page_isolation.c 8.36 KB
Newer Older
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 12
#include "internal.h"

13 14 15
#define CREATE_TRACE_POINTS
#include <trace/events/page_isolation.h>

16 17
static int set_migratetype_isolate(struct page *page,
				bool skip_hwpoisoned_pages)
18 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
{
	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.
	 */
53 54
	if (!has_unmovable_pages(zone, page, arg.pages_found,
				 skip_hwpoisoned_pages))
55 56 57
		ret = 0;

	/*
58
	 * immobile means "not-on-lru" pages. If immobile is larger than
59 60 61 62 63
	 * removable-by-driver pages reported by notifier, we'll fail.
	 */

out:
	if (!ret) {
64
		unsigned long nr_pages;
65
		int migratetype = get_pageblock_migratetype(page);
66

67
		set_pageblock_migratetype(page, MIGRATE_ISOLATE);
68
		zone->nr_isolate_pageblock++;
69 70
		nr_pages = move_freepages_block(zone, page, MIGRATE_ISOLATE);

71
		__mod_zone_freepage_state(zone, -nr_pages, migratetype);
72 73 74 75
	}

	spin_unlock_irqrestore(&zone->lock, flags);
	if (!ret)
76
		drain_all_pages(zone);
77 78 79
	return ret;
}

80
static void unset_migratetype_isolate(struct page *page, unsigned migratetype)
81 82
{
	struct zone *zone;
83
	unsigned long flags, nr_pages;
84
	bool isolated_page = false;
85 86 87
	unsigned int order;
	unsigned long page_idx, buddy_idx;
	struct page *buddy;
88

89 90 91 92
	zone = page_zone(page);
	spin_lock_irqsave(&zone->lock, flags);
	if (get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
		goto out;
93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108

	/*
	 * 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) {
			page_idx = page_to_pfn(page) & ((1 << MAX_ORDER) - 1);
			buddy_idx = __find_buddy_index(page_idx, order);
			buddy = page + (buddy_idx - page_idx);

109 110
			if (pfn_valid_within(page_to_pfn(buddy)) &&
			    !is_migrate_isolate_page(buddy)) {
111
				__isolate_free_page(page, order);
112
				isolated_page = true;
113 114 115 116 117 118 119 120 121 122 123 124 125
			}
		}
	}

	/*
	 * 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) {
		nr_pages = move_freepages_block(zone, page, migratetype);
		__mod_zone_freepage_state(zone, nr_pages, migratetype);
	}
126
	set_pageblock_migratetype(page, migratetype);
127
	zone->nr_isolate_pageblock--;
128 129
out:
	spin_unlock_irqrestore(&zone->lock, flags);
130
	if (isolated_page) {
131
		post_alloc_hook(page, order, __GFP_MOVABLE);
132
		__free_pages(page, order);
133
	}
134 135
}

136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152
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.
153
 * @migratetype: migrate type to set in error recovery.
154 155 156 157 158 159 160 161
 *
 * 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.
 */
162
int start_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
163
			     unsigned migratetype, bool skip_hwpoisoned_pages)
164 165 166 167 168
{
	unsigned long pfn;
	unsigned long undo_pfn;
	struct page *page;

169 170
	BUG_ON(!IS_ALIGNED(start_pfn, pageblock_nr_pages));
	BUG_ON(!IS_ALIGNED(end_pfn, pageblock_nr_pages));
171 172 173 174 175

	for (pfn = start_pfn;
	     pfn < end_pfn;
	     pfn += pageblock_nr_pages) {
		page = __first_valid_page(pfn, pageblock_nr_pages);
176 177
		if (page &&
		    set_migratetype_isolate(page, skip_hwpoisoned_pages)) {
178 179 180 181 182 183 184
			undo_pfn = pfn;
			goto undo;
		}
	}
	return 0;
undo:
	for (pfn = start_pfn;
185
	     pfn < undo_pfn;
186
	     pfn += pageblock_nr_pages)
187
		unset_migratetype_isolate(pfn_to_page(pfn), migratetype);
188 189 190 191 192 193 194

	return -EBUSY;
}

/*
 * Make isolated pages available again.
 */
195 196
int undo_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
			    unsigned migratetype)
197 198 199
{
	unsigned long pfn;
	struct page *page;
200 201 202 203

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

204 205 206 207
	for (pfn = start_pfn;
	     pfn < end_pfn;
	     pfn += pageblock_nr_pages) {
		page = __first_valid_page(pfn, pageblock_nr_pages);
208
		if (!page || get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
209
			continue;
210
		unset_migratetype_isolate(page, migratetype);
211 212 213 214 215 216 217 218
	}
	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.
 *
219
 * Returns the last tested pfn.
220
 */
221
static unsigned long
222 223
__test_page_isolated_in_pageblock(unsigned long pfn, unsigned long end_pfn,
				  bool skip_hwpoisoned_pages)
224 225 226 227 228 229 230 231 232
{
	struct page *page;

	while (pfn < end_pfn) {
		if (!pfn_valid_within(pfn)) {
			pfn++;
			continue;
		}
		page = pfn_to_page(pfn);
233
		if (PageBuddy(page))
234
			/*
235 236 237
			 * 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.
238
			 */
239
			pfn += 1 << page_order(page);
240 241
		else if (skip_hwpoisoned_pages && PageHWPoison(page))
			/* A HWPoisoned page cannot be also PageBuddy */
242
			pfn++;
243 244 245
		else
			break;
	}
246 247

	return pfn;
248 249
}

250
/* Caller should ensure that requested range is in a single zone */
251 252
int test_pages_isolated(unsigned long start_pfn, unsigned long end_pfn,
			bool skip_hwpoisoned_pages)
253
{
254
	unsigned long pfn, flags;
255
	struct page *page;
256
	struct zone *zone;
257 258

	/*
259 260 261
	 * Note: pageblock_nr_pages != MAX_ORDER. Then, chunks of free pages
	 * are not aligned to pageblock_nr_pages.
	 * Then we just check migratetype first.
262 263 264
	 */
	for (pfn = start_pfn; pfn < end_pfn; pfn += pageblock_nr_pages) {
		page = __first_valid_page(pfn, pageblock_nr_pages);
265
		if (page && get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
266 267
			break;
	}
268 269
	page = __first_valid_page(start_pfn, end_pfn - start_pfn);
	if ((pfn < end_pfn) || !page)
270
		return -EBUSY;
271
	/* Check all pages are free or marked as ISOLATED */
272
	zone = page_zone(page);
273
	spin_lock_irqsave(&zone->lock, flags);
274
	pfn = __test_page_isolated_in_pageblock(start_pfn, end_pfn,
275
						skip_hwpoisoned_pages);
276
	spin_unlock_irqrestore(&zone->lock, flags);
277

278 279
	trace_test_pages_isolated(start_pfn, end_pfn, pfn);

280
	return pfn < end_pfn ? -EBUSY : 0;
281
}
282 283 284 285 286 287

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

288 289 290 291 292
	/*
	 * TODO: allocate a destination hugepage from a nearest neighbor node,
	 * accordance with memory policy of the user process if possible. For
	 * now as a simple work-around, we use the next node for destination.
	 */
293
	if (PageHuge(page))
294
		return alloc_huge_page_node(page_hstate(compound_head(page)),
295 296
					    next_node_in(page_to_nid(page),
							 node_online_map));
297

298 299 300 301 302
	if (PageHighMem(page))
		gfp_mask |= __GFP_HIGHMEM;

	return alloc_page(gfp_mask);
}