page_isolation.c 8.0 KB
Newer Older
K
KAMEZAWA Hiroyuki 已提交
1 2 3 4 5 6 7
/*
 * linux/mm/page_isolation.c
 */

#include <linux/mm.h>
#include <linux/page-isolation.h>
#include <linux/pageblock-flags.h>
8
#include <linux/memory.h>
9
#include <linux/hugetlb.h>
10
#include <linux/page_owner.h>
11
#include <linux/migrate.h>
K
KAMEZAWA Hiroyuki 已提交
12 13
#include "internal.h"

14 15 16
#define CREATE_TRACE_POINTS
#include <trace/events/page_isolation.h>

17 18
static int set_migratetype_isolate(struct page *page,
				bool skip_hwpoisoned_pages)
19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
{
	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.
	 */
54 55
	if (!has_unmovable_pages(zone, page, arg.pages_found,
				 skip_hwpoisoned_pages))
56 57 58
		ret = 0;

	/*
59
	 * immobile means "not-on-lru" pages. If immobile is larger than
60 61 62 63 64
	 * removable-by-driver pages reported by notifier, we'll fail.
	 */

out:
	if (!ret) {
65
		unsigned long nr_pages;
66
		int migratetype = get_pageblock_migratetype(page);
67

68
		set_pageblock_migratetype(page, MIGRATE_ISOLATE);
69
		zone->nr_isolate_pageblock++;
70 71
		nr_pages = move_freepages_block(zone, page, MIGRATE_ISOLATE,
									NULL);
72

73
		__mod_zone_freepage_state(zone, -nr_pages, migratetype);
74 75 76 77
	}

	spin_unlock_irqrestore(&zone->lock, flags);
	if (!ret)
78
		drain_all_pages(zone);
79 80 81
	return ret;
}

82
static void unset_migratetype_isolate(struct page *page, unsigned migratetype)
83 84
{
	struct zone *zone;
85
	unsigned long flags, nr_pages;
86
	bool isolated_page = false;
87
	unsigned int order;
88
	unsigned long pfn, buddy_pfn;
89
	struct page *buddy;
90

91 92
	zone = page_zone(page);
	spin_lock_irqsave(&zone->lock, flags);
93
	if (!is_migrate_isolate_page(page))
94
		goto out;
95 96 97 98 99 100 101 102 103 104 105 106

	/*
	 * Because freepage with more than pageblock_order on isolated
	 * pageblock is restricted to merge due to freepage counting problem,
	 * it is possible that there is free buddy page.
	 * move_freepages_block() doesn't care of merge so we need other
	 * approach in order to merge them. Isolation and free will make
	 * these pages to be merged.
	 */
	if (PageBuddy(page)) {
		order = page_order(page);
		if (order >= pageblock_order) {
107 108 109
			pfn = page_to_pfn(page);
			buddy_pfn = __find_buddy_pfn(pfn, order);
			buddy = page + (buddy_pfn - pfn);
110

111
			if (pfn_valid_within(buddy_pfn) &&
112
			    !is_migrate_isolate_page(buddy)) {
113
				__isolate_free_page(page, order);
114
				isolated_page = true;
115 116 117 118 119 120 121 122 123 124
			}
		}
	}

	/*
	 * If we isolate freepage with more than pageblock_order, there
	 * should be no freepage in the range, so we could avoid costly
	 * pageblock scanning for freepage moving.
	 */
	if (!isolated_page) {
125
		nr_pages = move_freepages_block(zone, page, migratetype, NULL);
126 127
		__mod_zone_freepage_state(zone, nr_pages, migratetype);
	}
128
	set_pageblock_migratetype(page, migratetype);
129
	zone->nr_isolate_pageblock--;
130 131
out:
	spin_unlock_irqrestore(&zone->lock, flags);
132
	if (isolated_page) {
133
		post_alloc_hook(page, order, __GFP_MOVABLE);
134
		__free_pages(page, order);
135
	}
136 137
}

K
KAMEZAWA Hiroyuki 已提交
138 139 140 141
static inline struct page *
__first_valid_page(unsigned long pfn, unsigned long nr_pages)
{
	int i;
142 143 144 145 146 147 148 149 150 151 152 153

	for (i = 0; i < nr_pages; i++) {
		struct page *page;

		if (!pfn_valid_within(pfn + i))
			continue;
		page = pfn_to_online_page(pfn + i);
		if (!page)
			continue;
		return page;
	}
	return NULL;
K
KAMEZAWA Hiroyuki 已提交
154 155 156 157 158 159 160
}

/*
 * 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.
161
 * @migratetype: migrate type to set in error recovery.
K
KAMEZAWA Hiroyuki 已提交
162 163 164 165 166 167 168 169
 *
 * 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.
 */
170
int start_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
171
			     unsigned migratetype, bool skip_hwpoisoned_pages)
K
KAMEZAWA Hiroyuki 已提交
172 173 174 175 176
{
	unsigned long pfn;
	unsigned long undo_pfn;
	struct page *page;

177 178
	BUG_ON(!IS_ALIGNED(start_pfn, pageblock_nr_pages));
	BUG_ON(!IS_ALIGNED(end_pfn, pageblock_nr_pages));
K
KAMEZAWA Hiroyuki 已提交
179 180 181 182 183

	for (pfn = start_pfn;
	     pfn < end_pfn;
	     pfn += pageblock_nr_pages) {
		page = __first_valid_page(pfn, pageblock_nr_pages);
184 185
		if (page &&
		    set_migratetype_isolate(page, skip_hwpoisoned_pages)) {
K
KAMEZAWA Hiroyuki 已提交
186 187 188 189 190 191 192
			undo_pfn = pfn;
			goto undo;
		}
	}
	return 0;
undo:
	for (pfn = start_pfn;
193
	     pfn < undo_pfn;
194 195 196 197 198 199
	     pfn += pageblock_nr_pages) {
		struct page *page = pfn_to_online_page(pfn);
		if (!page)
			continue;
		unset_migratetype_isolate(page, migratetype);
	}
K
KAMEZAWA Hiroyuki 已提交
200 201 202 203 204 205 206

	return -EBUSY;
}

/*
 * Make isolated pages available again.
 */
207 208
int undo_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
			    unsigned migratetype)
K
KAMEZAWA Hiroyuki 已提交
209 210 211
{
	unsigned long pfn;
	struct page *page;
212 213 214 215

	BUG_ON(!IS_ALIGNED(start_pfn, pageblock_nr_pages));
	BUG_ON(!IS_ALIGNED(end_pfn, pageblock_nr_pages));

K
KAMEZAWA Hiroyuki 已提交
216 217 218 219
	for (pfn = start_pfn;
	     pfn < end_pfn;
	     pfn += pageblock_nr_pages) {
		page = __first_valid_page(pfn, pageblock_nr_pages);
220
		if (!page || !is_migrate_isolate_page(page))
K
KAMEZAWA Hiroyuki 已提交
221
			continue;
222
		unset_migratetype_isolate(page, migratetype);
K
KAMEZAWA Hiroyuki 已提交
223 224 225 226 227 228 229 230
	}
	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.
 *
231
 * Returns the last tested pfn.
K
KAMEZAWA Hiroyuki 已提交
232
 */
233
static unsigned long
234 235
__test_page_isolated_in_pageblock(unsigned long pfn, unsigned long end_pfn,
				  bool skip_hwpoisoned_pages)
K
KAMEZAWA Hiroyuki 已提交
236 237 238 239 240 241 242 243 244
{
	struct page *page;

	while (pfn < end_pfn) {
		if (!pfn_valid_within(pfn)) {
			pfn++;
			continue;
		}
		page = pfn_to_page(pfn);
245
		if (PageBuddy(page))
246
			/*
247 248 249
			 * If the page is on a free list, it has to be on
			 * the correct MIGRATE_ISOLATE freelist. There is no
			 * simple way to verify that as VM_BUG_ON(), though.
250
			 */
K
KAMEZAWA Hiroyuki 已提交
251
			pfn += 1 << page_order(page);
252 253
		else if (skip_hwpoisoned_pages && PageHWPoison(page))
			/* A HWPoisoned page cannot be also PageBuddy */
254
			pfn++;
K
KAMEZAWA Hiroyuki 已提交
255 256 257
		else
			break;
	}
258 259

	return pfn;
K
KAMEZAWA Hiroyuki 已提交
260 261
}

262
/* Caller should ensure that requested range is in a single zone */
263 264
int test_pages_isolated(unsigned long start_pfn, unsigned long end_pfn,
			bool skip_hwpoisoned_pages)
K
KAMEZAWA Hiroyuki 已提交
265
{
266
	unsigned long pfn, flags;
K
KAMEZAWA Hiroyuki 已提交
267
	struct page *page;
268
	struct zone *zone;
K
KAMEZAWA Hiroyuki 已提交
269 270

	/*
271 272 273
	 * Note: pageblock_nr_pages != MAX_ORDER. Then, chunks of free pages
	 * are not aligned to pageblock_nr_pages.
	 * Then we just check migratetype first.
K
KAMEZAWA Hiroyuki 已提交
274 275 276
	 */
	for (pfn = start_pfn; pfn < end_pfn; pfn += pageblock_nr_pages) {
		page = __first_valid_page(pfn, pageblock_nr_pages);
277
		if (page && !is_migrate_isolate_page(page))
K
KAMEZAWA Hiroyuki 已提交
278 279
			break;
	}
280 281
	page = __first_valid_page(start_pfn, end_pfn - start_pfn);
	if ((pfn < end_pfn) || !page)
K
KAMEZAWA Hiroyuki 已提交
282
		return -EBUSY;
283
	/* Check all pages are free or marked as ISOLATED */
284
	zone = page_zone(page);
285
	spin_lock_irqsave(&zone->lock, flags);
286
	pfn = __test_page_isolated_in_pageblock(start_pfn, end_pfn,
287
						skip_hwpoisoned_pages);
288
	spin_unlock_irqrestore(&zone->lock, flags);
289

290 291
	trace_test_pages_isolated(start_pfn, end_pfn, pfn);

292
	return pfn < end_pfn ? -EBUSY : 0;
K
KAMEZAWA Hiroyuki 已提交
293
}
294 295 296 297

struct page *alloc_migrate_target(struct page *page, unsigned long private,
				  int **resultp)
{
298
	return new_page_nodemask(page, numa_node_id(), &node_states[N_MEMORY]);
299
}