *tsk;
	struct cgroup_pidlist *l;

	lockdep_assert_held(&cgrp->pidlist_mutex);

	/*
	 * If cgroup gets more users after we read count, we won't have
	 * enough space - tough.  This race is indistinguishable to the
	 * caller from the case that the additional cgroup users didn't
	 * show up until sometime later on.
	 */
	length = cgroup_task_count(cgrp);
	array = pidlist_allocate(length);
	if (!array)
		return -ENOMEM;
	/* now, populate the array */
	css_task_iter_start(&cgrp->dummy_css, &it);
	while ((tsk = css_task_iter_next(&it))) {
		if (unlikely(n == length))
			break;
		/* get tgid or pid for procs or tasks file respectively */
		if (type == CGROUP_FILE_PROCS)
			pid = task_tgid_vnr(tsk);
		else
			pid = task_pid_vnr(tsk);
		if (pid > 0) /* make sure to only use valid results */
			array[n++] = pid;
	}
	css_task_iter_end(&it);
	length = n;
	/* now sort & (if procs) strip out duplicates */
	if (cgroup_sane_behavior(cgrp))
		sort(array, length, sizeof(pid_t), fried_cmppid, NULL);
	else
		sort(array, length, sizeof(pid_t), cmppid, NULL);
	if (type == CGROUP_FILE_PROCS)
		length = pidlist_uniq(array, length);

	l = cgroup_pidlist_find_create(cgrp, type);
	if (!l) {
		mutex_unlock(&cgrp->pidlist_mutex);
		pidlist_free(array);
		return -ENOMEM;
	}

	/* store array, freeing old if necessary */
	pidlist_free(l->list);
	l->list = array;
	l->length = length;
	*lp = l;
	return 0;
}

/**
 * cgroupstats_build - build and fill cgroupstats
 * @stats: cgroupstats to fill information into
 * @dentry: A dentry entry belonging to the cgroup for which stats have
 * been requested.
 *
 * Build and fill cgroupstats so that taskstats can export it to user
 * space.
 */
int cgroupstats_build(struct cgroupstats *stats, struct dentry *dentry)
{
	struct kernfs_node *kn = kernfs_node_from_dentry(dentry);
	struct cgroup *cgrp;
	struct css_task_iter it;
	struct task_struct *tsk;

	/* it should be kernfs_node belonging to cgroupfs and is a directory */
	if (dentry->d_sb->s_type != &cgroup_fs_type || !kn ||
	    kernfs_type(kn) != KERNFS_DIR)
		return -EINVAL;

	/*
	 * We aren't being called from kernfs and there's no guarantee on
	 * @kn->priv's validity.  For this and css_tryget_from_dir(),
	 * @kn->priv is RCU safe.  Let's do the RCU dancing.
	 */
	rcu_read_lock();
	cgrp = rcu_dereference(kn->priv);
	if (!cgrp) {
		rcu_read_unlock();
		return -ENOENT;
	}

	css_task_iter_start(&cgrp->dummy_css, &it);
	while ((tsk = css_task_iter_next(&it))) {
		switch (tsk->state) {
		case TASK_RUNNING:
			stats->nr_running++;
			break;
		case TASK_INTERRUPTIBLE:
			stats->nr_sleeping++;
			break;
		case TASK_UNINTERRUPTIBLE:
			stats->nr_uninterruptible++;
			break;
		case TASK_STOPPED:
			stats->nr_stopped++;
			break;
		default:
			if (delayacct_is_task_waiting_on_io(tsk))
				stats->nr_io_wait++;
			break;
		}
	}
	css_task_iter_end(&it);

	rcu_read_unlock();
	return 0;
}


/*
 * seq_file methods for the tasks/procs files. The seq_file position is the
 * next pid to display; the seq_file iterator is a pointer to the pid
 * in the cgroup->l->list array.
 */

static void *cgroup_pidlist_start(struct seq_file *s, loff_t *pos)
{
	/*
	 * Initially we receive a position value that corresponds to
	 * one more than the last pid shown (or 0 on the first call or
	 * after a seek to the start). Use a binary-search to find the
	 * next pid to display, if any
	 */
	struct kernfs_open_file *of = s->private;
	struct cgroup *cgrp = seq_css(s)->cgroup;
	struct cgroup_pidlist *l;
	enum cgroup_filetype type = seq_cft(s)->private;
	int index = 0, pid = *pos;
	int *iter, ret;

	mutex_lock(&cgrp->pidlist_mutex);

	/*
	 * !NULL @of->priv indicates that this isn't the first start()
	 * after open.  If the matching pidlist is around, we can use that.
	 * Look for it.  Note that @of->priv can't be used directly.  It
	 * could already have been destroyed.
	 */
	if (of->priv)
		of->priv = cgroup_pidlist_find(cgrp, type);

	/*
	 * Either this is the first start() after open or the matching
	 * pidlist has been destroyed inbetween.  Create a new one.
	 */
	if (!of->priv) {
		ret = pidlist_array_load(cgrp, type,
					 (struct cgroup_pidlist **)&of->priv);
		if (ret)
			return ERR_PTR(ret);
	}
	l = of->priv;

	if (pid) {
		int end = l->length;

		while (index < end) {
			int mid = (index + end) / 2;
			if (cgroup_pid_fry(cgrp, l->list[mid]) == pid) {
				index = mid;
				break;
			} else if (cgroup_pid_fry(cgrp, l->list[mid]) <= pid)
				index = mid + 1;
			else
				end = mid;
		}
	}
	/* If we're off the end of the array, we're done */
	if (index >= l->length)
		return NULL;
	/* Update the abstract position to be the actual pid that we found */
	iter = l->list + index;
	*pos = cgroup_pid_fry(cgrp, *iter);
	return iter;
}

static void cgroup_pidlist_stop(struct seq_file *s, void *v)
{
	struct kernfs_open_file *of = s->private;
	struct cgroup_pidlist *l = of->priv;

	if (l)
		mod_delayed_work(cgroup_pidlist_destroy_wq, &l->destroy_dwork,
				 CGROUP_PIDLIST_DESTROY_DELAY);
	mutex_unlock(&seq_css(s)->cgroup->pidlist_mutex);
}

static void *cgroup_pidlist_next(struct seq_file *s, void *v, loff_t *pos)
{
	struct kernfs_open_file *of = s->private;
	struct cgroup_pidlist *l = of->priv;
	pid_t *p = v;
	pid_t *end = l->list + l->length;
	/*
	 * Advance to the next pid in the array. If this goes off the
	 * end, we're done
	 */
	p++;
	if (p >= end) {
		return NULL;
	} else {
		*pos = cgroup_pid_fry(seq_css(s)->cgroup, *p);
		return p;
	}
}

static int cgroup_pidlist_show(struct seq_file *s, void *v)
{
	return seq_printf(s, "%d\n", *(int *)v);
}

/*
 * seq_operations functions for iterating on pidlists through seq_file -
 * independent of whether it's tasks or procs
 */
static const struct seq_operations cgroup_pidlist_seq_operations = {
	.start = cgroup_pidlist_start,
	.stop = cgroup_pidlist_stop,
	.next = cgroup_pidlist_next,
	.show = cgroup_pidlist_show,
};

static u64 cgroup_read_notify_on_release(struct cgroup_subsys_state *css,
					 struct cftype *cft)
{
	return notify_on_release(css->cgroup);
}

static int cgroup_write_notify_on_release(struct cgroup_subsys_state *css,
					  struct cftype *cft, u64 val)
{
	clear_bit(CGRP_RELEASABLE, &css->cgroup->flags);
	if (val)
		set_bit(CGRP_NOTIFY_ON_RELEASE, &css->cgroup->flags);
	else
		clear_bit(CGRP_NOTIFY_ON_RELEASE, &css->cgroup->flags);
	return 0;
}

static u64 cgroup_clone_children_read(struct cgroup_subsys_state *css,
				      struct cftype *cft)
{
	return test_bit(CGRP_CPUSET_CLONE_CHILDREN, &css->cgroup->flags);
}

static int cgroup_clone_children_write(struct cgroup_subsys_state *css,
				       struct cftype *cft, u64 val)
{
	if (val)
		set_bit(CGRP_CPUSET_CLONE_CHILDREN, &css->cgroup->flags);
	else
		clear_bit(CGRP_CPUSET_CLONE_CHILDREN, &css->cgroup->flags);
	return 0;
}

static struct cftype cgroup_base_files[] = {
	{
		.name = "cgroup.procs",
		.seq_start = cgroup_pidlist_start,
		.seq_next = cgroup_pidlist_next,
		.seq_stop = cgroup_pidlist_stop,
		.seq_show = cgroup_pidlist_show,
		.private = CGROUP_FILE_PROCS,
		.write_u64 = cgroup_procs_write,
		.mode = S_IRUGO | S_IWUSR,
	},
	{
		.name = "cgroup.clone_children",
		.flags = CFTYPE_INSANE,
		.read_u64 = cgroup_clone_children_read,
		.write_u64 = cgroup_clone_children_write,
	},
	{
		.name = "cgroup.sane_behavior",
		.flags = CFTYPE_ONLY_ON_ROOT,
		.seq_show = cgroup_sane_behavior_show,
	},

	/*
	 * Historical crazy stuff.  These don't have "cgroup."  prefix and
	 * don't exist if sane_behavior.  If you're depending on these, be
	 * prepared to be burned.
	 */
	{
		.name = "tasks",
		.flags = CFTYPE_INSANE,		/* use "procs" instead */
		.seq_start = cgroup_pidlist_start,
		.seq_next = cgroup_pidlist_next,
		.seq_stop = cgroup_pidlist_stop,
		.seq_show = cgroup_pidlist_show,
		.private = CGROUP_FILE_TASKS,
		.write_u64 = cgroup_tasks_write,
		.mode = S_IRUGO | S_IWUSR,
	},
	{
		.name = "notify_on_release",
		.flags = CFTYPE_INSANE,
		.read_u64 = cgroup_read_notify_on_release,
		.write_u64 = cgroup_write_notify_on_release,
	},
	{
		.name = "release_agent",
		.flags = CFTYPE_INSANE | CFTYPE_ONLY_ON_ROOT,
		.seq_show = cgroup_release_agent_show,
		.write_string = cgroup_release_agent_write,
		.max_write_len = PATH_MAX - 1,
	},
	{ }	/* terminate */
};

/**
 * cgroup_populate_dir - create subsys files in a cgroup directory
 * @cgrp: target cgroup
 * @subsys_mask: mask of the subsystem ids whose files should be added
 *
 * On failure, no file is added.
 */
static int cgroup_populate_dir(struct cgroup *cgrp, unsigned long subsys_mask)
{
	struct cgroup_subsys *ss;
	int i, ret = 0;

	/* process cftsets of each subsystem */
	for_each_subsys(ss, i) {
		struct cftype_set *set;

		if (!test_bit(i, &subsys_mask))
			continue;

		list_for_each_entry(set, &ss->cftsets, node) {
			ret = cgroup_addrm_files(cgrp, set->cfts, true);
			if (ret < 0)
				goto err;
		}
	}
	return 0;
err:
	cgroup_clear_dir(cgrp, subsys_mask);
	return ret;
}

/*
 * css destruction is four-stage process.
 *
 * 1. Destruction starts.  Killing of the percpu_ref is initiated.
 *    Implemented in kill_css().
 *
 * 2. When the percpu_ref is confirmed to be visible as killed on all CPUs
 *    and thus css_tryget() is guaranteed to fail, the css can be offlined
 *    by invoking offline_css().  After offlining, the base ref is put.
 *    Implemented in css_killed_work_fn().
 *
 * 3. When the percpu_ref reaches zero, the only possible remaining
 *    accessors are inside RCU read sections.  css_release() schedules the
 *    RCU callback.
 *
 * 4. After the grace period, the css can be freed.  Implemented in
 *    css_free_work_fn().
 *
 * It is actually hairier because both step 2 and 4 require process context
 * and thus involve punting to css->destroy_work adding two additional
 * steps to the already complex sequence.
 */
static void css_free_work_fn(struct work_struct *work)
{
	struct cgroup_subsys_state *css =
		container_of(work, struct cgroup_subsys_state, destroy_work);
	struct cgroup *cgrp = css->cgroup;

	if (css->parent)
		css_put(css->parent);

	css->ss->css_free(css);
	cgroup_put(cgrp);
}

static void css_free_rcu_fn(struct rcu_head *rcu_head)
{
	struct cgroup_subsys_state *css =
		container_of(rcu_head, struct cgroup_subsys_state, rcu_head);

	INIT_WORK(&css->destroy_work, css_free_work_fn);
	queue_work(cgroup_destroy_wq, &css->destroy_work);
}

static void css_release(struct percpu_ref *ref)
{
	struct cgroup_subsys_state *css =
		container_of(ref, struct cgroup_subsys_state, refcnt);

	rcu_assign_pointer(css->cgroup->subsys[css->ss->id], NULL);
	call_rcu(&css->rcu_head, css_free_rcu_fn);
}

static void init_css(struct cgroup_subsys_state *css, struct cgroup_subsys *ss,
		     struct cgroup *cgrp)
{
	css->cgroup = cgrp;
	css->ss = ss;
	css->flags = 0;

	if (cgrp->parent)
		css->parent = cgroup_css(cgrp->parent, ss);
	else
		css->flags |= CSS_ROOT;

	BUG_ON(cgroup_css(cgrp, ss));
}

/* invoke ->css_online() on a new CSS and mark it online if successful */
static int online_css(struct cgroup_subsys_state *css)
{
	struct cgroup_subsys *ss = css->ss;
	int ret = 0;

	lockdep_assert_held(&cgroup_tree_mutex);
	lockdep_assert_held(&cgroup_mutex);

	if (ss->css_online)
		ret = ss->css_online(css);
	if (!ret) {
		css->flags |= CSS_ONLINE;
		css->cgroup->nr_css++;
		rcu_assign_pointer(css->cgroup->subsys[ss->id], css);
	}
	return ret;
}

/* if the CSS is online, invoke ->css_offline() on it and mark it offline */
static void offline_css(struct cgroup_subsys_state *css)
{
	struct cgroup_subsys *ss = css->ss;

	lockdep_assert_held(&cgroup_tree_mutex);
	lockdep_assert_held(&cgroup_mutex);

	if (!(css->flags & CSS_ONLINE))
		return;

	if (ss->css_offline)
		ss->css_offline(css);

	css->flags &= ~CSS_ONLINE;
	css->cgroup->nr_css--;
	RCU_INIT_POINTER(css->cgroup->subsys[ss->id], css);
}

/**
 * create_css - create a cgroup_subsys_state
 * @cgrp: the cgroup new css will be associated with
 * @ss: the subsys of new css
 *
 * Create a new css associated with @cgrp - @ss pair.  On success, the new
 * css is online and installed in @cgrp with all interface files created.
 * Returns 0 on success, -errno on failure.
 */
static int create_css(struct cgroup *cgrp, struct cgroup_subsys *ss)
{
	struct cgroup *parent = cgrp->parent;
	struct cgroup_subsys_state *css;
	int err;

	lockdep_assert_held(&cgroup_mutex);

	css = ss->css_alloc(cgroup_css(parent, ss));
	if (IS_ERR(css))
		return PTR_ERR(css);

	err = percpu_ref_init(&css->refcnt, css_release);
	if (err)
		goto err_free;

	init_css(css, ss, cgrp);

	err = cgroup_populate_dir(cgrp, 1 << ss->id);
	if (err)
		goto err_free;

	err = online_css(css);
	if (err)
		goto err_free;

	cgroup_get(cgrp);
	css_get(css->parent);

	if (ss->broken_hierarchy && !ss->warned_broken_hierarchy &&
	    parent->parent) {
		pr_warning("cgroup: %s (%d) created nested cgroup for controller \"%s\" which has incomplete hierarchy support. Nested cgroups may change behavior in the future.\n",
			   current->comm, current->pid, ss->name);
		if (!strcmp(ss->name, "memory"))
			pr_warning("cgroup: \"memory\" requires setting use_hierarchy to 1 on the root.\n");
		ss->warned_broken_hierarchy = true;
	}

	return 0;

err_free:
	percpu_ref_cancel_init(&css->refcnt);
	ss->css_free(css);
	return err;
}

/**
 * cgroup_create - create a cgroup
 * @parent: cgroup that will be parent of the new cgroup
 * @name_str: name of the new cgroup
 * @mode: mode to set on new cgroup
 */
static long cgroup_create(struct cgroup *parent, const char *name_str,
			  umode_t mode)
{
	struct cgroup *cgrp;
	struct cgroup_name *name;
	struct cgroupfs_root *root = parent->root;
	int ssid, err;
	struct cgroup_subsys *ss;
	struct kernfs_node *kn;

	/* allocate the cgroup and its ID, 0 is reserved for the root */
	cgrp = kzalloc(sizeof(*cgrp), GFP_KERNEL);
	if (!cgrp)
		return -ENOMEM;

	name = cgroup_alloc_name(name_str);
	if (!name) {
		err = -ENOMEM;
		goto err_free_cgrp;
	}
	rcu_assign_pointer(cgrp->name, name);

	mutex_lock(&cgroup_tree_mutex);

	/*
	 * Only live parents can have children.  Note that the liveliness
	 * check isn't strictly necessary because cgroup_mkdir() and
	 * cgroup_rmdir() are fully synchronized by i_mutex; however, do it
	 * anyway so that locking is contained inside cgroup proper and we
	 * don't get nasty surprises if we ever grow another caller.
	 */
	if (!cgroup_lock_live_group(parent)) {
		err = -ENODEV;
		goto err_unlock_tree;
	}

	/*
	 * Temporarily set the pointer to NULL, so idr_find() won't return
	 * a half-baked cgroup.
	 */
	cgrp->id = idr_alloc(&root->cgroup_idr, NULL, 1, 0, GFP_KERNEL);
	if (cgrp->id < 0) {
		err = -ENOMEM;
		goto err_unlock;
	}

	init_cgroup_housekeeping(cgrp);

	cgrp->parent = parent;
	cgrp->dummy_css.parent = &parent->dummy_css;
	cgrp->root = parent->root;

	if (notify_on_release(parent))
		set_bit(CGRP_NOTIFY_ON_RELEASE, &cgrp->flags);

	if (test_bit(CGRP_CPUSET_CLONE_CHILDREN, &parent->flags))
		set_bit(CGRP_CPUSET_CLONE_CHILDREN, &cgrp->flags);

	/* create the directory */
	kn = kernfs_create_dir(parent->kn, name->name, mode, cgrp);
	if (IS_ERR(kn)) {
		err = PTR_ERR(kn);
		goto err_free_id;
	}
	cgrp->kn = kn;

	cgrp->serial_nr = cgroup_serial_nr_next++;

	/* allocation complete, commit to creation */
	list_add_tail_rcu(&cgrp->sibling, &cgrp->parent->children);
	root->number_of_cgroups++;

	/*
	 * Grab a reference on the root and parent so that they don't get
	 * deleted while there are child cgroups.
	 */
	cgroup_get_root(root);
	cgroup_get(parent);

	/*
	 * @cgrp is now fully operational.  If something fails after this
	 * point, it'll be released via the normal destruction path.
	 */
	idr_replace(&root->cgroup_idr, cgrp, cgrp->id);

	err = cgroup_addrm_files(cgrp, cgroup_base_files, true);
	if (err)
		goto err_destroy;

	/* let's create and online css's */
	for_each_subsys(ss, ssid) {
		if (root->subsys_mask & (1 << ssid)) {
			err = create_css(cgrp, ss);
			if (err)
				goto err_destroy;
		}
	}

	kernfs_activate(kn);

	mutex_unlock(&cgroup_mutex);
	mutex_unlock(&cgroup_tree_mutex);

	return 0;

err_free_id:
	idr_remove(&root->cgroup_idr, cgrp->id);
err_unlock:
	mutex_unlock(&cgroup_mutex);
err_unlock_tree:
	mutex_unlock(&cgroup_tree_mutex);
	kfree(rcu_dereference_raw(cgrp->name));
err_free_cgrp:
	kfree(cgrp);
	return err;

err_destroy:
	cgroup_destroy_locked(cgrp);
	mutex_unlock(&cgroup_mutex);
	mutex_unlock(&cgroup_tree_mutex);
	return err;
}

static int cgroup_mkdir(struct kernfs_node *parent_kn, const char *name,
			umode_t mode)
{
	struct cgroup *parent = parent_kn->priv;

	return cgroup_create(parent, name, mode);
}

/*
 * This is called when the refcnt of a css is confirmed to be killed.
 * css_tryget() is now guaranteed to fail.
 */
static void css_killed_work_fn(struct work_struct *work)
{
	struct cgroup_subsys_state *css =
		container_of(work, struct cgroup_subsys_state, destroy_work);
	struct cgroup *cgrp = css->cgroup;

	mutex_lock(&cgroup_tree_mutex);
	mutex_lock(&cgroup_mutex);

	/*
	 * css_tryget() is guaranteed to fail now.  Tell subsystems to
	 * initate destruction.
	 */
	offline_css(css);

	/*
	 * If @cgrp is marked dead, it's waiting for refs of all css's to
	 * be disabled before proceeding to the second phase of cgroup
	 * destruction.  If we are the last one, kick it off.
	 */
	if (!cgrp->nr_css && cgroup_is_dead(cgrp))
		cgroup_destroy_css_killed(cgrp);

	mutex_unlock(&cgroup_mutex);
	mutex_unlock(&cgroup_tree_mutex);

	/*
	 * Put the css refs from kill_css().  Each css holds an extra
	 * reference to the cgroup's dentry and cgroup removal proceeds
	 * regardless of css refs.  On the last put of each css, whenever
	 * that may be, the extra dentry ref is put so that dentry
	 * destruction happens only after all css's are released.
	 */
	css_put(css);
}

/* css kill confirmation processing requires process context, bounce */
static void css_killed_ref_fn(struct percpu_ref *ref)
{
	struct cgroup_subsys_state *css =
		container_of(ref, struct cgroup_subsys_state, refcnt);

	INIT_WORK(&css->destroy_work, css_killed_work_fn);
	queue_work(cgroup_destroy_wq, &css->destroy_work);
}

/**
 * kill_css - destroy a css
 * @css: css to destroy
 *
 * This function initiates destruction of @css by removing cgroup interface
 * files and putting its base reference.  ->css_offline() will be invoked
 * asynchronously once css_tryget() is guaranteed to fail and when the
 * reference count reaches zero, @css will be released.
 */
static void kill_css(struct cgroup_subsys_state *css)
{
	/*
	 * This must happen before css is disassociated with its cgroup.
	 * See seq_css() for details.
	 */
	cgroup_clear_dir(css->cgroup, 1 << css->ss->id);

	/*
	 * Killing would put the base ref, but we need to keep it alive
	 * until after ->css_offline().
	 */
	css_get(css);

	/*
	 * cgroup core guarantees that, by the time ->css_offline() is
	 * invoked, no new css reference will be given out via
	 * css_tryget().  We can't simply call percpu_ref_kill() and
	 * proceed to offlining css's because percpu_ref_kill() doesn't
	 * guarantee that the ref is seen as killed on all CPUs on return.
	 *
	 * Use percpu_ref_kill_and_confirm() to get notifications as each
	 * css is confirmed to be seen as killed on all CPUs.
	 */
	percpu_ref_kill_and_confirm(&css->refcnt, css_killed_ref_fn);
}

/**
 * cgroup_destroy_locked - the first stage of cgroup destruction
 * @cgrp: cgroup to be destroyed
 *
 * css's make use of percpu refcnts whose killing latency shouldn't be
 * exposed to userland and are RCU protected.  Also, cgroup core needs to
 * guarantee that css_tryget() won't succeed by the time ->css_offline() is
 * invoked.  To satisfy all the requirements, destruction is implemented in
 * the following two steps.
 *
 * s1. Verify @cgrp can be destroyed and mark it dying.  Remove all
 *     userland visible parts and start killing the percpu refcnts of
 *     css's.  Set up so that the next stage will be kicked off once all
 *     the percpu refcnts are confirmed to be killed.
 *
 * s2. Invoke ->css_offline(), mark the cgroup dead and proceed with the
 *     rest of destruction.  Once all cgroup references are gone, the
 *     cgroup is RCU-freed.
 *
 * This function implements s1.  After this step, @cgrp is gone as far as
 * the userland is concerned and a new cgroup with the same name may be
 * created.  As cgroup doesn't care about the names internally, this
 * doesn't cause any problem.