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

/*
 * linux/mm/page_isolation.c
 */

#include <linux/mm.h>
#include <linux/page-isolation.h>
#include <linux/pageblock-flags.h>
#include <linux/memory.h>
#include "internal.h"

/* called while holding zone->lock */
static void set_pageblock_isolate(struct page *page)
{
	if (get_pageblock_migratetype(page) == MIGRATE_ISOLATE)
		return;

	set_pageblock_migratetype(page, MIGRATE_ISOLATE);
	page_zone(page)->nr_pageblock_isolate++;
}

/* called while holding zone->lock */
static void restore_pageblock_isolate(struct page *page, int migratetype)
{
	struct zone *zone = page_zone(page);
	if (WARN_ON(get_pageblock_migratetype(page) != MIGRATE_ISOLATE))
		return;

	BUG_ON(zone->nr_pageblock_isolate <= 0);
	set_pageblock_migratetype(page, migratetype);
	zone->nr_pageblock_isolate--;
}

int set_migratetype_isolate(struct page *page, bool skip_hwpoisoned_pages)
{
	struct zone *zone;
	unsigned long flags, pfn;
	struct memory_isolate_notify arg;
	int notifier_ret;
	int ret = -EBUSY;

	zone = page_zone(page);

	spin_lock_irqsave(&zone->lock, flags);

	pfn = page_to_pfn(page);
	arg.start_pfn = pfn;
	arg.nr_pages = pageblock_nr_pages;
	arg.pages_found = 0;

	/*
	 * It may be possible to isolate a pageblock even if the
	 * migratetype is not MIGRATE_MOVABLE. The memory isolation
	 * notifier chain is used by balloon drivers to return the
	 * number of pages in a range that are held by the balloon
	 * driver to shrink memory. If all the pages are accounted for
	 * by balloons, are free, or on the LRU, isolation can continue.
	 * Later, for example, when memory hotplug notifier runs, these
	 * pages reported as "can be isolated" should be isolated(freed)
	 * by the balloon driver through the memory notifier chain.
	 */
	notifier_ret = memory_isolate_notify(MEM_ISOLATE_COUNT, &arg);
	notifier_ret = notifier_to_errno(notifier_ret);
	if (notifier_ret)
		goto out;
	/*
	 * FIXME: Now, memory hotplug doesn't call shrink_slab() by itself.
	 * We just check MOVABLE pages.
	 */
	if (!has_unmovable_pages(zone, page, arg.pages_found,
				 skip_hwpoisoned_pages))
		ret = 0;

	/*
	 * immobile means "not-on-lru" paes. If immobile is larger than
	 * removable-by-driver pages reported by notifier, we'll fail.
	 */

out:
	if (!ret) {
		unsigned long nr_pages;
		int migratetype = get_pageblock_migratetype(page);

		set_pageblock_isolate(page);
		nr_pages = move_freepages_block(zone, page, MIGRATE_ISOLATE);

		__mod_zone_freepage_state(zone, -nr_pages, migratetype);
	}

	spin_unlock_irqrestore(&zone->lock, flags);
	if (!ret)
		drain_all_pages();
	return ret;
}

void unset_migratetype_isolate(struct page *page, unsigned migratetype)
{
	struct zone *zone;
	unsigned long flags, nr_pages;

	zone = page_zone(page);
	spin_lock_irqsave(&zone->lock, flags);
	if (get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
		goto out;
	nr_pages = move_freepages_block(zone, page, migratetype);
	__mod_zone_freepage_state(zone, nr_pages, migratetype);
	restore_pageblock_isolate(page, migratetype);
out:
	spin_unlock_irqrestore(&zone->lock, flags);
}

static inline struct page *
__first_valid_page(unsigned long pfn, unsigned long nr_pages)
{
	int i;
	for (i = 0; i < nr_pages; i++)
		if (pfn_valid_within(pfn + i))
			break;
	if (unlikely(i == nr_pages))
		return NULL;
	return pfn_to_page(pfn + i);
}

/*
 * start_isolate_page_range() -- make page-allocation-type of range of pages
 * to be MIGRATE_ISOLATE.
 * @start_pfn: The lower PFN of the range to be isolated.
 * @end_pfn: The upper PFN of the range to be isolated.
 * @migratetype: migrate type to set in error recovery.
 *
 * Making page-allocation-type to be MIGRATE_ISOLATE means free pages in
 * the range will never be allocated. Any free pages and pages freed in the
 * future will not be allocated again.
 *
 * start_pfn/end_pfn must be aligned to pageblock_order.
 * Returns 0 on success and -EBUSY if any part of range cannot be isolated.
 */
int start_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
			     unsigned migratetype, bool skip_hwpoisoned_pages)
{
	unsigned long pfn;
	unsigned long undo_pfn;
	struct page *page;

	BUG_ON((start_pfn) & (pageblock_nr_pages - 1));
	BUG_ON((end_pfn) & (pageblock_nr_pages - 1));

	for (pfn = start_pfn;
	     pfn < end_pfn;
	     pfn += pageblock_nr_pages) {
		page = __first_valid_page(pfn, pageblock_nr_pages);
		if (page &&
		    set_migratetype_isolate(page, skip_hwpoisoned_pages)) {
			undo_pfn = pfn;
			goto undo;
		}
	}
	return 0;
undo:
	for (pfn = start_pfn;
	     pfn < undo_pfn;
	     pfn += pageblock_nr_pages)
		unset_migratetype_isolate(pfn_to_page(pfn), migratetype);

	return -EBUSY;
}

/*
 * Make isolated pages available again.
 */
int undo_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
			    unsigned migratetype)
{
	unsigned long pfn;
	struct page *page;
	BUG_ON((start_pfn) & (pageblock_nr_pages - 1));
	BUG_ON((end_pfn) & (pageblock_nr_pages - 1));
	for (pfn = start_pfn;
	     pfn < end_pfn;
	     pfn += pageblock_nr_pages) {
		page = __first_valid_page(pfn, pageblock_nr_pages);
		if (!page || get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
			continue;
		unset_migratetype_isolate(page, migratetype);
	}
	return 0;
}
/*
 * Test all pages in the range is free(means isolated) or not.
 * all pages in [start_pfn...end_pfn) must be in the same zone.
 * zone->lock must be held before call this.
 *
 * Returns 1 if all pages in the range are isolated.
 */
static int
__test_page_isolated_in_pageblock(unsigned long pfn, unsigned long end_pfn,
				  bool skip_hwpoisoned_pages)
{
	struct page *page;

	while (pfn < end_pfn) {
		if (!pfn_valid_within(pfn)) {
			pfn++;
			continue;
		}
		page = pfn_to_page(pfn);
		if (PageBuddy(page)) {
			/*
			 * If race between isolatation and allocation happens,
			 * some free pages could be in MIGRATE_MOVABLE list
			 * although pageblock's migratation type of the page
			 * is MIGRATE_ISOLATE. Catch it and move the page into
			 * MIGRATE_ISOLATE list.
			 */
			if (get_freepage_migratetype(page) != MIGRATE_ISOLATE) {
				struct page *end_page;

				end_page = page + (1 << page_order(page)) - 1;
				move_freepages(page_zone(page), page, end_page,
						MIGRATE_ISOLATE);
			}
			pfn += 1 << page_order(page);
		}
		else if (page_count(page) == 0 &&
			get_freepage_migratetype(page) == MIGRATE_ISOLATE)
			pfn += 1;
		else if (skip_hwpoisoned_pages && PageHWPoison(page)) {
			/*
			 * The HWPoisoned page may be not in buddy
			 * system, and page_count() is not 0.
			 */
			pfn++;
			continue;
		}
		else
			break;
	}
	if (pfn < end_pfn)
		return 0;
	return 1;
}

int test_pages_isolated(unsigned long start_pfn, unsigned long end_pfn,
			bool skip_hwpoisoned_pages)
{
	unsigned long pfn, flags;
	struct page *page;
	struct zone *zone;
	int ret;

	/*
	 * Note: pageblock_nr_page != MAX_ORDER. Then, chunks of free page
	 * is not aligned to pageblock_nr_pages.
	 * Then we just check pagetype fist.
	 */
	for (pfn = start_pfn; pfn < end_pfn; pfn += pageblock_nr_pages) {
		page = __first_valid_page(pfn, pageblock_nr_pages);
		if (page && get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
			break;
	}
	page = __first_valid_page(start_pfn, end_pfn - start_pfn);
	if ((pfn < end_pfn) || !page)
		return -EBUSY;
	/* Check all pages are free or Marked as ISOLATED */
	zone = page_zone(page);
	spin_lock_irqsave(&zone->lock, flags);
	ret = __test_page_isolated_in_pageblock(start_pfn, end_pfn,
						skip_hwpoisoned_pages);
	spin_unlock_irqrestore(&zone->lock, flags);
	return ret ? 0 : -EBUSY;
}

struct page *alloc_migrate_target(struct page *page, unsigned long private,
				  int **resultp)
{
	gfp_t gfp_mask = GFP_USER | __GFP_MOVABLE;

	if (PageHighMem(page))
		gfp_mask |= __GFP_HIGHMEM;

	return alloc_page(gfp_mask);
}
Commit	Line	Data
a5d76b54 KH	1	/*
	2	* linux/mm/page_isolation.c
	3	*/
	4
a5d76b54 KH	5	#include <linux/mm.h>
	6	#include <linux/page-isolation.h>
	7	#include <linux/pageblock-flags.h>
ee6f509c	8	#include <linux/memory.h>
a5d76b54 KH	9	#include "internal.h"
a5d76b54 KH	10
702d1a6e MK	11	/* called while holding zone->lock */
	12	static void set_pageblock_isolate(struct page *page)
	13	{
	14	if (get_pageblock_migratetype(page) == MIGRATE_ISOLATE)
	15	return;
	16
	17	set_pageblock_migratetype(page, MIGRATE_ISOLATE);
	18	page_zone(page)->nr_pageblock_isolate++;
	19	}
	20
	21	/* called while holding zone->lock */
	22	static void restore_pageblock_isolate(struct page *page, int migratetype)
	23	{
	24	struct zone *zone = page_zone(page);
	25	if (WARN_ON(get_pageblock_migratetype(page) != MIGRATE_ISOLATE))
	26	return;
	27
	28	BUG_ON(zone->nr_pageblock_isolate <= 0);
	29	set_pageblock_migratetype(page, migratetype);
	30	zone->nr_pageblock_isolate--;
	31	}
	32
b023f468	33	int set_migratetype_isolate(struct page *page, bool skip_hwpoisoned_pages)
ee6f509c MK	34	{
	35	struct zone *zone;
	36	unsigned long flags, pfn;
	37	struct memory_isolate_notify arg;
	38	int notifier_ret;
	39	int ret = -EBUSY;
	40
	41	zone = page_zone(page);
	42
	43	spin_lock_irqsave(&zone->lock, flags);
	44
	45	pfn = page_to_pfn(page);
	46	arg.start_pfn = pfn;
	47	arg.nr_pages = pageblock_nr_pages;
	48	arg.pages_found = 0;
	49
	50	/*
	51	* It may be possible to isolate a pageblock even if the
	52	* migratetype is not MIGRATE_MOVABLE. The memory isolation
	53	* notifier chain is used by balloon drivers to return the
	54	* number of pages in a range that are held by the balloon
	55	* driver to shrink memory. If all the pages are accounted for
	56	* by balloons, are free, or on the LRU, isolation can continue.
	57	* Later, for example, when memory hotplug notifier runs, these
	58	* pages reported as "can be isolated" should be isolated(freed)
	59	* by the balloon driver through the memory notifier chain.
	60	*/
	61	notifier_ret = memory_isolate_notify(MEM_ISOLATE_COUNT, &arg);
	62	notifier_ret = notifier_to_errno(notifier_ret);
	63	if (notifier_ret)
	64	goto out;
	65	/*
	66	* FIXME: Now, memory hotplug doesn't call shrink_slab() by itself.
	67	* We just check MOVABLE pages.
	68	*/
b023f468 WC	69	if (!has_unmovable_pages(zone, page, arg.pages_found,
b023f468 WC	70	skip_hwpoisoned_pages))
ee6f509c MK	71	ret = 0;
	72
	73	/*
	74	* immobile means "not-on-lru" paes. If immobile is larger than
	75	* removable-by-driver pages reported by notifier, we'll fail.
	76	*/
	77
	78	out:
	79	if (!ret) {
2139cbe6	80	unsigned long nr_pages;
d1ce749a	81	int migratetype = get_pageblock_migratetype(page);
2139cbe6	82
702d1a6e	83	set_pageblock_isolate(page);
2139cbe6 BZ	84	nr_pages = move_freepages_block(zone, page, MIGRATE_ISOLATE);
2139cbe6 BZ	85
d1ce749a	86	__mod_zone_freepage_state(zone, -nr_pages, migratetype);
ee6f509c MK	87	}
	88
	89	spin_unlock_irqrestore(&zone->lock, flags);
	90	if (!ret)
	91	drain_all_pages();
	92	return ret;
	93	}
	94
	95	void unset_migratetype_isolate(struct page *page, unsigned migratetype)
	96	{
	97	struct zone *zone;
2139cbe6 BZ	98	unsigned long flags, nr_pages;
2139cbe6 BZ	99
ee6f509c MK	100	zone = page_zone(page);
	101	spin_lock_irqsave(&zone->lock, flags);
	102	if (get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
	103	goto out;
2139cbe6	104	nr_pages = move_freepages_block(zone, page, migratetype);
d1ce749a	105	__mod_zone_freepage_state(zone, nr_pages, migratetype);
702d1a6e	106	restore_pageblock_isolate(page, migratetype);
ee6f509c MK	107	out:
	108	spin_unlock_irqrestore(&zone->lock, flags);
	109	}
	110
a5d76b54 KH	111	static inline struct page *
	112	__first_valid_page(unsigned long pfn, unsigned long nr_pages)
	113	{
	114	int i;
	115	for (i = 0; i < nr_pages; i++)
	116	if (pfn_valid_within(pfn + i))
	117	break;
	118	if (unlikely(i == nr_pages))
	119	return NULL;
	120	return pfn_to_page(pfn + i);
	121	}
	122
	123	/*
	124	* start_isolate_page_range() -- make page-allocation-type of range of pages
	125	* to be MIGRATE_ISOLATE.
	126	* @start_pfn: The lower PFN of the range to be isolated.
	127	* @end_pfn: The upper PFN of the range to be isolated.
0815f3d8	128	* @migratetype: migrate type to set in error recovery.
a5d76b54 KH	129	*
	130	* Making page-allocation-type to be MIGRATE_ISOLATE means free pages in
	131	* the range will never be allocated. Any free pages and pages freed in the
	132	* future will not be allocated again.
	133	*
	134	* start_pfn/end_pfn must be aligned to pageblock_order.
	135	* Returns 0 on success and -EBUSY if any part of range cannot be isolated.
	136	*/
0815f3d8	137	int start_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
b023f468	138	unsigned migratetype, bool skip_hwpoisoned_pages)
a5d76b54 KH	139	{
	140	unsigned long pfn;
	141	unsigned long undo_pfn;
	142	struct page *page;
	143
	144	BUG_ON((start_pfn) & (pageblock_nr_pages - 1));
	145	BUG_ON((end_pfn) & (pageblock_nr_pages - 1));
	146
	147	for (pfn = start_pfn;
	148	pfn < end_pfn;
	149	pfn += pageblock_nr_pages) {
	150	page = __first_valid_page(pfn, pageblock_nr_pages);
b023f468 WC	151	if (page &&
b023f468 WC	152	set_migratetype_isolate(page, skip_hwpoisoned_pages)) {
a5d76b54 KH	153	undo_pfn = pfn;
	154	goto undo;
	155	}
	156	}
	157	return 0;
	158	undo:
	159	for (pfn = start_pfn;
dbc0e4ce	160	pfn < undo_pfn;
a5d76b54	161	pfn += pageblock_nr_pages)
0815f3d8	162	unset_migratetype_isolate(pfn_to_page(pfn), migratetype);
a5d76b54 KH	163
	164	return -EBUSY;
	165	}
	166
	167	/*
	168	* Make isolated pages available again.
	169	*/
0815f3d8 MN	170	int undo_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
0815f3d8 MN	171	unsigned migratetype)
a5d76b54 KH	172	{
	173	unsigned long pfn;
	174	struct page *page;
	175	BUG_ON((start_pfn) & (pageblock_nr_pages - 1));
	176	BUG_ON((end_pfn) & (pageblock_nr_pages - 1));
	177	for (pfn = start_pfn;
	178	pfn < end_pfn;
	179	pfn += pageblock_nr_pages) {
	180	page = __first_valid_page(pfn, pageblock_nr_pages);
dbc0e4ce	181	if (!page \|\| get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
a5d76b54	182	continue;
0815f3d8	183	unset_migratetype_isolate(page, migratetype);
a5d76b54 KH	184	}
	185	return 0;
	186	}
	187	/*
	188	* Test all pages in the range is free(means isolated) or not.
	189	* all pages in [start_pfn...end_pfn) must be in the same zone.
	190	* zone->lock must be held before call this.
	191	*
0815f3d8	192	* Returns 1 if all pages in the range are isolated.
a5d76b54 KH	193	*/
a5d76b54 KH	194	static int
b023f468 WC	195	__test_page_isolated_in_pageblock(unsigned long pfn, unsigned long end_pfn,
b023f468 WC	196	bool skip_hwpoisoned_pages)
a5d76b54 KH	197	{
	198	struct page *page;
	199
	200	while (pfn < end_pfn) {
	201	if (!pfn_valid_within(pfn)) {
	202	pfn++;
	203	continue;
	204	}
	205	page = pfn_to_page(pfn);
41d575ad	206	if (PageBuddy(page)) {
435b405c MK	207	/*
	208	* If race between isolatation and allocation happens,
	209	* some free pages could be in MIGRATE_MOVABLE list
	210	* although pageblock's migratation type of the page
	211	* is MIGRATE_ISOLATE. Catch it and move the page into
	212	* MIGRATE_ISOLATE list.
	213	*/
	214	if (get_freepage_migratetype(page) != MIGRATE_ISOLATE) {
	215	struct page *end_page;
	216
	217	end_page = page + (1 << page_order(page)) - 1;
	218	move_freepages(page_zone(page), page, end_page,
	219	MIGRATE_ISOLATE);
	220	}
a5d76b54	221	pfn += 1 << page_order(page);
41d575ad	222	}
a5d76b54	223	else if (page_count(page) == 0 &&
b12c4ad1	224	get_freepage_migratetype(page) == MIGRATE_ISOLATE)
a5d76b54	225	pfn += 1;
b023f468 WC	226	else if (skip_hwpoisoned_pages && PageHWPoison(page)) {
	227	/*
	228	* The HWPoisoned page may be not in buddy
	229	* system, and page_count() is not 0.
	230	*/
	231	pfn++;
	232	continue;
	233	}
a5d76b54 KH	234	else
	235	break;
	236	}
	237	if (pfn < end_pfn)
	238	return 0;
	239	return 1;
	240	}
	241
b023f468 WC	242	int test_pages_isolated(unsigned long start_pfn, unsigned long end_pfn,
b023f468 WC	243	bool skip_hwpoisoned_pages)
a5d76b54	244	{
6c1b7f68	245	unsigned long pfn, flags;
a5d76b54	246	struct page *page;
6c1b7f68 GS	247	struct zone *zone;
6c1b7f68 GS	248	int ret;
a5d76b54	249
a5d76b54 KH	250	/*
	251	* Note: pageblock_nr_page != MAX_ORDER. Then, chunks of free page
	252	* is not aligned to pageblock_nr_pages.
	253	* Then we just check pagetype fist.
	254	*/
	255	for (pfn = start_pfn; pfn < end_pfn; pfn += pageblock_nr_pages) {
	256	page = __first_valid_page(pfn, pageblock_nr_pages);
dbc0e4ce	257	if (page && get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
a5d76b54 KH	258	break;
a5d76b54 KH	259	}
a70dcb96 GS	260	page = __first_valid_page(start_pfn, end_pfn - start_pfn);
a70dcb96 GS	261	if ((pfn < end_pfn) \|\| !page)
a5d76b54 KH	262	return -EBUSY;
a5d76b54 KH	263	/* Check all pages are free or Marked as ISOLATED */
a70dcb96	264	zone = page_zone(page);
6c1b7f68	265	spin_lock_irqsave(&zone->lock, flags);
b023f468 WC	266	ret = __test_page_isolated_in_pageblock(start_pfn, end_pfn,
b023f468 WC	267	skip_hwpoisoned_pages);
6c1b7f68 GS	268	spin_unlock_irqrestore(&zone->lock, flags);
6c1b7f68 GS	269	return ret ? 0 : -EBUSY;
a5d76b54	270	}
723a0644 MK	271
	272	struct page alloc_migrate_target(struct page page, unsigned long private,
	273	int **resultp)
	274	{
	275	gfp_t gfp_mask = GFP_USER \| __GFP_MOVABLE;
	276
	277	if (PageHighMem(page))
	278	gfp_mask \|= __GFP_HIGHMEM;
	279
	280	return alloc_page(gfp_mask);
	281	}