[deliverable/linux.git] / mm / hugetlb_cgroup.c

/*
 *
 * Copyright IBM Corporation, 2012
 * Author Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms of version 2.1 of the GNU Lesser General Public License
 * as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it would be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
 *
 */

#include <linux/cgroup.h>
#include <linux/slab.h>
#include <linux/hugetlb.h>
#include <linux/hugetlb_cgroup.h>

struct hugetlb_cgroup {
	struct cgroup_subsys_state css;
	/*
	 * the counter to account for hugepages from hugetlb.
	 */
	struct res_counter hugepage[HUGE_MAX_HSTATE];
};

#define MEMFILE_PRIVATE(x, val)	(((x) << 16) | (val))
#define MEMFILE_IDX(val)	(((val) >> 16) & 0xffff)
#define MEMFILE_ATTR(val)	((val) & 0xffff)

static struct hugetlb_cgroup *root_h_cgroup __read_mostly;

static inline
struct hugetlb_cgroup *hugetlb_cgroup_from_css(struct cgroup_subsys_state *s)
{
	return s ? container_of(s, struct hugetlb_cgroup, css) : NULL;
}

static inline
struct hugetlb_cgroup *hugetlb_cgroup_from_task(struct task_struct *task)
{
	return hugetlb_cgroup_from_css(task_css(task, hugetlb_cgrp_id));
}

static inline bool hugetlb_cgroup_is_root(struct hugetlb_cgroup *h_cg)
{
	return (h_cg == root_h_cgroup);
}

static inline struct hugetlb_cgroup *
parent_hugetlb_cgroup(struct hugetlb_cgroup *h_cg)
{
	return hugetlb_cgroup_from_css(h_cg->css.parent);
}

static inline bool hugetlb_cgroup_have_usage(struct hugetlb_cgroup *h_cg)
{
	int idx;

	for (idx = 0; idx < hugetlb_max_hstate; idx++) {
		if ((res_counter_read_u64(&h_cg->hugepage[idx], RES_USAGE)) > 0)
			return true;
	}
	return false;
}

static struct cgroup_subsys_state *
hugetlb_cgroup_css_alloc(struct cgroup_subsys_state *parent_css)
{
	struct hugetlb_cgroup *parent_h_cgroup = hugetlb_cgroup_from_css(parent_css);
	struct hugetlb_cgroup *h_cgroup;
	int idx;

	h_cgroup = kzalloc(sizeof(*h_cgroup), GFP_KERNEL);
	if (!h_cgroup)
		return ERR_PTR(-ENOMEM);

	if (parent_h_cgroup) {
		for (idx = 0; idx < HUGE_MAX_HSTATE; idx++)
			res_counter_init(&h_cgroup->hugepage[idx],
					 &parent_h_cgroup->hugepage[idx]);
	} else {
		root_h_cgroup = h_cgroup;
		for (idx = 0; idx < HUGE_MAX_HSTATE; idx++)
			res_counter_init(&h_cgroup->hugepage[idx], NULL);
	}
	return &h_cgroup->css;
}

static void hugetlb_cgroup_css_free(struct cgroup_subsys_state *css)
{
	struct hugetlb_cgroup *h_cgroup;

	h_cgroup = hugetlb_cgroup_from_css(css);
	kfree(h_cgroup);
}


/*
 * Should be called with hugetlb_lock held.
 * Since we are holding hugetlb_lock, pages cannot get moved from
 * active list or uncharged from the cgroup, So no need to get
 * page reference and test for page active here. This function
 * cannot fail.
 */
static void hugetlb_cgroup_move_parent(int idx, struct hugetlb_cgroup *h_cg,
				       struct page *page)
{
	int csize;
	struct res_counter *counter;
	struct res_counter *fail_res;
	struct hugetlb_cgroup *page_hcg;
	struct hugetlb_cgroup *parent = parent_hugetlb_cgroup(h_cg);

	page_hcg = hugetlb_cgroup_from_page(page);
	/*
	 * We can have pages in active list without any cgroup
	 * ie, hugepage with less than 3 pages. We can safely
	 * ignore those pages.
	 */
	if (!page_hcg || page_hcg != h_cg)
		goto out;

	csize = PAGE_SIZE << compound_order(page);
	if (!parent) {
		parent = root_h_cgroup;
		/* root has no limit */
		res_counter_charge_nofail(&parent->hugepage[idx],
					  csize, &fail_res);
	}
	counter = &h_cg->hugepage[idx];
	res_counter_uncharge_until(counter, counter->parent, csize);

	set_hugetlb_cgroup(page, parent);
out:
	return;
}

/*
 * Force the hugetlb cgroup to empty the hugetlb resources by moving them to
 * the parent cgroup.
 */
static void hugetlb_cgroup_css_offline(struct cgroup_subsys_state *css)
{
	struct hugetlb_cgroup *h_cg = hugetlb_cgroup_from_css(css);
	struct hstate *h;
	struct page *page;
	int idx = 0;

	do {
		for_each_hstate(h) {
			spin_lock(&hugetlb_lock);
			list_for_each_entry(page, &h->hugepage_activelist, lru)
				hugetlb_cgroup_move_parent(idx, h_cg, page);

			spin_unlock(&hugetlb_lock);
			idx++;
		}
		cond_resched();
	} while (hugetlb_cgroup_have_usage(h_cg));
}

int hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages,
				 struct hugetlb_cgroup **ptr)
{
	int ret = 0;
	struct res_counter *fail_res;
	struct hugetlb_cgroup *h_cg = NULL;
	unsigned long csize = nr_pages * PAGE_SIZE;

	if (hugetlb_cgroup_disabled())
		goto done;
	/*
	 * We don't charge any cgroup if the compound page have less
	 * than 3 pages.
	 */
	if (huge_page_order(&hstates[idx]) < HUGETLB_CGROUP_MIN_ORDER)
		goto done;
again:
	rcu_read_lock();
	h_cg = hugetlb_cgroup_from_task(current);
	if (!css_tryget_online(&h_cg->css)) {
		rcu_read_unlock();
		goto again;
	}
	rcu_read_unlock();

	ret = res_counter_charge(&h_cg->hugepage[idx], csize, &fail_res);
	css_put(&h_cg->css);
done:
	*ptr = h_cg;
	return ret;
}

/* Should be called with hugetlb_lock held */
void hugetlb_cgroup_commit_charge(int idx, unsigned long nr_pages,
				  struct hugetlb_cgroup *h_cg,
				  struct page *page)
{
	if (hugetlb_cgroup_disabled() || !h_cg)
		return;

	set_hugetlb_cgroup(page, h_cg);
	return;
}

/*
 * Should be called with hugetlb_lock held
 */
void hugetlb_cgroup_uncharge_page(int idx, unsigned long nr_pages,
				  struct page *page)
{
	struct hugetlb_cgroup *h_cg;
	unsigned long csize = nr_pages * PAGE_SIZE;

	if (hugetlb_cgroup_disabled())
		return;
	VM_BUG_ON(!spin_is_locked(&hugetlb_lock));
	h_cg = hugetlb_cgroup_from_page(page);
	if (unlikely(!h_cg))
		return;
	set_hugetlb_cgroup(page, NULL);
	res_counter_uncharge(&h_cg->hugepage[idx], csize);
	return;
}

void hugetlb_cgroup_uncharge_cgroup(int idx, unsigned long nr_pages,
				    struct hugetlb_cgroup *h_cg)
{
	unsigned long csize = nr_pages * PAGE_SIZE;

	if (hugetlb_cgroup_disabled() || !h_cg)
		return;

	if (huge_page_order(&hstates[idx]) < HUGETLB_CGROUP_MIN_ORDER)
		return;

	res_counter_uncharge(&h_cg->hugepage[idx], csize);
	return;
}

static u64 hugetlb_cgroup_read_u64(struct cgroup_subsys_state *css,
				   struct cftype *cft)
{
	int idx, name;
	struct hugetlb_cgroup *h_cg = hugetlb_cgroup_from_css(css);

	idx = MEMFILE_IDX(cft->private);
	name = MEMFILE_ATTR(cft->private);

	return res_counter_read_u64(&h_cg->hugepage[idx], name);
}

static ssize_t hugetlb_cgroup_write(struct kernfs_open_file *of,
				    char *buf, size_t nbytes, loff_t off)
{
	int idx, name, ret;
	unsigned long long val;
	struct hugetlb_cgroup *h_cg = hugetlb_cgroup_from_css(of_css(of));

	buf = strstrip(buf);
	idx = MEMFILE_IDX(of_cft(of)->private);
	name = MEMFILE_ATTR(of_cft(of)->private);

	switch (name) {
	case RES_LIMIT:
		if (hugetlb_cgroup_is_root(h_cg)) {
			/* Can't set limit on root */
			ret = -EINVAL;
			break;
		}
		/* This function does all necessary parse...reuse it */
		ret = res_counter_memparse_write_strategy(buf, &val);
		if (ret)
			break;
		ret = res_counter_set_limit(&h_cg->hugepage[idx], val);
		break;
	default:
		ret = -EINVAL;
		break;
	}
	return ret ?: nbytes;
}

static ssize_t hugetlb_cgroup_reset(struct kernfs_open_file *of,
				    char *buf, size_t nbytes, loff_t off)
{
	int idx, name, ret = 0;
	struct hugetlb_cgroup *h_cg = hugetlb_cgroup_from_css(of_css(of));

	idx = MEMFILE_IDX(of_cft(of)->private);
	name = MEMFILE_ATTR(of_cft(of)->private);

	switch (name) {
	case RES_MAX_USAGE:
		res_counter_reset_max(&h_cg->hugepage[idx]);
		break;
	case RES_FAILCNT:
		res_counter_reset_failcnt(&h_cg->hugepage[idx]);
		break;
	default:
		ret = -EINVAL;
		break;
	}
	return ret ?: nbytes;
}

static char *mem_fmt(char *buf, int size, unsigned long hsize)
{
	if (hsize >= (1UL << 30))
		snprintf(buf, size, "%luGB", hsize >> 30);
	else if (hsize >= (1UL << 20))
		snprintf(buf, size, "%luMB", hsize >> 20);
	else
		snprintf(buf, size, "%luKB", hsize >> 10);
	return buf;
}

static void __init __hugetlb_cgroup_file_init(int idx)
{
	char buf[32];
	struct cftype *cft;
	struct hstate *h = &hstates[idx];

	/* format the size */
	mem_fmt(buf, 32, huge_page_size(h));

	/* Add the limit file */
	cft = &h->cgroup_files[0];
	snprintf(cft->name, MAX_CFTYPE_NAME, "%s.limit_in_bytes", buf);
	cft->private = MEMFILE_PRIVATE(idx, RES_LIMIT);
	cft->read_u64 = hugetlb_cgroup_read_u64;
	cft->write = hugetlb_cgroup_write;

	/* Add the usage file */
	cft = &h->cgroup_files[1];
	snprintf(cft->name, MAX_CFTYPE_NAME, "%s.usage_in_bytes", buf);
	cft->private = MEMFILE_PRIVATE(idx, RES_USAGE);
	cft->read_u64 = hugetlb_cgroup_read_u64;

	/* Add the MAX usage file */
	cft = &h->cgroup_files[2];
	snprintf(cft->name, MAX_CFTYPE_NAME, "%s.max_usage_in_bytes", buf);
	cft->private = MEMFILE_PRIVATE(idx, RES_MAX_USAGE);
	cft->write = hugetlb_cgroup_reset;
	cft->read_u64 = hugetlb_cgroup_read_u64;

	/* Add the failcntfile */
	cft = &h->cgroup_files[3];
	snprintf(cft->name, MAX_CFTYPE_NAME, "%s.failcnt", buf);
	cft->private  = MEMFILE_PRIVATE(idx, RES_FAILCNT);
	cft->write = hugetlb_cgroup_reset;
	cft->read_u64 = hugetlb_cgroup_read_u64;

	/* NULL terminate the last cft */
	cft = &h->cgroup_files[4];
	memset(cft, 0, sizeof(*cft));

	WARN_ON(cgroup_add_cftypes(&hugetlb_cgrp_subsys, h->cgroup_files));

	return;
}

void __init hugetlb_cgroup_file_init(void)
{
	struct hstate *h;

	for_each_hstate(h) {
		/*
		 * Add cgroup control files only if the huge page consists
		 * of more than two normal pages. This is because we use
		 * page[2].lru.next for storing cgroup details.
		 */
		if (huge_page_order(h) >= HUGETLB_CGROUP_MIN_ORDER)
			__hugetlb_cgroup_file_init(hstate_index(h));
	}
}

/*
 * hugetlb_lock will make sure a parallel cgroup rmdir won't happen
 * when we migrate hugepages
 */
void hugetlb_cgroup_migrate(struct page *oldhpage, struct page *newhpage)
{
	struct hugetlb_cgroup *h_cg;
	struct hstate *h = page_hstate(oldhpage);

	if (hugetlb_cgroup_disabled())
		return;

	VM_BUG_ON_PAGE(!PageHuge(oldhpage), oldhpage);
	spin_lock(&hugetlb_lock);
	h_cg = hugetlb_cgroup_from_page(oldhpage);
	set_hugetlb_cgroup(oldhpage, NULL);

	/* move the h_cg details to new cgroup */
	set_hugetlb_cgroup(newhpage, h_cg);
	list_move(&newhpage->lru, &h->hugepage_activelist);
	spin_unlock(&hugetlb_lock);
	return;
}

struct cgroup_subsys hugetlb_cgrp_subsys = {
	.css_alloc	= hugetlb_cgroup_css_alloc,
	.css_offline	= hugetlb_cgroup_css_offline,
	.css_free	= hugetlb_cgroup_css_free,
};
Commit	Line	Data
	1	/*
	2	*
	3	* Copyright IBM Corporation, 2012
	4	* Author Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
	5	*
	6	* This program is free software; you can redistribute it and/or modify it
	7	* under the terms of version 2.1 of the GNU Lesser General Public License
	8	* as published by the Free Software Foundation.
	9	*
	10	* This program is distributed in the hope that it would be useful, but
	11	* WITHOUT ANY WARRANTY; without even the implied warranty of
	12	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
	13	*
	14	*/
	15
	16	#include <linux/cgroup.h>
	17	#include <linux/slab.h>
	18	#include <linux/hugetlb.h>
	19	#include <linux/hugetlb_cgroup.h>
	20
	21	struct hugetlb_cgroup {
	22	struct cgroup_subsys_state css;
	23	/*
	24	* the counter to account for hugepages from hugetlb.
	25	*/
	26	struct res_counter hugepage[HUGE_MAX_HSTATE];
	27	};
	28
	29	#define MEMFILE_PRIVATE(x, val) (((x) << 16) \| (val))
	30	#define MEMFILE_IDX(val) (((val) >> 16) & 0xffff)
	31	#define MEMFILE_ATTR(val) ((val) & 0xffff)
	32
	33	static struct hugetlb_cgroup *root_h_cgroup __read_mostly;
	34
	35	static inline
	36	struct hugetlb_cgroup hugetlb_cgroup_from_css(struct cgroup_subsys_state s)
	37	{
	38	return s ? container_of(s, struct hugetlb_cgroup, css) : NULL;
	39	}
	40
	41	static inline
	42	struct hugetlb_cgroup hugetlb_cgroup_from_task(struct task_struct task)
	43	{
	44	return hugetlb_cgroup_from_css(task_css(task, hugetlb_cgrp_id));
	45	}
	46
	47	static inline bool hugetlb_cgroup_is_root(struct hugetlb_cgroup *h_cg)
	48	{
	49	return (h_cg == root_h_cgroup);
	50	}
	51
	52	static inline struct hugetlb_cgroup *
	53	parent_hugetlb_cgroup(struct hugetlb_cgroup *h_cg)
	54	{
	55	return hugetlb_cgroup_from_css(h_cg->css.parent);
	56	}
	57
	58	static inline bool hugetlb_cgroup_have_usage(struct hugetlb_cgroup *h_cg)
	59	{
	60	int idx;
	61
	62	for (idx = 0; idx < hugetlb_max_hstate; idx++) {
	63	if ((res_counter_read_u64(&h_cg->hugepage[idx], RES_USAGE)) > 0)
	64	return true;
	65	}
	66	return false;
	67	}
	68
	69	static struct cgroup_subsys_state *
	70	hugetlb_cgroup_css_alloc(struct cgroup_subsys_state *parent_css)
	71	{
	72	struct hugetlb_cgroup *parent_h_cgroup = hugetlb_cgroup_from_css(parent_css);
	73	struct hugetlb_cgroup *h_cgroup;
	74	int idx;
	75
	76	h_cgroup = kzalloc(sizeof(*h_cgroup), GFP_KERNEL);
	77	if (!h_cgroup)
	78	return ERR_PTR(-ENOMEM);
	79
	80	if (parent_h_cgroup) {
	81	for (idx = 0; idx < HUGE_MAX_HSTATE; idx++)
	82	res_counter_init(&h_cgroup->hugepage[idx],
	83	&parent_h_cgroup->hugepage[idx]);
	84	} else {
	85	root_h_cgroup = h_cgroup;
	86	for (idx = 0; idx < HUGE_MAX_HSTATE; idx++)
	87	res_counter_init(&h_cgroup->hugepage[idx], NULL);
	88	}
	89	return &h_cgroup->css;
	90	}
	91
	92	static void hugetlb_cgroup_css_free(struct cgroup_subsys_state *css)
	93	{
	94	struct hugetlb_cgroup *h_cgroup;
	95
	96	h_cgroup = hugetlb_cgroup_from_css(css);
	97	kfree(h_cgroup);
	98	}
	99
	100
	101	/*
	102	* Should be called with hugetlb_lock held.
	103	* Since we are holding hugetlb_lock, pages cannot get moved from
	104	* active list or uncharged from the cgroup, So no need to get
	105	* page reference and test for page active here. This function
	106	* cannot fail.
	107	*/
	108	static void hugetlb_cgroup_move_parent(int idx, struct hugetlb_cgroup *h_cg,
	109	struct page *page)
	110	{
	111	int csize;
	112	struct res_counter *counter;
	113	struct res_counter *fail_res;
	114	struct hugetlb_cgroup *page_hcg;
	115	struct hugetlb_cgroup *parent = parent_hugetlb_cgroup(h_cg);
	116
	117	page_hcg = hugetlb_cgroup_from_page(page);
	118	/*
	119	* We can have pages in active list without any cgroup
	120	* ie, hugepage with less than 3 pages. We can safely
	121	* ignore those pages.
	122	*/
	123	if (!page_hcg \|\| page_hcg != h_cg)
	124	goto out;
	125
	126	csize = PAGE_SIZE << compound_order(page);
	127	if (!parent) {
	128	parent = root_h_cgroup;
	129	/* root has no limit */
	130	res_counter_charge_nofail(&parent->hugepage[idx],
	131	csize, &fail_res);
	132	}
	133	counter = &h_cg->hugepage[idx];
	134	res_counter_uncharge_until(counter, counter->parent, csize);
	135
	136	set_hugetlb_cgroup(page, parent);
	137	out:
	138	return;
	139	}
	140
	141	/*
	142	* Force the hugetlb cgroup to empty the hugetlb resources by moving them to
	143	* the parent cgroup.
	144	*/
	145	static void hugetlb_cgroup_css_offline(struct cgroup_subsys_state *css)
	146	{
	147	struct hugetlb_cgroup *h_cg = hugetlb_cgroup_from_css(css);
	148	struct hstate *h;
	149	struct page *page;
	150	int idx = 0;
	151
	152	do {
	153	for_each_hstate(h) {
	154	spin_lock(&hugetlb_lock);
	155	list_for_each_entry(page, &h->hugepage_activelist, lru)
	156	hugetlb_cgroup_move_parent(idx, h_cg, page);
	157
	158	spin_unlock(&hugetlb_lock);
	159	idx++;
	160	}
	161	cond_resched();
	162	} while (hugetlb_cgroup_have_usage(h_cg));
	163	}
	164
	165	int hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages,
	166	struct hugetlb_cgroup **ptr)
	167	{
	168	int ret = 0;
	169	struct res_counter *fail_res;
	170	struct hugetlb_cgroup *h_cg = NULL;
	171	unsigned long csize = nr_pages * PAGE_SIZE;
	172
	173	if (hugetlb_cgroup_disabled())
	174	goto done;
	175	/*
	176	* We don't charge any cgroup if the compound page have less
	177	* than 3 pages.
	178	*/
	179	if (huge_page_order(&hstates[idx]) < HUGETLB_CGROUP_MIN_ORDER)
	180	goto done;
	181	again:
	182	rcu_read_lock();
	183	h_cg = hugetlb_cgroup_from_task(current);
	184	if (!css_tryget_online(&h_cg->css)) {
	185	rcu_read_unlock();
	186	goto again;
	187	}
	188	rcu_read_unlock();
	189
	190	ret = res_counter_charge(&h_cg->hugepage[idx], csize, &fail_res);
	191	css_put(&h_cg->css);
	192	done:
	193	*ptr = h_cg;
	194	return ret;
	195	}
	196
	197	/* Should be called with hugetlb_lock held */
	198	void hugetlb_cgroup_commit_charge(int idx, unsigned long nr_pages,
	199	struct hugetlb_cgroup *h_cg,
	200	struct page *page)
	201	{
	202	if (hugetlb_cgroup_disabled() \|\| !h_cg)
	203	return;
	204
	205	set_hugetlb_cgroup(page, h_cg);
	206	return;
	207	}
	208
	209	/*
	210	* Should be called with hugetlb_lock held
	211	*/
	212	void hugetlb_cgroup_uncharge_page(int idx, unsigned long nr_pages,
	213	struct page *page)
	214	{
	215	struct hugetlb_cgroup *h_cg;
	216	unsigned long csize = nr_pages * PAGE_SIZE;
	217
	218	if (hugetlb_cgroup_disabled())
	219	return;
	220	VM_BUG_ON(!spin_is_locked(&hugetlb_lock));
	221	h_cg = hugetlb_cgroup_from_page(page);
	222	if (unlikely(!h_cg))
	223	return;
	224	set_hugetlb_cgroup(page, NULL);
	225	res_counter_uncharge(&h_cg->hugepage[idx], csize);
	226	return;
	227	}
	228
	229	void hugetlb_cgroup_uncharge_cgroup(int idx, unsigned long nr_pages,
	230	struct hugetlb_cgroup *h_cg)
	231	{
	232	unsigned long csize = nr_pages * PAGE_SIZE;
	233
	234	if (hugetlb_cgroup_disabled() \|\| !h_cg)
	235	return;
	236
	237	if (huge_page_order(&hstates[idx]) < HUGETLB_CGROUP_MIN_ORDER)
	238	return;
	239
	240	res_counter_uncharge(&h_cg->hugepage[idx], csize);
	241	return;
	242	}
	243
	244	static u64 hugetlb_cgroup_read_u64(struct cgroup_subsys_state *css,
	245	struct cftype *cft)
	246	{
	247	int idx, name;
	248	struct hugetlb_cgroup *h_cg = hugetlb_cgroup_from_css(css);
	249
	250	idx = MEMFILE_IDX(cft->private);
	251	name = MEMFILE_ATTR(cft->private);
	252
	253	return res_counter_read_u64(&h_cg->hugepage[idx], name);
	254	}
	255
	256	static ssize_t hugetlb_cgroup_write(struct kernfs_open_file *of,
	257	char *buf, size_t nbytes, loff_t off)
	258	{
	259	int idx, name, ret;
	260	unsigned long long val;
	261	struct hugetlb_cgroup *h_cg = hugetlb_cgroup_from_css(of_css(of));
	262
	263	buf = strstrip(buf);
	264	idx = MEMFILE_IDX(of_cft(of)->private);
	265	name = MEMFILE_ATTR(of_cft(of)->private);
	266
	267	switch (name) {
	268	case RES_LIMIT:
	269	if (hugetlb_cgroup_is_root(h_cg)) {
	270	/* Can't set limit on root */
	271	ret = -EINVAL;
	272	break;
	273	}
	274	/* This function does all necessary parse...reuse it */
	275	ret = res_counter_memparse_write_strategy(buf, &val);
	276	if (ret)
	277	break;
	278	ret = res_counter_set_limit(&h_cg->hugepage[idx], val);
	279	break;
	280	default:
	281	ret = -EINVAL;
	282	break;
	283	}
	284	return ret ?: nbytes;
	285	}
	286
	287	static ssize_t hugetlb_cgroup_reset(struct kernfs_open_file *of,
	288	char *buf, size_t nbytes, loff_t off)
	289	{
	290	int idx, name, ret = 0;
	291	struct hugetlb_cgroup *h_cg = hugetlb_cgroup_from_css(of_css(of));
	292
	293	idx = MEMFILE_IDX(of_cft(of)->private);
	294	name = MEMFILE_ATTR(of_cft(of)->private);
	295
	296	switch (name) {
	297	case RES_MAX_USAGE:
	298	res_counter_reset_max(&h_cg->hugepage[idx]);
	299	break;
	300	case RES_FAILCNT:
	301	res_counter_reset_failcnt(&h_cg->hugepage[idx]);
	302	break;
	303	default:
	304	ret = -EINVAL;
	305	break;
	306	}
	307	return ret ?: nbytes;
	308	}
	309
	310	static char mem_fmt(char buf, int size, unsigned long hsize)
	311	{
	312	if (hsize >= (1UL << 30))
	313	snprintf(buf, size, "%luGB", hsize >> 30);
	314	else if (hsize >= (1UL << 20))
	315	snprintf(buf, size, "%luMB", hsize >> 20);
	316	else
	317	snprintf(buf, size, "%luKB", hsize >> 10);
	318	return buf;
	319	}
	320
	321	static void __init __hugetlb_cgroup_file_init(int idx)
	322	{
	323	char buf[32];
	324	struct cftype *cft;
	325	struct hstate *h = &hstates[idx];
	326
	327	/* format the size */
	328	mem_fmt(buf, 32, huge_page_size(h));
	329
	330	/* Add the limit file */
	331	cft = &h->cgroup_files[0];
	332	snprintf(cft->name, MAX_CFTYPE_NAME, "%s.limit_in_bytes", buf);
	333	cft->private = MEMFILE_PRIVATE(idx, RES_LIMIT);
	334	cft->read_u64 = hugetlb_cgroup_read_u64;
	335	cft->write = hugetlb_cgroup_write;
	336
	337	/* Add the usage file */
	338	cft = &h->cgroup_files[1];
	339	snprintf(cft->name, MAX_CFTYPE_NAME, "%s.usage_in_bytes", buf);
	340	cft->private = MEMFILE_PRIVATE(idx, RES_USAGE);
	341	cft->read_u64 = hugetlb_cgroup_read_u64;
	342
	343	/* Add the MAX usage file */
	344	cft = &h->cgroup_files[2];
	345	snprintf(cft->name, MAX_CFTYPE_NAME, "%s.max_usage_in_bytes", buf);
	346	cft->private = MEMFILE_PRIVATE(idx, RES_MAX_USAGE);
	347	cft->write = hugetlb_cgroup_reset;
	348	cft->read_u64 = hugetlb_cgroup_read_u64;
	349
	350	/* Add the failcntfile */
	351	cft = &h->cgroup_files[3];
	352	snprintf(cft->name, MAX_CFTYPE_NAME, "%s.failcnt", buf);
	353	cft->private = MEMFILE_PRIVATE(idx, RES_FAILCNT);
	354	cft->write = hugetlb_cgroup_reset;
	355	cft->read_u64 = hugetlb_cgroup_read_u64;
	356
	357	/* NULL terminate the last cft */
	358	cft = &h->cgroup_files[4];
	359	memset(cft, 0, sizeof(*cft));
	360
	361	WARN_ON(cgroup_add_cftypes(&hugetlb_cgrp_subsys, h->cgroup_files));
	362
	363	return;
	364	}
	365
	366	void __init hugetlb_cgroup_file_init(void)
	367	{
	368	struct hstate *h;
	369
	370	for_each_hstate(h) {
	371	/*
	372	* Add cgroup control files only if the huge page consists
	373	* of more than two normal pages. This is because we use
	374	* page[2].lru.next for storing cgroup details.
	375	*/
	376	if (huge_page_order(h) >= HUGETLB_CGROUP_MIN_ORDER)
	377	__hugetlb_cgroup_file_init(hstate_index(h));
	378	}
	379	}
	380
	381	/*
	382	* hugetlb_lock will make sure a parallel cgroup rmdir won't happen
	383	* when we migrate hugepages
	384	*/
	385	void hugetlb_cgroup_migrate(struct page oldhpage, struct page newhpage)
	386	{
	387	struct hugetlb_cgroup *h_cg;
	388	struct hstate *h = page_hstate(oldhpage);
	389
	390	if (hugetlb_cgroup_disabled())
	391	return;
	392
	393	VM_BUG_ON_PAGE(!PageHuge(oldhpage), oldhpage);
	394	spin_lock(&hugetlb_lock);
	395	h_cg = hugetlb_cgroup_from_page(oldhpage);
	396	set_hugetlb_cgroup(oldhpage, NULL);
	397
	398	/* move the h_cg details to new cgroup */
	399	set_hugetlb_cgroup(newhpage, h_cg);
	400	list_move(&newhpage->lru, &h->hugepage_activelist);
	401	spin_unlock(&hugetlb_lock);
	402	return;
	403	}
	404
	405	struct cgroup_subsys hugetlb_cgrp_subsys = {
	406	.css_alloc = hugetlb_cgroup_css_alloc,
	407	.css_offline = hugetlb_cgroup_css_offline,
	408	.css_free = hugetlb_cgroup_css_free,
	409	};