[GitHub/mt8127/android_kernel_alcatel_ttab.git] / mm / page_cgroup.c

#include <linux/mm.h>
#include <linux/mmzone.h>
#include <linux/bootmem.h>
#include <linux/bit_spinlock.h>
#include <linux/page_cgroup.h>
#include <linux/hash.h>
#include <linux/slab.h>
#include <linux/memory.h>
#include <linux/vmalloc.h>
#include <linux/cgroup.h>
#include <linux/swapops.h>
#include <linux/kmemleak.h>

static unsigned long total_usage;

#if !defined(CONFIG_SPARSEMEM)


void __meminit pgdat_page_cgroup_init(struct pglist_data *pgdat)
{
	pgdat->node_page_cgroup = NULL;
}

struct page_cgroup *lookup_page_cgroup(struct page *page)
{
	unsigned long pfn = page_to_pfn(page);
	unsigned long offset;
	struct page_cgroup *base;

	base = NODE_DATA(page_to_nid(page))->node_page_cgroup;
#ifdef CONFIG_DEBUG_VM
	/*
	 * The sanity checks the page allocator does upon freeing a
	 * page can reach here before the page_cgroup arrays are
	 * allocated when feeding a range of pages to the allocator
	 * for the first time during bootup or memory hotplug.
	 */
	if (unlikely(!base))
		return NULL;
#endif
	offset = pfn - NODE_DATA(page_to_nid(page))->node_start_pfn;
	return base + offset;
}

static int __init alloc_node_page_cgroup(int nid)
{
	struct page_cgroup *base;
	unsigned long table_size;
	unsigned long nr_pages;

	nr_pages = NODE_DATA(nid)->node_spanned_pages;
	if (!nr_pages)
		return 0;

	table_size = sizeof(struct page_cgroup) * nr_pages;

	base = __alloc_bootmem_node_nopanic(NODE_DATA(nid),
			table_size, PAGE_SIZE, __pa(MAX_DMA_ADDRESS));
	if (!base)
		return -ENOMEM;
	NODE_DATA(nid)->node_page_cgroup = base;
	total_usage += table_size;
	return 0;
}

void __init page_cgroup_init_flatmem(void)
{

	int nid, fail;

	if (mem_cgroup_disabled())
		return;

	for_each_online_node(nid)  {
		fail = alloc_node_page_cgroup(nid);
		if (fail)
			goto fail;
	}
	printk(KERN_INFO "allocated %ld bytes of page_cgroup\n", total_usage);
	printk(KERN_INFO "please try 'cgroup_disable=memory' option if you"
	" don't want memory cgroups\n");
	return;
fail:
	printk(KERN_CRIT "allocation of page_cgroup failed.\n");
	printk(KERN_CRIT "please try 'cgroup_disable=memory' boot option\n");
	panic("Out of memory");
}

#else /* CONFIG_FLAT_NODE_MEM_MAP */

struct page_cgroup *lookup_page_cgroup(struct page *page)
{
	unsigned long pfn = page_to_pfn(page);
	struct mem_section *section = __pfn_to_section(pfn);
#ifdef CONFIG_DEBUG_VM
	/*
	 * The sanity checks the page allocator does upon freeing a
	 * page can reach here before the page_cgroup arrays are
	 * allocated when feeding a range of pages to the allocator
	 * for the first time during bootup or memory hotplug.
	 */
	if (!section->page_cgroup)
		return NULL;
#endif
	return section->page_cgroup + pfn;
}

static void *__meminit alloc_page_cgroup(size_t size, int nid)
{
	gfp_t flags = GFP_KERNEL | __GFP_ZERO | __GFP_NOWARN;
	void *addr = NULL;

	addr = alloc_pages_exact_nid(nid, size, flags);
	if (addr) {
		kmemleak_alloc(addr, size, 1, flags);
		return addr;
	}

	if (node_state(nid, N_HIGH_MEMORY))
		addr = vzalloc_node(size, nid);
	else
		addr = vzalloc(size);

	return addr;
}

#ifdef CONFIG_MEMORY_HOTPLUG
static void free_page_cgroup(void *addr)
{
	if (is_vmalloc_addr(addr)) {
		vfree(addr);
	} else {
		struct page *page = virt_to_page(addr);
		size_t table_size =
			sizeof(struct page_cgroup) * PAGES_PER_SECTION;

		BUG_ON(PageReserved(page));
		free_pages_exact(addr, table_size);
	}
}
#endif

static int __meminit init_section_page_cgroup(unsigned long pfn, int nid)
{
	struct mem_section *section;
	struct page_cgroup *base;
	unsigned long table_size;

	section = __pfn_to_section(pfn);

	if (section->page_cgroup)
		return 0;

	table_size = sizeof(struct page_cgroup) * PAGES_PER_SECTION;
	base = alloc_page_cgroup(table_size, nid);

	/*
	 * The value stored in section->page_cgroup is (base - pfn)
	 * and it does not point to the memory block allocated above,
	 * causing kmemleak false positives.
	 */
	kmemleak_not_leak(base);

	if (!base) {
		printk(KERN_ERR "page cgroup allocation failure\n");
		return -ENOMEM;
	}

	/*
	 * The passed "pfn" may not be aligned to SECTION.  For the calculation
	 * we need to apply a mask.
	 */
	pfn &= PAGE_SECTION_MASK;
	section->page_cgroup = base - pfn;
	total_usage += table_size;
	return 0;
}
#ifdef CONFIG_MEMORY_HOTPLUG
void __free_page_cgroup(unsigned long pfn)
{
	struct mem_section *ms;
	struct page_cgroup *base;

	ms = __pfn_to_section(pfn);
	if (!ms || !ms->page_cgroup)
		return;
	base = ms->page_cgroup + pfn;
	free_page_cgroup(base);
	ms->page_cgroup = NULL;
}

int __meminit online_page_cgroup(unsigned long start_pfn,
			unsigned long nr_pages,
			int nid)
{
	unsigned long start, end, pfn;
	int fail = 0;

	start = SECTION_ALIGN_DOWN(start_pfn);
	end = SECTION_ALIGN_UP(start_pfn + nr_pages);

	if (nid == -1) {
		/*
		 * In this case, "nid" already exists and contains valid memory.
		 * "start_pfn" passed to us is a pfn which is an arg for
		 * online__pages(), and start_pfn should exist.
		 */
		nid = pfn_to_nid(start_pfn);
		VM_BUG_ON(!node_state(nid, N_ONLINE));
	}

	for (pfn = start; !fail && pfn < end; pfn += PAGES_PER_SECTION) {
		if (!pfn_present(pfn))
			continue;
		fail = init_section_page_cgroup(pfn, nid);
	}
	if (!fail)
		return 0;

	/* rollback */
	for (pfn = start; pfn < end; pfn += PAGES_PER_SECTION)
		__free_page_cgroup(pfn);

	return -ENOMEM;
}

int __meminit offline_page_cgroup(unsigned long start_pfn,
		unsigned long nr_pages, int nid)
{
	unsigned long start, end, pfn;

	start = SECTION_ALIGN_DOWN(start_pfn);
	end = SECTION_ALIGN_UP(start_pfn + nr_pages);

	for (pfn = start; pfn < end; pfn += PAGES_PER_SECTION)
		__free_page_cgroup(pfn);
	return 0;

}

static int __meminit page_cgroup_callback(struct notifier_block *self,
			       unsigned long action, void *arg)
{
	struct memory_notify *mn = arg;
	int ret = 0;
	switch (action) {
	case MEM_GOING_ONLINE:
		ret = online_page_cgroup(mn->start_pfn,
				   mn->nr_pages, mn->status_change_nid);
		break;
	case MEM_OFFLINE:
		offline_page_cgroup(mn->start_pfn,
				mn->nr_pages, mn->status_change_nid);
		break;
	case MEM_CANCEL_ONLINE:
	case MEM_GOING_OFFLINE:
		break;
	case MEM_ONLINE:
	case MEM_CANCEL_OFFLINE:
		break;
	}

	return notifier_from_errno(ret);
}

#endif

void __init page_cgroup_init(void)
{
	unsigned long pfn;
	int nid;

	if (mem_cgroup_disabled())
		return;

	for_each_node_state(nid, N_HIGH_MEMORY) {
		unsigned long start_pfn, end_pfn;

		start_pfn = node_start_pfn(nid);
		end_pfn = node_end_pfn(nid);
		/*
		 * start_pfn and end_pfn may not be aligned to SECTION and the
		 * page->flags of out of node pages are not initialized.  So we
		 * scan [start_pfn, the biggest section's pfn < end_pfn) here.
		 */
		for (pfn = start_pfn;
		     pfn < end_pfn;
                     pfn = ALIGN(pfn + 1, PAGES_PER_SECTION)) {

			if (!pfn_valid(pfn))
				continue;
			/*
			 * Nodes's pfns can be overlapping.
			 * We know some arch can have a nodes layout such as
			 * -------------pfn-------------->
			 * N0 | N1 | N2 | N0 | N1 | N2|....
			 */
			if (pfn_to_nid(pfn) != nid)
				continue;
			if (init_section_page_cgroup(pfn, nid))
				goto oom;
		}
	}
	hotplug_memory_notifier(page_cgroup_callback, 0);
	printk(KERN_INFO "allocated %ld bytes of page_cgroup\n", total_usage);
	printk(KERN_INFO "please try 'cgroup_disable=memory' option if you "
			 "don't want memory cgroups\n");
	return;
oom:
	printk(KERN_CRIT "try 'cgroup_disable=memory' boot option\n");
	panic("Out of memory");
}

void __meminit pgdat_page_cgroup_init(struct pglist_data *pgdat)
{
	return;
}

#endif


#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP

static DEFINE_MUTEX(swap_cgroup_mutex);
struct swap_cgroup_ctrl {
	struct page **map;
	unsigned long length;
	spinlock_t	lock;
};

static struct swap_cgroup_ctrl swap_cgroup_ctrl[MAX_SWAPFILES];

struct swap_cgroup {
	unsigned short		id;
};
#define SC_PER_PAGE	(PAGE_SIZE/sizeof(struct swap_cgroup))
#define SC_POS_MASK	(SC_PER_PAGE - 1)

/*
 * SwapCgroup implements "lookup" and "exchange" operations.
 * In typical usage, this swap_cgroup is accessed via memcg's charge/uncharge
 * against SwapCache. At swap_free(), this is accessed directly from swap.
 *
 * This means,
 *  - we have no race in "exchange" when we're accessed via SwapCache because
 *    SwapCache(and its swp_entry) is under lock.
 *  - When called via swap_free(), there is no user of this entry and no race.
 * Then, we don't need lock around "exchange".
 *
 * TODO: we can push these buffers out to HIGHMEM.
 */

/*
 * allocate buffer for swap_cgroup.
 */
static int swap_cgroup_prepare(int type)
{
	struct page *page;
	struct swap_cgroup_ctrl *ctrl;
	unsigned long idx, max;

	ctrl = &swap_cgroup_ctrl[type];

	for (idx = 0; idx < ctrl->length; idx++) {
		page = alloc_page(GFP_KERNEL | __GFP_ZERO);
		if (!page)
			goto not_enough_page;
		ctrl->map[idx] = page;
	}
	return 0;
not_enough_page:
	max = idx;
	for (idx = 0; idx < max; idx++)
		__free_page(ctrl->map[idx]);

	return -ENOMEM;
}

/**
 * swap_cgroup_cmpxchg - cmpxchg mem_cgroup's id for this swp_entry.
 * @end: swap entry to be cmpxchged
 * @old: old id
 * @new: new id
 *
 * Returns old id at success, 0 at failure.
 * (There is no mem_cgroup using 0 as its id)
 */
unsigned short swap_cgroup_cmpxchg(swp_entry_t ent,
					unsigned short old, unsigned short new)
{
	int type = swp_type(ent);
	unsigned long offset = swp_offset(ent);
	unsigned long idx = offset / SC_PER_PAGE;
	unsigned long pos = offset & SC_POS_MASK;
	struct swap_cgroup_ctrl *ctrl;
	struct page *mappage;
	struct swap_cgroup *sc;
	unsigned long flags;
	unsigned short retval;

	ctrl = &swap_cgroup_ctrl[type];

	mappage = ctrl->map[idx];
	sc = page_address(mappage);
	sc += pos;
	spin_lock_irqsave(&ctrl->lock, flags);
	retval = sc->id;
	if (retval == old)
		sc->id = new;
	else
		retval = 0;
	spin_unlock_irqrestore(&ctrl->lock, flags);
	return retval;
}

/**
 * swap_cgroup_record - record mem_cgroup for this swp_entry.
 * @ent: swap entry to be recorded into
 * @mem: mem_cgroup to be recorded
 *
 * Returns old value at success, 0 at failure.
 * (Of course, old value can be 0.)
 */
unsigned short swap_cgroup_record(swp_entry_t ent, unsigned short id)
{
	int type = swp_type(ent);
	unsigned long offset = swp_offset(ent);
	unsigned long idx = offset / SC_PER_PAGE;
	unsigned long pos = offset & SC_POS_MASK;
	struct swap_cgroup_ctrl *ctrl;
	struct page *mappage;
	struct swap_cgroup *sc;
	unsigned short old;
	unsigned long flags;

	ctrl = &swap_cgroup_ctrl[type];

	mappage = ctrl->map[idx];
	sc = page_address(mappage);
	sc += pos;
	spin_lock_irqsave(&ctrl->lock, flags);
	old = sc->id;
	sc->id = id;
	spin_unlock_irqrestore(&ctrl->lock, flags);

	return old;
}

/**
 * lookup_swap_cgroup - lookup mem_cgroup tied to swap entry
 * @ent: swap entry to be looked up.
 *
 * Returns CSS ID of mem_cgroup at success. 0 at failure. (0 is invalid ID)
 */
unsigned short lookup_swap_cgroup(swp_entry_t ent)
{
	int type = swp_type(ent);
	unsigned long offset = swp_offset(ent);
	unsigned long idx = offset / SC_PER_PAGE;
	unsigned long pos = offset & SC_POS_MASK;
	struct swap_cgroup_ctrl *ctrl;
	struct page *mappage;
	struct swap_cgroup *sc;
	unsigned short ret;

	ctrl = &swap_cgroup_ctrl[type];
	mappage = ctrl->map[idx];
	sc = page_address(mappage);
	sc += pos;
	ret = sc->id;
	return ret;
}

int swap_cgroup_swapon(int type, unsigned long max_pages)
{
	void *array;
	unsigned long array_size;
	unsigned long length;
	struct swap_cgroup_ctrl *ctrl;

	if (!do_swap_account)
		return 0;

	length = DIV_ROUND_UP(max_pages, SC_PER_PAGE);
	array_size = length * sizeof(void *);

	array = vzalloc(array_size);
	if (!array)
		goto nomem;

	ctrl = &swap_cgroup_ctrl[type];
	mutex_lock(&swap_cgroup_mutex);
	ctrl->length = length;
	ctrl->map = array;
	spin_lock_init(&ctrl->lock);
	if (swap_cgroup_prepare(type)) {
		/* memory shortage */
		ctrl->map = NULL;
		ctrl->length = 0;
		mutex_unlock(&swap_cgroup_mutex);
		vfree(array);
		goto nomem;
	}
	mutex_unlock(&swap_cgroup_mutex);

	return 0;
nomem:
	printk(KERN_INFO "couldn't allocate enough memory for swap_cgroup.\n");
	printk(KERN_INFO
		"swap_cgroup can be disabled by swapaccount=0 boot option\n");
	return -ENOMEM;
}

void swap_cgroup_swapoff(int type)
{
	struct page **map;
	unsigned long i, length;
	struct swap_cgroup_ctrl *ctrl;

	if (!do_swap_account)
		return;

	mutex_lock(&swap_cgroup_mutex);
	ctrl = &swap_cgroup_ctrl[type];
	map = ctrl->map;
	length = ctrl->length;
	ctrl->map = NULL;
	ctrl->length = 0;
	mutex_unlock(&swap_cgroup_mutex);

	if (map) {
		for (i = 0; i < length; i++) {
			struct page *page = map[i];
			if (page)
				__free_page(page);
		}
		vfree(map);
	}
}

#endif
Commit	Line	Data
52d4b9ac KH	1	#include <linux/mm.h>
	2	#include <linux/mmzone.h>
	3	#include <linux/bootmem.h>
	4	#include <linux/bit_spinlock.h>
	5	#include <linux/page_cgroup.h>
	6	#include <linux/hash.h>
94b6da5a	7	#include <linux/slab.h>
52d4b9ac	8	#include <linux/memory.h>
4c821042	9	#include <linux/vmalloc.h>
94b6da5a	10	#include <linux/cgroup.h>
27a7faa0	11	#include <linux/swapops.h>
7952f988	12	#include <linux/kmemleak.h>
52d4b9ac	13
52d4b9ac KH	14	static unsigned long total_usage;
	15
	16	#if !defined(CONFIG_SPARSEMEM)
	17
	18
31168481	19	void __meminit pgdat_page_cgroup_init(struct pglist_data *pgdat)
52d4b9ac KH	20	{
	21	pgdat->node_page_cgroup = NULL;
	22	}
	23
	24	struct page_cgroup lookup_page_cgroup(struct page page)
	25	{
	26	unsigned long pfn = page_to_pfn(page);
	27	unsigned long offset;
	28	struct page_cgroup *base;
	29
	30	base = NODE_DATA(page_to_nid(page))->node_page_cgroup;
00c54c0b JW	31	#ifdef CONFIG_DEBUG_VM
	32	/*
	33	* The sanity checks the page allocator does upon freeing a
	34	* page can reach here before the page_cgroup arrays are
	35	* allocated when feeding a range of pages to the allocator
	36	* for the first time during bootup or memory hotplug.
	37	*/
52d4b9ac KH	38	if (unlikely(!base))
52d4b9ac KH	39	return NULL;
00c54c0b	40	#endif
52d4b9ac KH	41	offset = pfn - NODE_DATA(page_to_nid(page))->node_start_pfn;
	42	return base + offset;
	43	}
	44
	45	static int __init alloc_node_page_cgroup(int nid)
	46	{
6b208e3f	47	struct page_cgroup *base;
52d4b9ac	48	unsigned long table_size;
6b208e3f	49	unsigned long nr_pages;
52d4b9ac	50
52d4b9ac	51	nr_pages = NODE_DATA(nid)->node_spanned_pages;
653d22c0 KH	52	if (!nr_pages)
	53	return 0;
	54
52d4b9ac	55	table_size = sizeof(struct page_cgroup) * nr_pages;
ca371c0d KH	56
	57	base = __alloc_bootmem_node_nopanic(NODE_DATA(nid),
	58	table_size, PAGE_SIZE, __pa(MAX_DMA_ADDRESS));
	59	if (!base)
52d4b9ac	60	return -ENOMEM;
52d4b9ac KH	61	NODE_DATA(nid)->node_page_cgroup = base;
	62	total_usage += table_size;
	63	return 0;
	64	}
	65
ca371c0d	66	void __init page_cgroup_init_flatmem(void)
52d4b9ac KH	67	{
	68
	69	int nid, fail;
	70
f8d66542	71	if (mem_cgroup_disabled())
94b6da5a KH	72	return;
94b6da5a KH	73
52d4b9ac KH	74	for_each_online_node(nid) {
	75	fail = alloc_node_page_cgroup(nid);
	76	if (fail)
	77	goto fail;
	78	}
	79	printk(KERN_INFO "allocated %ld bytes of page_cgroup\n", total_usage);
8ca739e3 RD	80	printk(KERN_INFO "please try 'cgroup_disable=memory' option if you"
8ca739e3 RD	81	" don't want memory cgroups\n");
52d4b9ac KH	82	return;
52d4b9ac KH	83	fail:
8ca739e3 RD	84	printk(KERN_CRIT "allocation of page_cgroup failed.\n");
8ca739e3 RD	85	printk(KERN_CRIT "please try 'cgroup_disable=memory' boot option\n");
52d4b9ac KH	86	panic("Out of memory");
	87	}
	88
	89	#else /* CONFIG_FLAT_NODE_MEM_MAP */
	90
	91	struct page_cgroup lookup_page_cgroup(struct page page)
	92	{
	93	unsigned long pfn = page_to_pfn(page);
	94	struct mem_section *section = __pfn_to_section(pfn);
00c54c0b JW	95	#ifdef CONFIG_DEBUG_VM
	96	/*
	97	* The sanity checks the page allocator does upon freeing a
	98	* page can reach here before the page_cgroup arrays are
	99	* allocated when feeding a range of pages to the allocator
	100	* for the first time during bootup or memory hotplug.
	101	*/
d69b042f BS	102	if (!section->page_cgroup)
d69b042f BS	103	return NULL;
00c54c0b	104	#endif
52d4b9ac KH	105	return section->page_cgroup + pfn;
	106	}
	107
268433b8	108	static void *__meminit alloc_page_cgroup(size_t size, int nid)
dde79e00	109	{
6b208e3f	110	gfp_t flags = GFP_KERNEL \| __GFP_ZERO \| __GFP_NOWARN;
dde79e00 MH	111	void *addr = NULL;
dde79e00 MH	112
ff7ee93f SR	113	addr = alloc_pages_exact_nid(nid, size, flags);
	114	if (addr) {
	115	kmemleak_alloc(addr, size, 1, flags);
dde79e00	116	return addr;
ff7ee93f	117	}
dde79e00 MH	118
dde79e00 MH	119	if (node_state(nid, N_HIGH_MEMORY))
6b208e3f	120	addr = vzalloc_node(size, nid);
dde79e00	121	else
6b208e3f	122	addr = vzalloc(size);
dde79e00 MH	123
	124	return addr;
	125	}
	126
	127	#ifdef CONFIG_MEMORY_HOTPLUG
	128	static void free_page_cgroup(void *addr)
	129	{
	130	if (is_vmalloc_addr(addr)) {
	131	vfree(addr);
	132	} else {
	133	struct page *page = virt_to_page(addr);
6cfddb26 MH	134	size_t table_size =
	135	sizeof(struct page_cgroup) * PAGES_PER_SECTION;
	136
	137	BUG_ON(PageReserved(page));
	138	free_pages_exact(addr, table_size);
dde79e00 MH	139	}
	140	}
	141	#endif
	142
37573e8c	143	static int __meminit init_section_page_cgroup(unsigned long pfn, int nid)
52d4b9ac	144	{
6b3ae58e	145	struct mem_section *section;
6b208e3f	146	struct page_cgroup *base;
52d4b9ac	147	unsigned long table_size;
52d4b9ac	148
6b208e3f	149	section = __pfn_to_section(pfn);
6b3ae58e JW	150
	151	if (section->page_cgroup)
	152	return 0;
	153
6b3ae58e	154	table_size = sizeof(struct page_cgroup) * PAGES_PER_SECTION;
dde79e00 MH	155	base = alloc_page_cgroup(table_size, nid);
dde79e00 MH	156
6b3ae58e JW	157	/*
	158	* The value stored in section->page_cgroup is (base - pfn)
	159	* and it does not point to the memory block allocated above,
	160	* causing kmemleak false positives.
	161	*/
	162	kmemleak_not_leak(base);
52d4b9ac KH	163
	164	if (!base) {
	165	printk(KERN_ERR "page cgroup allocation failure\n");
	166	return -ENOMEM;
	167	}
	168
37573e8c KH	169	/*
	170	* The passed "pfn" may not be aligned to SECTION. For the calculation
	171	* we need to apply a mask.
	172	*/
	173	pfn &= PAGE_SECTION_MASK;
52d4b9ac KH	174	section->page_cgroup = base - pfn;
	175	total_usage += table_size;
	176	return 0;
	177	}
	178	#ifdef CONFIG_MEMORY_HOTPLUG
	179	void __free_page_cgroup(unsigned long pfn)
	180	{
	181	struct mem_section *ms;
	182	struct page_cgroup *base;
	183
	184	ms = __pfn_to_section(pfn);
	185	if (!ms \|\| !ms->page_cgroup)
	186	return;
	187	base = ms->page_cgroup + pfn;
dde79e00 MH	188	free_page_cgroup(base);
dde79e00 MH	189	ms->page_cgroup = NULL;
52d4b9ac KH	190	}
52d4b9ac KH	191
31168481	192	int __meminit online_page_cgroup(unsigned long start_pfn,
52d4b9ac KH	193	unsigned long nr_pages,
	194	int nid)
	195	{
	196	unsigned long start, end, pfn;
	197	int fail = 0;
	198
1bb36fbd DK	199	start = SECTION_ALIGN_DOWN(start_pfn);
1bb36fbd DK	200	end = SECTION_ALIGN_UP(start_pfn + nr_pages);
52d4b9ac	201
37573e8c KH	202	if (nid == -1) {
	203	/*
	204	* In this case, "nid" already exists and contains valid memory.
	205	* "start_pfn" passed to us is a pfn which is an arg for
	206	* online__pages(), and start_pfn should exist.
	207	*/
	208	nid = pfn_to_nid(start_pfn);
	209	VM_BUG_ON(!node_state(nid, N_ONLINE));
	210	}
	211
52d4b9ac KH	212	for (pfn = start; !fail && pfn < end; pfn += PAGES_PER_SECTION) {
	213	if (!pfn_present(pfn))
	214	continue;
37573e8c	215	fail = init_section_page_cgroup(pfn, nid);
52d4b9ac KH	216	}
	217	if (!fail)
	218	return 0;
	219
	220	/* rollback */
	221	for (pfn = start; pfn < end; pfn += PAGES_PER_SECTION)
	222	__free_page_cgroup(pfn);
	223
	224	return -ENOMEM;
	225	}
	226
31168481	227	int __meminit offline_page_cgroup(unsigned long start_pfn,
52d4b9ac KH	228	unsigned long nr_pages, int nid)
	229	{
	230	unsigned long start, end, pfn;
	231
1bb36fbd DK	232	start = SECTION_ALIGN_DOWN(start_pfn);
1bb36fbd DK	233	end = SECTION_ALIGN_UP(start_pfn + nr_pages);
52d4b9ac KH	234
	235	for (pfn = start; pfn < end; pfn += PAGES_PER_SECTION)
	236	__free_page_cgroup(pfn);
	237	return 0;
	238
	239	}
	240
31168481	241	static int __meminit page_cgroup_callback(struct notifier_block *self,
52d4b9ac KH	242	unsigned long action, void *arg)
	243	{
	244	struct memory_notify *mn = arg;
	245	int ret = 0;
	246	switch (action) {
	247	case MEM_GOING_ONLINE:
	248	ret = online_page_cgroup(mn->start_pfn,
	249	mn->nr_pages, mn->status_change_nid);
	250	break;
52d4b9ac KH	251	case MEM_OFFLINE:
	252	offline_page_cgroup(mn->start_pfn,
	253	mn->nr_pages, mn->status_change_nid);
	254	break;
dc19f9db	255	case MEM_CANCEL_ONLINE:
52d4b9ac KH	256	case MEM_GOING_OFFLINE:
	257	break;
	258	case MEM_ONLINE:
	259	case MEM_CANCEL_OFFLINE:
	260	break;
	261	}
dc19f9db	262
5fda1bd5	263	return notifier_from_errno(ret);
52d4b9ac KH	264	}
	265
	266	#endif
	267
	268	void __init page_cgroup_init(void)
	269	{
	270	unsigned long pfn;
37573e8c	271	int nid;
52d4b9ac	272
f8d66542	273	if (mem_cgroup_disabled())
94b6da5a KH	274	return;
94b6da5a KH	275
37573e8c KH	276	for_each_node_state(nid, N_HIGH_MEMORY) {
	277	unsigned long start_pfn, end_pfn;
	278
	279	start_pfn = node_start_pfn(nid);
	280	end_pfn = node_end_pfn(nid);
	281	/*
	282	* start_pfn and end_pfn may not be aligned to SECTION and the
	283	* page->flags of out of node pages are not initialized. So we
	284	* scan [start_pfn, the biggest section's pfn < end_pfn) here.
	285	*/
	286	for (pfn = start_pfn;
	287	pfn < end_pfn;
	288	pfn = ALIGN(pfn + 1, PAGES_PER_SECTION)) {
	289
	290	if (!pfn_valid(pfn))
	291	continue;
	292	/*
	293	* Nodes's pfns can be overlapping.
	294	* We know some arch can have a nodes layout such as
	295	* -------------pfn-------------->
	296	* N0 \| N1 \| N2 \| N0 \| N1 \| N2\|....
	297	*/
	298	if (pfn_to_nid(pfn) != nid)
	299	continue;
	300	if (init_section_page_cgroup(pfn, nid))
	301	goto oom;
	302	}
52d4b9ac	303	}
37573e8c	304	hotplug_memory_notifier(page_cgroup_callback, 0);
52d4b9ac	305	printk(KERN_INFO "allocated %ld bytes of page_cgroup\n", total_usage);
37573e8c KH	306	printk(KERN_INFO "please try 'cgroup_disable=memory' option if you "
	307	"don't want memory cgroups\n");
	308	return;
	309	oom:
	310	printk(KERN_CRIT "try 'cgroup_disable=memory' boot option\n");
	311	panic("Out of memory");
52d4b9ac KH	312	}
52d4b9ac KH	313
31168481	314	void __meminit pgdat_page_cgroup_init(struct pglist_data *pgdat)
52d4b9ac KH	315	{
	316	return;
	317	}
	318
	319	#endif
27a7faa0 KH	320
	321
	322	#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
	323
	324	static DEFINE_MUTEX(swap_cgroup_mutex);
	325	struct swap_cgroup_ctrl {
	326	struct page **map;
	327	unsigned long length;
e9e58a4e	328	spinlock_t lock;
27a7faa0 KH	329	};
27a7faa0 KH	330
61600f57	331	static struct swap_cgroup_ctrl swap_cgroup_ctrl[MAX_SWAPFILES];
27a7faa0	332
27a7faa0	333	struct swap_cgroup {
a3b2d692	334	unsigned short id;
27a7faa0 KH	335	};
	336	#define SC_PER_PAGE (PAGE_SIZE/sizeof(struct swap_cgroup))
	337	#define SC_POS_MASK (SC_PER_PAGE - 1)
	338
	339	/*
	340	* SwapCgroup implements "lookup" and "exchange" operations.
	341	* In typical usage, this swap_cgroup is accessed via memcg's charge/uncharge
	342	* against SwapCache. At swap_free(), this is accessed directly from swap.
	343	*
	344	* This means,
	345	* - we have no race in "exchange" when we're accessed via SwapCache because
	346	* SwapCache(and its swp_entry) is under lock.
	347	* - When called via swap_free(), there is no user of this entry and no race.
	348	* Then, we don't need lock around "exchange".
	349	*
	350	* TODO: we can push these buffers out to HIGHMEM.
	351	*/
	352
	353	/*
	354	* allocate buffer for swap_cgroup.
	355	*/
	356	static int swap_cgroup_prepare(int type)
	357	{
	358	struct page *page;
	359	struct swap_cgroup_ctrl *ctrl;
	360	unsigned long idx, max;
	361
27a7faa0 KH	362	ctrl = &swap_cgroup_ctrl[type];
	363
	364	for (idx = 0; idx < ctrl->length; idx++) {
	365	page = alloc_page(GFP_KERNEL \| __GFP_ZERO);
	366	if (!page)
	367	goto not_enough_page;
	368	ctrl->map[idx] = page;
	369	}
	370	return 0;
	371	not_enough_page:
	372	max = idx;
	373	for (idx = 0; idx < max; idx++)
	374	__free_page(ctrl->map[idx]);
	375
	376	return -ENOMEM;
	377	}
	378
02491447 DN	379	/**
	380	* swap_cgroup_cmpxchg - cmpxchg mem_cgroup's id for this swp_entry.
	381	* @end: swap entry to be cmpxchged
	382	* @old: old id
	383	* @new: new id
	384	*
	385	* Returns old id at success, 0 at failure.
25985edc	386	* (There is no mem_cgroup using 0 as its id)
02491447 DN	387	*/
	388	unsigned short swap_cgroup_cmpxchg(swp_entry_t ent,
	389	unsigned short old, unsigned short new)
	390	{
	391	int type = swp_type(ent);
	392	unsigned long offset = swp_offset(ent);
	393	unsigned long idx = offset / SC_PER_PAGE;
	394	unsigned long pos = offset & SC_POS_MASK;
	395	struct swap_cgroup_ctrl *ctrl;
	396	struct page *mappage;
	397	struct swap_cgroup *sc;
e9e58a4e KH	398	unsigned long flags;
e9e58a4e KH	399	unsigned short retval;
02491447 DN	400
	401	ctrl = &swap_cgroup_ctrl[type];
	402
	403	mappage = ctrl->map[idx];
	404	sc = page_address(mappage);
	405	sc += pos;
e9e58a4e KH	406	spin_lock_irqsave(&ctrl->lock, flags);
	407	retval = sc->id;
	408	if (retval == old)
	409	sc->id = new;
02491447	410	else
e9e58a4e KH	411	retval = 0;
	412	spin_unlock_irqrestore(&ctrl->lock, flags);
	413	return retval;
02491447 DN	414	}
02491447 DN	415
27a7faa0 KH	416	/**
	417	* swap_cgroup_record - record mem_cgroup for this swp_entry.
	418	* @ent: swap entry to be recorded into
	419	* @mem: mem_cgroup to be recorded
	420	*
a3b2d692 KH	421	* Returns old value at success, 0 at failure.
a3b2d692 KH	422	* (Of course, old value can be 0.)
27a7faa0	423	*/
a3b2d692	424	unsigned short swap_cgroup_record(swp_entry_t ent, unsigned short id)
27a7faa0 KH	425	{
	426	int type = swp_type(ent);
	427	unsigned long offset = swp_offset(ent);
	428	unsigned long idx = offset / SC_PER_PAGE;
	429	unsigned long pos = offset & SC_POS_MASK;
	430	struct swap_cgroup_ctrl *ctrl;
	431	struct page *mappage;
	432	struct swap_cgroup *sc;
a3b2d692	433	unsigned short old;
e9e58a4e	434	unsigned long flags;
27a7faa0	435
27a7faa0 KH	436	ctrl = &swap_cgroup_ctrl[type];
	437
	438	mappage = ctrl->map[idx];
	439	sc = page_address(mappage);
	440	sc += pos;
e9e58a4e KH	441	spin_lock_irqsave(&ctrl->lock, flags);
	442	old = sc->id;
	443	sc->id = id;
	444	spin_unlock_irqrestore(&ctrl->lock, flags);
27a7faa0 KH	445
	446	return old;
	447	}
	448
	449	/**
	450	* lookup_swap_cgroup - lookup mem_cgroup tied to swap entry
	451	* @ent: swap entry to be looked up.
	452	*
a3b2d692	453	* Returns CSS ID of mem_cgroup at success. 0 at failure. (0 is invalid ID)
27a7faa0	454	*/
a3b2d692	455	unsigned short lookup_swap_cgroup(swp_entry_t ent)
27a7faa0 KH	456	{
	457	int type = swp_type(ent);
	458	unsigned long offset = swp_offset(ent);
	459	unsigned long idx = offset / SC_PER_PAGE;
	460	unsigned long pos = offset & SC_POS_MASK;
	461	struct swap_cgroup_ctrl *ctrl;
	462	struct page *mappage;
	463	struct swap_cgroup *sc;
a3b2d692	464	unsigned short ret;
27a7faa0	465
27a7faa0 KH	466	ctrl = &swap_cgroup_ctrl[type];
	467	mappage = ctrl->map[idx];
	468	sc = page_address(mappage);
	469	sc += pos;
a3b2d692	470	ret = sc->id;
27a7faa0 KH	471	return ret;
	472	}
	473
	474	int swap_cgroup_swapon(int type, unsigned long max_pages)
	475	{
	476	void *array;
	477	unsigned long array_size;
	478	unsigned long length;
	479	struct swap_cgroup_ctrl *ctrl;
	480
	481	if (!do_swap_account)
	482	return 0;
	483
33278f7f	484	length = DIV_ROUND_UP(max_pages, SC_PER_PAGE);
27a7faa0 KH	485	array_size = length * sizeof(void *);
27a7faa0 KH	486
8c1fec1b	487	array = vzalloc(array_size);
27a7faa0 KH	488	if (!array)
	489	goto nomem;
	490
27a7faa0 KH	491	ctrl = &swap_cgroup_ctrl[type];
	492	mutex_lock(&swap_cgroup_mutex);
	493	ctrl->length = length;
	494	ctrl->map = array;
e9e58a4e	495	spin_lock_init(&ctrl->lock);
27a7faa0 KH	496	if (swap_cgroup_prepare(type)) {
	497	/* memory shortage */
	498	ctrl->map = NULL;
	499	ctrl->length = 0;
27a7faa0	500	mutex_unlock(&swap_cgroup_mutex);
6a5b18d2	501	vfree(array);
27a7faa0 KH	502	goto nomem;
	503	}
	504	mutex_unlock(&swap_cgroup_mutex);
	505
27a7faa0 KH	506	return 0;
	507	nomem:
	508	printk(KERN_INFO "couldn't allocate enough memory for swap_cgroup.\n");
	509	printk(KERN_INFO
00a66d29	510	"swap_cgroup can be disabled by swapaccount=0 boot option\n");
27a7faa0 KH	511	return -ENOMEM;
	512	}
	513
	514	void swap_cgroup_swapoff(int type)
	515	{
6a5b18d2 NK	516	struct page **map;
6a5b18d2 NK	517	unsigned long i, length;
27a7faa0 KH	518	struct swap_cgroup_ctrl *ctrl;
	519
	520	if (!do_swap_account)
	521	return;
	522
	523	mutex_lock(&swap_cgroup_mutex);
	524	ctrl = &swap_cgroup_ctrl[type];
6a5b18d2 NK	525	map = ctrl->map;
	526	length = ctrl->length;
	527	ctrl->map = NULL;
	528	ctrl->length = 0;
	529	mutex_unlock(&swap_cgroup_mutex);
	530
	531	if (map) {
	532	for (i = 0; i < length; i++) {
	533	struct page *page = map[i];
27a7faa0 KH	534	if (page)
	535	__free_page(page);
	536	}
6a5b18d2	537	vfree(map);
27a7faa0	538	}
27a7faa0 KH	539	}
	540
	541	#endif