hugetlb.c 99.9 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 * Generic hugetlb support.
3
 * (C) Nadia Yvette Chambers, April 2004
L
Linus Torvalds 已提交
4 5 6 7 8
 */
#include <linux/list.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/mm.h>
9
#include <linux/seq_file.h>
L
Linus Torvalds 已提交
10 11
#include <linux/sysctl.h>
#include <linux/highmem.h>
A
Andrea Arcangeli 已提交
12
#include <linux/mmu_notifier.h>
L
Linus Torvalds 已提交
13
#include <linux/nodemask.h>
D
David Gibson 已提交
14
#include <linux/pagemap.h>
15
#include <linux/mempolicy.h>
16
#include <linux/compiler.h>
17
#include <linux/cpuset.h>
18
#include <linux/mutex.h>
19
#include <linux/bootmem.h>
20
#include <linux/sysfs.h>
21
#include <linux/slab.h>
22
#include <linux/rmap.h>
23 24
#include <linux/swap.h>
#include <linux/swapops.h>
25
#include <linux/page-isolation.h>
26
#include <linux/jhash.h>
27

D
David Gibson 已提交
28 29
#include <asm/page.h>
#include <asm/pgtable.h>
30
#include <asm/tlb.h>
D
David Gibson 已提交
31

32
#include <linux/io.h>
D
David Gibson 已提交
33
#include <linux/hugetlb.h>
34
#include <linux/hugetlb_cgroup.h>
35
#include <linux/node.h>
36
#include "internal.h"
L
Linus Torvalds 已提交
37

38
int hugepages_treat_as_movable;
39

40
int hugetlb_max_hstate __read_mostly;
41 42 43
unsigned int default_hstate_idx;
struct hstate hstates[HUGE_MAX_HSTATE];

44 45
__initdata LIST_HEAD(huge_boot_pages);

46 47 48
/* for command line parsing */
static struct hstate * __initdata parsed_hstate;
static unsigned long __initdata default_hstate_max_huge_pages;
49
static unsigned long __initdata default_hstate_size;
50

51
/*
52 53
 * Protects updates to hugepage_freelists, hugepage_activelist, nr_huge_pages,
 * free_huge_pages, and surplus_huge_pages.
54
 */
55
DEFINE_SPINLOCK(hugetlb_lock);
56

57 58 59 60 61 62 63
/*
 * Serializes faults on the same logical page.  This is used to
 * prevent spurious OOMs when the hugepage pool is fully utilized.
 */
static int num_fault_mutexes;
static struct mutex *htlb_fault_mutex_table ____cacheline_aligned_in_smp;

64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138
static inline void unlock_or_release_subpool(struct hugepage_subpool *spool)
{
	bool free = (spool->count == 0) && (spool->used_hpages == 0);

	spin_unlock(&spool->lock);

	/* If no pages are used, and no other handles to the subpool
	 * remain, free the subpool the subpool remain */
	if (free)
		kfree(spool);
}

struct hugepage_subpool *hugepage_new_subpool(long nr_blocks)
{
	struct hugepage_subpool *spool;

	spool = kmalloc(sizeof(*spool), GFP_KERNEL);
	if (!spool)
		return NULL;

	spin_lock_init(&spool->lock);
	spool->count = 1;
	spool->max_hpages = nr_blocks;
	spool->used_hpages = 0;

	return spool;
}

void hugepage_put_subpool(struct hugepage_subpool *spool)
{
	spin_lock(&spool->lock);
	BUG_ON(!spool->count);
	spool->count--;
	unlock_or_release_subpool(spool);
}

static int hugepage_subpool_get_pages(struct hugepage_subpool *spool,
				      long delta)
{
	int ret = 0;

	if (!spool)
		return 0;

	spin_lock(&spool->lock);
	if ((spool->used_hpages + delta) <= spool->max_hpages) {
		spool->used_hpages += delta;
	} else {
		ret = -ENOMEM;
	}
	spin_unlock(&spool->lock);

	return ret;
}

static void hugepage_subpool_put_pages(struct hugepage_subpool *spool,
				       long delta)
{
	if (!spool)
		return;

	spin_lock(&spool->lock);
	spool->used_hpages -= delta;
	/* If hugetlbfs_put_super couldn't free spool due to
	* an outstanding quota reference, free it now. */
	unlock_or_release_subpool(spool);
}

static inline struct hugepage_subpool *subpool_inode(struct inode *inode)
{
	return HUGETLBFS_SB(inode->i_sb)->spool;
}

static inline struct hugepage_subpool *subpool_vma(struct vm_area_struct *vma)
{
A
Al Viro 已提交
139
	return subpool_inode(file_inode(vma->vm_file));
140 141
}

142 143 144
/*
 * Region tracking -- allows tracking of reservations and instantiated pages
 *                    across the pages in a mapping.
145
 *
146 147
 * The region data structures are embedded into a resv_map and
 * protected by a resv_map's lock
148 149 150 151 152 153 154
 */
struct file_region {
	struct list_head link;
	long from;
	long to;
};

155
static long region_add(struct resv_map *resv, long f, long t)
156
{
157
	struct list_head *head = &resv->regions;
158 159
	struct file_region *rg, *nrg, *trg;

160
	spin_lock(&resv->lock);
161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189
	/* Locate the region we are either in or before. */
	list_for_each_entry(rg, head, link)
		if (f <= rg->to)
			break;

	/* Round our left edge to the current segment if it encloses us. */
	if (f > rg->from)
		f = rg->from;

	/* Check for and consume any regions we now overlap with. */
	nrg = rg;
	list_for_each_entry_safe(rg, trg, rg->link.prev, link) {
		if (&rg->link == head)
			break;
		if (rg->from > t)
			break;

		/* If this area reaches higher then extend our area to
		 * include it completely.  If this is not the first area
		 * which we intend to reuse, free it. */
		if (rg->to > t)
			t = rg->to;
		if (rg != nrg) {
			list_del(&rg->link);
			kfree(rg);
		}
	}
	nrg->from = f;
	nrg->to = t;
190
	spin_unlock(&resv->lock);
191 192 193
	return 0;
}

194
static long region_chg(struct resv_map *resv, long f, long t)
195
{
196
	struct list_head *head = &resv->regions;
197
	struct file_region *rg, *nrg = NULL;
198 199
	long chg = 0;

200 201
retry:
	spin_lock(&resv->lock);
202 203 204 205 206 207 208 209 210
	/* Locate the region we are before or in. */
	list_for_each_entry(rg, head, link)
		if (f <= rg->to)
			break;

	/* If we are below the current region then a new region is required.
	 * Subtle, allocate a new region at the position but make it zero
	 * size such that we can guarantee to record the reservation. */
	if (&rg->link == head || t < rg->from) {
211 212 213 214 215 216 217 218 219 220 221
		if (!nrg) {
			spin_unlock(&resv->lock);
			nrg = kmalloc(sizeof(*nrg), GFP_KERNEL);
			if (!nrg)
				return -ENOMEM;

			nrg->from = f;
			nrg->to   = f;
			INIT_LIST_HEAD(&nrg->link);
			goto retry;
		}
222

223 224 225
		list_add(&nrg->link, rg->link.prev);
		chg = t - f;
		goto out_nrg;
226 227 228 229 230 231 232 233 234 235 236 237
	}

	/* Round our left edge to the current segment if it encloses us. */
	if (f > rg->from)
		f = rg->from;
	chg = t - f;

	/* Check for and consume any regions we now overlap with. */
	list_for_each_entry(rg, rg->link.prev, link) {
		if (&rg->link == head)
			break;
		if (rg->from > t)
238
			goto out;
239

L
Lucas De Marchi 已提交
240
		/* We overlap with this area, if it extends further than
241 242 243 244 245 246 247 248
		 * us then we must extend ourselves.  Account for its
		 * existing reservation. */
		if (rg->to > t) {
			chg += rg->to - t;
			t = rg->to;
		}
		chg -= rg->to - rg->from;
	}
249 250 251 252 253 254 255 256

out:
	spin_unlock(&resv->lock);
	/*  We already know we raced and no longer need the new region */
	kfree(nrg);
	return chg;
out_nrg:
	spin_unlock(&resv->lock);
257 258 259
	return chg;
}

260
static long region_truncate(struct resv_map *resv, long end)
261
{
262
	struct list_head *head = &resv->regions;
263 264 265
	struct file_region *rg, *trg;
	long chg = 0;

266
	spin_lock(&resv->lock);
267 268 269 270 271
	/* Locate the region we are either in or before. */
	list_for_each_entry(rg, head, link)
		if (end <= rg->to)
			break;
	if (&rg->link == head)
272
		goto out;
273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288

	/* If we are in the middle of a region then adjust it. */
	if (end > rg->from) {
		chg = rg->to - end;
		rg->to = end;
		rg = list_entry(rg->link.next, typeof(*rg), link);
	}

	/* Drop any remaining regions. */
	list_for_each_entry_safe(rg, trg, rg->link.prev, link) {
		if (&rg->link == head)
			break;
		chg += rg->to - rg->from;
		list_del(&rg->link);
		kfree(rg);
	}
289 290 291

out:
	spin_unlock(&resv->lock);
292 293 294
	return chg;
}

295
static long region_count(struct resv_map *resv, long f, long t)
296
{
297
	struct list_head *head = &resv->regions;
298 299 300
	struct file_region *rg;
	long chg = 0;

301
	spin_lock(&resv->lock);
302 303
	/* Locate each segment we overlap with, and count that overlap. */
	list_for_each_entry(rg, head, link) {
304 305
		long seg_from;
		long seg_to;
306 307 308 309 310 311 312 313 314 315 316

		if (rg->to <= f)
			continue;
		if (rg->from >= t)
			break;

		seg_from = max(rg->from, f);
		seg_to = min(rg->to, t);

		chg += seg_to - seg_from;
	}
317
	spin_unlock(&resv->lock);
318 319 320 321

	return chg;
}

322 323 324 325
/*
 * Convert the address within this vma to the page offset within
 * the mapping, in pagecache page units; huge pages here.
 */
326 327
static pgoff_t vma_hugecache_offset(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
328
{
329 330
	return ((address - vma->vm_start) >> huge_page_shift(h)) +
			(vma->vm_pgoff >> huge_page_order(h));
331 332
}

333 334 335 336 337 338
pgoff_t linear_hugepage_index(struct vm_area_struct *vma,
				     unsigned long address)
{
	return vma_hugecache_offset(hstate_vma(vma), vma, address);
}

339 340 341 342 343 344 345 346 347 348 349 350 351
/*
 * Return the size of the pages allocated when backing a VMA. In the majority
 * cases this will be same size as used by the page table entries.
 */
unsigned long vma_kernel_pagesize(struct vm_area_struct *vma)
{
	struct hstate *hstate;

	if (!is_vm_hugetlb_page(vma))
		return PAGE_SIZE;

	hstate = hstate_vma(vma);

352
	return 1UL << huge_page_shift(hstate);
353
}
354
EXPORT_SYMBOL_GPL(vma_kernel_pagesize);
355

356 357 358 359 360 361 362 363 364 365 366 367 368
/*
 * Return the page size being used by the MMU to back a VMA. In the majority
 * of cases, the page size used by the kernel matches the MMU size. On
 * architectures where it differs, an architecture-specific version of this
 * function is required.
 */
#ifndef vma_mmu_pagesize
unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
{
	return vma_kernel_pagesize(vma);
}
#endif

369 370 371 372 373 374 375
/*
 * Flags for MAP_PRIVATE reservations.  These are stored in the bottom
 * bits of the reservation map pointer, which are always clear due to
 * alignment.
 */
#define HPAGE_RESV_OWNER    (1UL << 0)
#define HPAGE_RESV_UNMAPPED (1UL << 1)
376
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED)
377

378 379 380 381 382 383 384 385 386
/*
 * These helpers are used to track how many pages are reserved for
 * faults in a MAP_PRIVATE mapping. Only the process that called mmap()
 * is guaranteed to have their future faults succeed.
 *
 * With the exception of reset_vma_resv_huge_pages() which is called at fork(),
 * the reserve counters are updated with the hugetlb_lock held. It is safe
 * to reset the VMA at fork() time as it is not in use yet and there is no
 * chance of the global counters getting corrupted as a result of the values.
387 388 389 390 391 392 393 394 395
 *
 * The private mapping reservation is represented in a subtly different
 * manner to a shared mapping.  A shared mapping has a region map associated
 * with the underlying file, this region map represents the backing file
 * pages which have ever had a reservation assigned which this persists even
 * after the page is instantiated.  A private mapping has a region map
 * associated with the original mmap which is attached to all VMAs which
 * reference it, this region map represents those offsets which have consumed
 * reservation ie. where pages have been instantiated.
396
 */
397 398 399 400 401 402 403 404 405 406 407
static unsigned long get_vma_private_data(struct vm_area_struct *vma)
{
	return (unsigned long)vma->vm_private_data;
}

static void set_vma_private_data(struct vm_area_struct *vma,
							unsigned long value)
{
	vma->vm_private_data = (void *)value;
}

408
struct resv_map *resv_map_alloc(void)
409 410 411 412 413 414
{
	struct resv_map *resv_map = kmalloc(sizeof(*resv_map), GFP_KERNEL);
	if (!resv_map)
		return NULL;

	kref_init(&resv_map->refs);
415
	spin_lock_init(&resv_map->lock);
416 417 418 419 420
	INIT_LIST_HEAD(&resv_map->regions);

	return resv_map;
}

421
void resv_map_release(struct kref *ref)
422 423 424 425
{
	struct resv_map *resv_map = container_of(ref, struct resv_map, refs);

	/* Clear out any active regions before we release the map. */
426
	region_truncate(resv_map, 0);
427 428 429
	kfree(resv_map);
}

430 431 432 433 434
static inline struct resv_map *inode_resv_map(struct inode *inode)
{
	return inode->i_mapping->private_data;
}

435
static struct resv_map *vma_resv_map(struct vm_area_struct *vma)
436
{
437
	VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma);
438 439 440 441 442 443 444
	if (vma->vm_flags & VM_MAYSHARE) {
		struct address_space *mapping = vma->vm_file->f_mapping;
		struct inode *inode = mapping->host;

		return inode_resv_map(inode);

	} else {
445 446
		return (struct resv_map *)(get_vma_private_data(vma) &
							~HPAGE_RESV_MASK);
447
	}
448 449
}

450
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map)
451
{
452 453
	VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma);
	VM_BUG_ON_VMA(vma->vm_flags & VM_MAYSHARE, vma);
454

455 456
	set_vma_private_data(vma, (get_vma_private_data(vma) &
				HPAGE_RESV_MASK) | (unsigned long)map);
457 458 459 460
}

static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags)
{
461 462
	VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma);
	VM_BUG_ON_VMA(vma->vm_flags & VM_MAYSHARE, vma);
463 464

	set_vma_private_data(vma, get_vma_private_data(vma) | flags);
465 466 467 468
}

static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag)
{
469
	VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma);
470 471

	return (get_vma_private_data(vma) & flag) != 0;
472 473
}

474
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */
475 476
void reset_vma_resv_huge_pages(struct vm_area_struct *vma)
{
477
	VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma);
478
	if (!(vma->vm_flags & VM_MAYSHARE))
479 480 481 482
		vma->vm_private_data = (void *)0;
}

/* Returns true if the VMA has associated reserve pages */
483
static int vma_has_reserves(struct vm_area_struct *vma, long chg)
484
{
485 486 487 488 489 490 491 492 493 494 495 496 497 498 499
	if (vma->vm_flags & VM_NORESERVE) {
		/*
		 * This address is already reserved by other process(chg == 0),
		 * so, we should decrement reserved count. Without decrementing,
		 * reserve count remains after releasing inode, because this
		 * allocated page will go into page cache and is regarded as
		 * coming from reserved pool in releasing step.  Currently, we
		 * don't have any other solution to deal with this situation
		 * properly, so add work-around here.
		 */
		if (vma->vm_flags & VM_MAYSHARE && chg == 0)
			return 1;
		else
			return 0;
	}
500 501

	/* Shared mappings always use reserves */
502
	if (vma->vm_flags & VM_MAYSHARE)
503
		return 1;
504 505 506 507 508

	/*
	 * Only the process that called mmap() has reserves for
	 * private mappings.
	 */
509 510
	if (is_vma_resv_set(vma, HPAGE_RESV_OWNER))
		return 1;
511

512
	return 0;
513 514
}

515
static void enqueue_huge_page(struct hstate *h, struct page *page)
L
Linus Torvalds 已提交
516 517
{
	int nid = page_to_nid(page);
518
	list_move(&page->lru, &h->hugepage_freelists[nid]);
519 520
	h->free_huge_pages++;
	h->free_huge_pages_node[nid]++;
L
Linus Torvalds 已提交
521 522
}

523 524 525 526
static struct page *dequeue_huge_page_node(struct hstate *h, int nid)
{
	struct page *page;

527 528 529 530 531 532 533 534
	list_for_each_entry(page, &h->hugepage_freelists[nid], lru)
		if (!is_migrate_isolate_page(page))
			break;
	/*
	 * if 'non-isolated free hugepage' not found on the list,
	 * the allocation fails.
	 */
	if (&h->hugepage_freelists[nid] == &page->lru)
535
		return NULL;
536
	list_move(&page->lru, &h->hugepage_activelist);
537
	set_page_refcounted(page);
538 539 540 541 542
	h->free_huge_pages--;
	h->free_huge_pages_node[nid]--;
	return page;
}

543 544 545
/* Movability of hugepages depends on migration support. */
static inline gfp_t htlb_alloc_mask(struct hstate *h)
{
546
	if (hugepages_treat_as_movable || hugepage_migration_supported(h))
547 548 549 550 551
		return GFP_HIGHUSER_MOVABLE;
	else
		return GFP_HIGHUSER;
}

552 553
static struct page *dequeue_huge_page_vma(struct hstate *h,
				struct vm_area_struct *vma,
554 555
				unsigned long address, int avoid_reserve,
				long chg)
L
Linus Torvalds 已提交
556
{
557
	struct page *page = NULL;
558
	struct mempolicy *mpol;
559
	nodemask_t *nodemask;
560
	struct zonelist *zonelist;
561 562
	struct zone *zone;
	struct zoneref *z;
563
	unsigned int cpuset_mems_cookie;
L
Linus Torvalds 已提交
564

565 566 567 568 569
	/*
	 * A child process with MAP_PRIVATE mappings created by their parent
	 * have no page reserves. This check ensures that reservations are
	 * not "stolen". The child may still get SIGKILLed
	 */
570
	if (!vma_has_reserves(vma, chg) &&
571
			h->free_huge_pages - h->resv_huge_pages == 0)
572
		goto err;
573

574
	/* If reserves cannot be used, ensure enough pages are in the pool */
575
	if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0)
576
		goto err;
577

578
retry_cpuset:
579
	cpuset_mems_cookie = read_mems_allowed_begin();
580
	zonelist = huge_zonelist(vma, address,
581
					htlb_alloc_mask(h), &mpol, &nodemask);
582

583 584
	for_each_zone_zonelist_nodemask(zone, z, zonelist,
						MAX_NR_ZONES - 1, nodemask) {
585
		if (cpuset_zone_allowed(zone, htlb_alloc_mask(h))) {
586 587
			page = dequeue_huge_page_node(h, zone_to_nid(zone));
			if (page) {
588 589 590 591 592
				if (avoid_reserve)
					break;
				if (!vma_has_reserves(vma, chg))
					break;

593
				SetPagePrivate(page);
594
				h->resv_huge_pages--;
595 596
				break;
			}
A
Andrew Morton 已提交
597
		}
L
Linus Torvalds 已提交
598
	}
599

600
	mpol_cond_put(mpol);
601
	if (unlikely(!page && read_mems_allowed_retry(cpuset_mems_cookie)))
602
		goto retry_cpuset;
L
Linus Torvalds 已提交
603
	return page;
604 605 606

err:
	return NULL;
L
Linus Torvalds 已提交
607 608
}

609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681
/*
 * common helper functions for hstate_next_node_to_{alloc|free}.
 * We may have allocated or freed a huge page based on a different
 * nodes_allowed previously, so h->next_node_to_{alloc|free} might
 * be outside of *nodes_allowed.  Ensure that we use an allowed
 * node for alloc or free.
 */
static int next_node_allowed(int nid, nodemask_t *nodes_allowed)
{
	nid = next_node(nid, *nodes_allowed);
	if (nid == MAX_NUMNODES)
		nid = first_node(*nodes_allowed);
	VM_BUG_ON(nid >= MAX_NUMNODES);

	return nid;
}

static int get_valid_node_allowed(int nid, nodemask_t *nodes_allowed)
{
	if (!node_isset(nid, *nodes_allowed))
		nid = next_node_allowed(nid, nodes_allowed);
	return nid;
}

/*
 * returns the previously saved node ["this node"] from which to
 * allocate a persistent huge page for the pool and advance the
 * next node from which to allocate, handling wrap at end of node
 * mask.
 */
static int hstate_next_node_to_alloc(struct hstate *h,
					nodemask_t *nodes_allowed)
{
	int nid;

	VM_BUG_ON(!nodes_allowed);

	nid = get_valid_node_allowed(h->next_nid_to_alloc, nodes_allowed);
	h->next_nid_to_alloc = next_node_allowed(nid, nodes_allowed);

	return nid;
}

/*
 * helper for free_pool_huge_page() - return the previously saved
 * node ["this node"] from which to free a huge page.  Advance the
 * next node id whether or not we find a free huge page to free so
 * that the next attempt to free addresses the next node.
 */
static int hstate_next_node_to_free(struct hstate *h, nodemask_t *nodes_allowed)
{
	int nid;

	VM_BUG_ON(!nodes_allowed);

	nid = get_valid_node_allowed(h->next_nid_to_free, nodes_allowed);
	h->next_nid_to_free = next_node_allowed(nid, nodes_allowed);

	return nid;
}

#define for_each_node_mask_to_alloc(hs, nr_nodes, node, mask)		\
	for (nr_nodes = nodes_weight(*mask);				\
		nr_nodes > 0 &&						\
		((node = hstate_next_node_to_alloc(hs, mask)) || 1);	\
		nr_nodes--)

#define for_each_node_mask_to_free(hs, nr_nodes, node, mask)		\
	for (nr_nodes = nodes_weight(*mask);				\
		nr_nodes > 0 &&						\
		((node = hstate_next_node_to_free(hs, mask)) || 1);	\
		nr_nodes--)

682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819
#if defined(CONFIG_CMA) && defined(CONFIG_X86_64)
static void destroy_compound_gigantic_page(struct page *page,
					unsigned long order)
{
	int i;
	int nr_pages = 1 << order;
	struct page *p = page + 1;

	for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) {
		__ClearPageTail(p);
		set_page_refcounted(p);
		p->first_page = NULL;
	}

	set_compound_order(page, 0);
	__ClearPageHead(page);
}

static void free_gigantic_page(struct page *page, unsigned order)
{
	free_contig_range(page_to_pfn(page), 1 << order);
}

static int __alloc_gigantic_page(unsigned long start_pfn,
				unsigned long nr_pages)
{
	unsigned long end_pfn = start_pfn + nr_pages;
	return alloc_contig_range(start_pfn, end_pfn, MIGRATE_MOVABLE);
}

static bool pfn_range_valid_gigantic(unsigned long start_pfn,
				unsigned long nr_pages)
{
	unsigned long i, end_pfn = start_pfn + nr_pages;
	struct page *page;

	for (i = start_pfn; i < end_pfn; i++) {
		if (!pfn_valid(i))
			return false;

		page = pfn_to_page(i);

		if (PageReserved(page))
			return false;

		if (page_count(page) > 0)
			return false;

		if (PageHuge(page))
			return false;
	}

	return true;
}

static bool zone_spans_last_pfn(const struct zone *zone,
			unsigned long start_pfn, unsigned long nr_pages)
{
	unsigned long last_pfn = start_pfn + nr_pages - 1;
	return zone_spans_pfn(zone, last_pfn);
}

static struct page *alloc_gigantic_page(int nid, unsigned order)
{
	unsigned long nr_pages = 1 << order;
	unsigned long ret, pfn, flags;
	struct zone *z;

	z = NODE_DATA(nid)->node_zones;
	for (; z - NODE_DATA(nid)->node_zones < MAX_NR_ZONES; z++) {
		spin_lock_irqsave(&z->lock, flags);

		pfn = ALIGN(z->zone_start_pfn, nr_pages);
		while (zone_spans_last_pfn(z, pfn, nr_pages)) {
			if (pfn_range_valid_gigantic(pfn, nr_pages)) {
				/*
				 * We release the zone lock here because
				 * alloc_contig_range() will also lock the zone
				 * at some point. If there's an allocation
				 * spinning on this lock, it may win the race
				 * and cause alloc_contig_range() to fail...
				 */
				spin_unlock_irqrestore(&z->lock, flags);
				ret = __alloc_gigantic_page(pfn, nr_pages);
				if (!ret)
					return pfn_to_page(pfn);
				spin_lock_irqsave(&z->lock, flags);
			}
			pfn += nr_pages;
		}

		spin_unlock_irqrestore(&z->lock, flags);
	}

	return NULL;
}

static void prep_new_huge_page(struct hstate *h, struct page *page, int nid);
static void prep_compound_gigantic_page(struct page *page, unsigned long order);

static struct page *alloc_fresh_gigantic_page_node(struct hstate *h, int nid)
{
	struct page *page;

	page = alloc_gigantic_page(nid, huge_page_order(h));
	if (page) {
		prep_compound_gigantic_page(page, huge_page_order(h));
		prep_new_huge_page(h, page, nid);
	}

	return page;
}

static int alloc_fresh_gigantic_page(struct hstate *h,
				nodemask_t *nodes_allowed)
{
	struct page *page = NULL;
	int nr_nodes, node;

	for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) {
		page = alloc_fresh_gigantic_page_node(h, node);
		if (page)
			return 1;
	}

	return 0;
}

static inline bool gigantic_page_supported(void) { return true; }
#else
static inline bool gigantic_page_supported(void) { return false; }
static inline void free_gigantic_page(struct page *page, unsigned order) { }
static inline void destroy_compound_gigantic_page(struct page *page,
						unsigned long order) { }
static inline int alloc_fresh_gigantic_page(struct hstate *h,
					nodemask_t *nodes_allowed) { return 0; }
#endif

820
static void update_and_free_page(struct hstate *h, struct page *page)
A
Adam Litke 已提交
821 822
{
	int i;
823

824 825
	if (hstate_is_gigantic(h) && !gigantic_page_supported())
		return;
826

827 828 829
	h->nr_huge_pages--;
	h->nr_huge_pages_node[page_to_nid(page)]--;
	for (i = 0; i < pages_per_huge_page(h); i++) {
830 831
		page[i].flags &= ~(1 << PG_locked | 1 << PG_error |
				1 << PG_referenced | 1 << PG_dirty |
832 833
				1 << PG_active | 1 << PG_private |
				1 << PG_writeback);
A
Adam Litke 已提交
834
	}
835
	VM_BUG_ON_PAGE(hugetlb_cgroup_from_page(page), page);
A
Adam Litke 已提交
836 837
	set_compound_page_dtor(page, NULL);
	set_page_refcounted(page);
838 839 840 841 842 843 844
	if (hstate_is_gigantic(h)) {
		destroy_compound_gigantic_page(page, huge_page_order(h));
		free_gigantic_page(page, huge_page_order(h));
	} else {
		arch_release_hugepage(page);
		__free_pages(page, huge_page_order(h));
	}
A
Adam Litke 已提交
845 846
}

847 848 849 850 851 852 853 854 855 856 857
struct hstate *size_to_hstate(unsigned long size)
{
	struct hstate *h;

	for_each_hstate(h) {
		if (huge_page_size(h) == size)
			return h;
	}
	return NULL;
}

858
void free_huge_page(struct page *page)
859
{
860 861 862 863
	/*
	 * Can't pass hstate in here because it is called from the
	 * compound page destructor.
	 */
864
	struct hstate *h = page_hstate(page);
865
	int nid = page_to_nid(page);
866 867
	struct hugepage_subpool *spool =
		(struct hugepage_subpool *)page_private(page);
868
	bool restore_reserve;
869

870
	set_page_private(page, 0);
871
	page->mapping = NULL;
872
	BUG_ON(page_count(page));
873
	BUG_ON(page_mapcount(page));
874
	restore_reserve = PagePrivate(page);
875
	ClearPagePrivate(page);
876 877

	spin_lock(&hugetlb_lock);
878 879
	hugetlb_cgroup_uncharge_page(hstate_index(h),
				     pages_per_huge_page(h), page);
880 881 882
	if (restore_reserve)
		h->resv_huge_pages++;

883
	if (h->surplus_huge_pages_node[nid]) {
884 885
		/* remove the page from active list */
		list_del(&page->lru);
886 887 888
		update_and_free_page(h, page);
		h->surplus_huge_pages--;
		h->surplus_huge_pages_node[nid]--;
889
	} else {
890
		arch_clear_hugepage_flags(page);
891
		enqueue_huge_page(h, page);
892
	}
893
	spin_unlock(&hugetlb_lock);
894
	hugepage_subpool_put_pages(spool, 1);
895 896
}

897
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid)
898
{
899
	INIT_LIST_HEAD(&page->lru);
900 901
	set_compound_page_dtor(page, free_huge_page);
	spin_lock(&hugetlb_lock);
902
	set_hugetlb_cgroup(page, NULL);
903 904
	h->nr_huge_pages++;
	h->nr_huge_pages_node[nid]++;
905 906 907 908
	spin_unlock(&hugetlb_lock);
	put_page(page); /* free it into the hugepage allocator */
}

909
static void prep_compound_gigantic_page(struct page *page, unsigned long order)
910 911 912 913 914 915 916 917
{
	int i;
	int nr_pages = 1 << order;
	struct page *p = page + 1;

	/* we rely on prep_new_huge_page to set the destructor */
	set_compound_order(page, order);
	__SetPageHead(page);
918
	__ClearPageReserved(page);
919 920
	for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) {
		__SetPageTail(p);
921 922 923 924 925 926 927 928 929 930 931 932 933
		/*
		 * For gigantic hugepages allocated through bootmem at
		 * boot, it's safer to be consistent with the not-gigantic
		 * hugepages and clear the PG_reserved bit from all tail pages
		 * too.  Otherwse drivers using get_user_pages() to access tail
		 * pages may get the reference counting wrong if they see
		 * PG_reserved set on a tail page (despite the head page not
		 * having PG_reserved set).  Enforcing this consistency between
		 * head and tail pages allows drivers to optimize away a check
		 * on the head page when they need know if put_page() is needed
		 * after get_user_pages().
		 */
		__ClearPageReserved(p);
934
		set_page_count(p, 0);
935 936 937 938
		p->first_page = page;
	}
}

A
Andrew Morton 已提交
939 940 941 942 943
/*
 * PageHuge() only returns true for hugetlbfs pages, but not for normal or
 * transparent huge pages.  See the PageTransHuge() documentation for more
 * details.
 */
944 945 946 947 948 949
int PageHuge(struct page *page)
{
	if (!PageCompound(page))
		return 0;

	page = compound_head(page);
950
	return get_compound_page_dtor(page) == free_huge_page;
951
}
952 953
EXPORT_SYMBOL_GPL(PageHuge);

954 955 956 957 958 959 960 961 962
/*
 * PageHeadHuge() only returns true for hugetlbfs head page, but not for
 * normal or transparent huge pages.
 */
int PageHeadHuge(struct page *page_head)
{
	if (!PageHead(page_head))
		return 0;

963
	return get_compound_page_dtor(page_head) == free_huge_page;
964 965
}

966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982
pgoff_t __basepage_index(struct page *page)
{
	struct page *page_head = compound_head(page);
	pgoff_t index = page_index(page_head);
	unsigned long compound_idx;

	if (!PageHuge(page_head))
		return page_index(page);

	if (compound_order(page_head) >= MAX_ORDER)
		compound_idx = page_to_pfn(page) - page_to_pfn(page_head);
	else
		compound_idx = page - page_head;

	return (index << compound_order(page_head)) + compound_idx;
}

983
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid)
L
Linus Torvalds 已提交
984 985
{
	struct page *page;
986

987
	page = alloc_pages_exact_node(nid,
988
		htlb_alloc_mask(h)|__GFP_COMP|__GFP_THISNODE|
989
						__GFP_REPEAT|__GFP_NOWARN,
990
		huge_page_order(h));
L
Linus Torvalds 已提交
991
	if (page) {
992
		if (arch_prepare_hugepage(page)) {
993
			__free_pages(page, huge_page_order(h));
994
			return NULL;
995
		}
996
		prep_new_huge_page(h, page, nid);
L
Linus Torvalds 已提交
997
	}
998 999 1000 1001

	return page;
}

1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023
static int alloc_fresh_huge_page(struct hstate *h, nodemask_t *nodes_allowed)
{
	struct page *page;
	int nr_nodes, node;
	int ret = 0;

	for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) {
		page = alloc_fresh_huge_page_node(h, node);
		if (page) {
			ret = 1;
			break;
		}
	}

	if (ret)
		count_vm_event(HTLB_BUDDY_PGALLOC);
	else
		count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);

	return ret;
}

1024 1025 1026 1027 1028 1029
/*
 * Free huge page from pool from next node to free.
 * Attempt to keep persistent huge pages more or less
 * balanced over allowed nodes.
 * Called with hugetlb_lock locked.
 */
1030 1031
static int free_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed,
							 bool acct_surplus)
1032
{
1033
	int nr_nodes, node;
1034 1035
	int ret = 0;

1036
	for_each_node_mask_to_free(h, nr_nodes, node, nodes_allowed) {
1037 1038 1039 1040
		/*
		 * If we're returning unused surplus pages, only examine
		 * nodes with surplus pages.
		 */
1041 1042
		if ((!acct_surplus || h->surplus_huge_pages_node[node]) &&
		    !list_empty(&h->hugepage_freelists[node])) {
1043
			struct page *page =
1044
				list_entry(h->hugepage_freelists[node].next,
1045 1046 1047
					  struct page, lru);
			list_del(&page->lru);
			h->free_huge_pages--;
1048
			h->free_huge_pages_node[node]--;
1049 1050
			if (acct_surplus) {
				h->surplus_huge_pages--;
1051
				h->surplus_huge_pages_node[node]--;
1052
			}
1053 1054
			update_and_free_page(h, page);
			ret = 1;
1055
			break;
1056
		}
1057
	}
1058 1059 1060 1061

	return ret;
}

1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090
/*
 * Dissolve a given free hugepage into free buddy pages. This function does
 * nothing for in-use (including surplus) hugepages.
 */
static void dissolve_free_huge_page(struct page *page)
{
	spin_lock(&hugetlb_lock);
	if (PageHuge(page) && !page_count(page)) {
		struct hstate *h = page_hstate(page);
		int nid = page_to_nid(page);
		list_del(&page->lru);
		h->free_huge_pages--;
		h->free_huge_pages_node[nid]--;
		update_and_free_page(h, page);
	}
	spin_unlock(&hugetlb_lock);
}

/*
 * Dissolve free hugepages in a given pfn range. Used by memory hotplug to
 * make specified memory blocks removable from the system.
 * Note that start_pfn should aligned with (minimum) hugepage size.
 */
void dissolve_free_huge_pages(unsigned long start_pfn, unsigned long end_pfn)
{
	unsigned int order = 8 * sizeof(void *);
	unsigned long pfn;
	struct hstate *h;

1091 1092 1093
	if (!hugepages_supported())
		return;

1094 1095 1096 1097 1098 1099 1100 1101 1102
	/* Set scan step to minimum hugepage size */
	for_each_hstate(h)
		if (order > huge_page_order(h))
			order = huge_page_order(h);
	VM_BUG_ON(!IS_ALIGNED(start_pfn, 1 << order));
	for (pfn = start_pfn; pfn < end_pfn; pfn += 1 << order)
		dissolve_free_huge_page(pfn_to_page(pfn));
}

1103
static struct page *alloc_buddy_huge_page(struct hstate *h, int nid)
1104 1105
{
	struct page *page;
1106
	unsigned int r_nid;
1107

1108
	if (hstate_is_gigantic(h))
1109 1110
		return NULL;

1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134
	/*
	 * Assume we will successfully allocate the surplus page to
	 * prevent racing processes from causing the surplus to exceed
	 * overcommit
	 *
	 * This however introduces a different race, where a process B
	 * tries to grow the static hugepage pool while alloc_pages() is
	 * called by process A. B will only examine the per-node
	 * counters in determining if surplus huge pages can be
	 * converted to normal huge pages in adjust_pool_surplus(). A
	 * won't be able to increment the per-node counter, until the
	 * lock is dropped by B, but B doesn't drop hugetlb_lock until
	 * no more huge pages can be converted from surplus to normal
	 * state (and doesn't try to convert again). Thus, we have a
	 * case where a surplus huge page exists, the pool is grown, and
	 * the surplus huge page still exists after, even though it
	 * should just have been converted to a normal huge page. This
	 * does not leak memory, though, as the hugepage will be freed
	 * once it is out of use. It also does not allow the counters to
	 * go out of whack in adjust_pool_surplus() as we don't modify
	 * the node values until we've gotten the hugepage and only the
	 * per-node value is checked there.
	 */
	spin_lock(&hugetlb_lock);
1135
	if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) {
1136 1137 1138
		spin_unlock(&hugetlb_lock);
		return NULL;
	} else {
1139 1140
		h->nr_huge_pages++;
		h->surplus_huge_pages++;
1141 1142 1143
	}
	spin_unlock(&hugetlb_lock);

1144
	if (nid == NUMA_NO_NODE)
1145
		page = alloc_pages(htlb_alloc_mask(h)|__GFP_COMP|
1146 1147 1148 1149
				   __GFP_REPEAT|__GFP_NOWARN,
				   huge_page_order(h));
	else
		page = alloc_pages_exact_node(nid,
1150
			htlb_alloc_mask(h)|__GFP_COMP|__GFP_THISNODE|
1151
			__GFP_REPEAT|__GFP_NOWARN, huge_page_order(h));
1152

1153 1154
	if (page && arch_prepare_hugepage(page)) {
		__free_pages(page, huge_page_order(h));
1155
		page = NULL;
1156 1157
	}

1158
	spin_lock(&hugetlb_lock);
1159
	if (page) {
1160
		INIT_LIST_HEAD(&page->lru);
1161
		r_nid = page_to_nid(page);
1162
		set_compound_page_dtor(page, free_huge_page);
1163
		set_hugetlb_cgroup(page, NULL);
1164 1165 1166
		/*
		 * We incremented the global counters already
		 */
1167 1168
		h->nr_huge_pages_node[r_nid]++;
		h->surplus_huge_pages_node[r_nid]++;
1169
		__count_vm_event(HTLB_BUDDY_PGALLOC);
1170
	} else {
1171 1172
		h->nr_huge_pages--;
		h->surplus_huge_pages--;
1173
		__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);
1174
	}
1175
	spin_unlock(&hugetlb_lock);
1176 1177 1178 1179

	return page;
}

1180 1181 1182 1183 1184 1185 1186
/*
 * This allocation function is useful in the context where vma is irrelevant.
 * E.g. soft-offlining uses this function because it only cares physical
 * address of error page.
 */
struct page *alloc_huge_page_node(struct hstate *h, int nid)
{
1187
	struct page *page = NULL;
1188 1189

	spin_lock(&hugetlb_lock);
1190 1191
	if (h->free_huge_pages - h->resv_huge_pages > 0)
		page = dequeue_huge_page_node(h, nid);
1192 1193
	spin_unlock(&hugetlb_lock);

1194
	if (!page)
1195 1196 1197 1198 1199
		page = alloc_buddy_huge_page(h, nid);

	return page;
}

1200
/*
L
Lucas De Marchi 已提交
1201
 * Increase the hugetlb pool such that it can accommodate a reservation
1202 1203
 * of size 'delta'.
 */
1204
static int gather_surplus_pages(struct hstate *h, int delta)
1205 1206 1207 1208 1209
{
	struct list_head surplus_list;
	struct page *page, *tmp;
	int ret, i;
	int needed, allocated;
1210
	bool alloc_ok = true;
1211

1212
	needed = (h->resv_huge_pages + delta) - h->free_huge_pages;
1213
	if (needed <= 0) {
1214
		h->resv_huge_pages += delta;
1215
		return 0;
1216
	}
1217 1218 1219 1220 1221 1222 1223 1224

	allocated = 0;
	INIT_LIST_HEAD(&surplus_list);

	ret = -ENOMEM;
retry:
	spin_unlock(&hugetlb_lock);
	for (i = 0; i < needed; i++) {
1225
		page = alloc_buddy_huge_page(h, NUMA_NO_NODE);
1226 1227 1228 1229
		if (!page) {
			alloc_ok = false;
			break;
		}
1230 1231
		list_add(&page->lru, &surplus_list);
	}
1232
	allocated += i;
1233 1234 1235 1236 1237 1238

	/*
	 * After retaking hugetlb_lock, we need to recalculate 'needed'
	 * because either resv_huge_pages or free_huge_pages may have changed.
	 */
	spin_lock(&hugetlb_lock);
1239 1240
	needed = (h->resv_huge_pages + delta) -
			(h->free_huge_pages + allocated);
1241 1242 1243 1244 1245 1246 1247 1248 1249 1250
	if (needed > 0) {
		if (alloc_ok)
			goto retry;
		/*
		 * We were not able to allocate enough pages to
		 * satisfy the entire reservation so we free what
		 * we've allocated so far.
		 */
		goto free;
	}
1251 1252
	/*
	 * The surplus_list now contains _at_least_ the number of extra pages
L
Lucas De Marchi 已提交
1253
	 * needed to accommodate the reservation.  Add the appropriate number
1254
	 * of pages to the hugetlb pool and free the extras back to the buddy
1255 1256 1257
	 * allocator.  Commit the entire reservation here to prevent another
	 * process from stealing the pages as they are added to the pool but
	 * before they are reserved.
1258 1259
	 */
	needed += allocated;
1260
	h->resv_huge_pages += delta;
1261
	ret = 0;
1262

1263
	/* Free the needed pages to the hugetlb pool */
1264
	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
1265 1266
		if ((--needed) < 0)
			break;
1267 1268 1269 1270 1271
		/*
		 * This page is now managed by the hugetlb allocator and has
		 * no users -- drop the buddy allocator's reference.
		 */
		put_page_testzero(page);
1272
		VM_BUG_ON_PAGE(page_count(page), page);
1273
		enqueue_huge_page(h, page);
1274
	}
1275
free:
1276
	spin_unlock(&hugetlb_lock);
1277 1278

	/* Free unnecessary surplus pages to the buddy allocator */
1279 1280
	list_for_each_entry_safe(page, tmp, &surplus_list, lru)
		put_page(page);
1281
	spin_lock(&hugetlb_lock);
1282 1283 1284 1285 1286 1287 1288 1289

	return ret;
}

/*
 * When releasing a hugetlb pool reservation, any surplus pages that were
 * allocated to satisfy the reservation must be explicitly freed if they were
 * never used.
1290
 * Called with hugetlb_lock held.
1291
 */
1292 1293
static void return_unused_surplus_pages(struct hstate *h,
					unsigned long unused_resv_pages)
1294 1295 1296
{
	unsigned long nr_pages;

1297
	/* Uncommit the reservation */
1298
	h->resv_huge_pages -= unused_resv_pages;
1299

1300
	/* Cannot return gigantic pages currently */
1301
	if (hstate_is_gigantic(h))
1302 1303
		return;

1304
	nr_pages = min(unused_resv_pages, h->surplus_huge_pages);
1305

1306 1307
	/*
	 * We want to release as many surplus pages as possible, spread
1308 1309 1310 1311 1312
	 * evenly across all nodes with memory. Iterate across these nodes
	 * until we can no longer free unreserved surplus pages. This occurs
	 * when the nodes with surplus pages have no free pages.
	 * free_pool_huge_page() will balance the the freed pages across the
	 * on-line nodes with memory and will handle the hstate accounting.
1313 1314
	 */
	while (nr_pages--) {
1315
		if (!free_pool_huge_page(h, &node_states[N_MEMORY], 1))
1316
			break;
1317
		cond_resched_lock(&hugetlb_lock);
1318 1319 1320
	}
}

1321 1322 1323
/*
 * Determine if the huge page at addr within the vma has an associated
 * reservation.  Where it does not we will need to logically increase
1324 1325 1326 1327 1328 1329
 * reservation and actually increase subpool usage before an allocation
 * can occur.  Where any new reservation would be required the
 * reservation change is prepared, but not committed.  Once the page
 * has been allocated from the subpool and instantiated the change should
 * be committed via vma_commit_reservation.  No action is required on
 * failure.
1330
 */
1331
static long vma_needs_reservation(struct hstate *h,
1332
			struct vm_area_struct *vma, unsigned long addr)
1333
{
1334 1335 1336
	struct resv_map *resv;
	pgoff_t idx;
	long chg;
1337

1338 1339
	resv = vma_resv_map(vma);
	if (!resv)
1340
		return 1;
1341

1342 1343
	idx = vma_hugecache_offset(h, vma, addr);
	chg = region_chg(resv, idx, idx + 1);
1344

1345 1346 1347 1348
	if (vma->vm_flags & VM_MAYSHARE)
		return chg;
	else
		return chg < 0 ? chg : 0;
1349
}
1350 1351
static void vma_commit_reservation(struct hstate *h,
			struct vm_area_struct *vma, unsigned long addr)
1352
{
1353 1354
	struct resv_map *resv;
	pgoff_t idx;
1355

1356 1357 1358
	resv = vma_resv_map(vma);
	if (!resv)
		return;
1359

1360 1361
	idx = vma_hugecache_offset(h, vma, addr);
	region_add(resv, idx, idx + 1);
1362 1363
}

1364
static struct page *alloc_huge_page(struct vm_area_struct *vma,
1365
				    unsigned long addr, int avoid_reserve)
L
Linus Torvalds 已提交
1366
{
1367
	struct hugepage_subpool *spool = subpool_vma(vma);
1368
	struct hstate *h = hstate_vma(vma);
1369
	struct page *page;
1370
	long chg;
1371 1372
	int ret, idx;
	struct hugetlb_cgroup *h_cg;
1373

1374
	idx = hstate_index(h);
1375
	/*
1376 1377 1378 1379 1380 1381
	 * Processes that did not create the mapping will have no
	 * reserves and will not have accounted against subpool
	 * limit. Check that the subpool limit can be made before
	 * satisfying the allocation MAP_NORESERVE mappings may also
	 * need pages and subpool limit allocated allocated if no reserve
	 * mapping overlaps.
1382
	 */
1383
	chg = vma_needs_reservation(h, vma, addr);
1384
	if (chg < 0)
1385
		return ERR_PTR(-ENOMEM);
1386 1387
	if (chg || avoid_reserve)
		if (hugepage_subpool_get_pages(spool, 1))
1388
			return ERR_PTR(-ENOSPC);
L
Linus Torvalds 已提交
1389

1390
	ret = hugetlb_cgroup_charge_cgroup(idx, pages_per_huge_page(h), &h_cg);
1391 1392 1393
	if (ret)
		goto out_subpool_put;

L
Linus Torvalds 已提交
1394
	spin_lock(&hugetlb_lock);
1395
	page = dequeue_huge_page_vma(h, vma, addr, avoid_reserve, chg);
1396
	if (!page) {
1397
		spin_unlock(&hugetlb_lock);
1398
		page = alloc_buddy_huge_page(h, NUMA_NO_NODE);
1399 1400 1401
		if (!page)
			goto out_uncharge_cgroup;

1402 1403
		spin_lock(&hugetlb_lock);
		list_move(&page->lru, &h->hugepage_activelist);
1404
		/* Fall through */
K
Ken Chen 已提交
1405
	}
1406 1407
	hugetlb_cgroup_commit_charge(idx, pages_per_huge_page(h), h_cg, page);
	spin_unlock(&hugetlb_lock);
1408

1409
	set_page_private(page, (unsigned long)spool);
1410

1411
	vma_commit_reservation(h, vma, addr);
1412
	return page;
1413 1414 1415 1416 1417 1418 1419

out_uncharge_cgroup:
	hugetlb_cgroup_uncharge_cgroup(idx, pages_per_huge_page(h), h_cg);
out_subpool_put:
	if (chg || avoid_reserve)
		hugepage_subpool_put_pages(spool, 1);
	return ERR_PTR(-ENOSPC);
1420 1421
}

1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435
/*
 * alloc_huge_page()'s wrapper which simply returns the page if allocation
 * succeeds, otherwise NULL. This function is called from new_vma_page(),
 * where no ERR_VALUE is expected to be returned.
 */
struct page *alloc_huge_page_noerr(struct vm_area_struct *vma,
				unsigned long addr, int avoid_reserve)
{
	struct page *page = alloc_huge_page(vma, addr, avoid_reserve);
	if (IS_ERR(page))
		page = NULL;
	return page;
}

1436
int __weak alloc_bootmem_huge_page(struct hstate *h)
1437 1438
{
	struct huge_bootmem_page *m;
1439
	int nr_nodes, node;
1440

1441
	for_each_node_mask_to_alloc(h, nr_nodes, node, &node_states[N_MEMORY]) {
1442 1443
		void *addr;

1444 1445 1446
		addr = memblock_virt_alloc_try_nid_nopanic(
				huge_page_size(h), huge_page_size(h),
				0, BOOTMEM_ALLOC_ACCESSIBLE, node);
1447 1448 1449 1450 1451 1452 1453
		if (addr) {
			/*
			 * Use the beginning of the huge page to store the
			 * huge_bootmem_page struct (until gather_bootmem
			 * puts them into the mem_map).
			 */
			m = addr;
1454
			goto found;
1455 1456 1457 1458 1459
		}
	}
	return 0;

found:
1460
	BUG_ON(!IS_ALIGNED(virt_to_phys(m), huge_page_size(h)));
1461 1462 1463 1464 1465 1466
	/* Put them into a private list first because mem_map is not up yet */
	list_add(&m->list, &huge_boot_pages);
	m->hstate = h;
	return 1;
}

1467
static void __init prep_compound_huge_page(struct page *page, int order)
1468 1469 1470 1471 1472 1473 1474
{
	if (unlikely(order > (MAX_ORDER - 1)))
		prep_compound_gigantic_page(page, order);
	else
		prep_compound_page(page, order);
}

1475 1476 1477 1478 1479 1480 1481
/* Put bootmem huge pages into the standard lists after mem_map is up */
static void __init gather_bootmem_prealloc(void)
{
	struct huge_bootmem_page *m;

	list_for_each_entry(m, &huge_boot_pages, list) {
		struct hstate *h = m->hstate;
1482 1483 1484 1485
		struct page *page;

#ifdef CONFIG_HIGHMEM
		page = pfn_to_page(m->phys >> PAGE_SHIFT);
1486 1487
		memblock_free_late(__pa(m),
				   sizeof(struct huge_bootmem_page));
1488 1489 1490
#else
		page = virt_to_page(m);
#endif
1491
		WARN_ON(page_count(page) != 1);
1492
		prep_compound_huge_page(page, h->order);
1493
		WARN_ON(PageReserved(page));
1494
		prep_new_huge_page(h, page, page_to_nid(page));
1495 1496 1497 1498 1499 1500
		/*
		 * If we had gigantic hugepages allocated at boot time, we need
		 * to restore the 'stolen' pages to totalram_pages in order to
		 * fix confusing memory reports from free(1) and another
		 * side-effects, like CommitLimit going negative.
		 */
1501
		if (hstate_is_gigantic(h))
1502
			adjust_managed_page_count(page, 1 << h->order);
1503 1504 1505
	}
}

1506
static void __init hugetlb_hstate_alloc_pages(struct hstate *h)
L
Linus Torvalds 已提交
1507 1508
{
	unsigned long i;
1509

1510
	for (i = 0; i < h->max_huge_pages; ++i) {
1511
		if (hstate_is_gigantic(h)) {
1512 1513
			if (!alloc_bootmem_huge_page(h))
				break;
1514
		} else if (!alloc_fresh_huge_page(h,
1515
					 &node_states[N_MEMORY]))
L
Linus Torvalds 已提交
1516 1517
			break;
	}
1518
	h->max_huge_pages = i;
1519 1520 1521 1522 1523 1524 1525
}

static void __init hugetlb_init_hstates(void)
{
	struct hstate *h;

	for_each_hstate(h) {
1526
		/* oversize hugepages were init'ed in early boot */
1527
		if (!hstate_is_gigantic(h))
1528
			hugetlb_hstate_alloc_pages(h);
1529 1530 1531
	}
}

A
Andi Kleen 已提交
1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542
static char * __init memfmt(char *buf, unsigned long n)
{
	if (n >= (1UL << 30))
		sprintf(buf, "%lu GB", n >> 30);
	else if (n >= (1UL << 20))
		sprintf(buf, "%lu MB", n >> 20);
	else
		sprintf(buf, "%lu KB", n >> 10);
	return buf;
}

1543 1544 1545 1546 1547
static void __init report_hugepages(void)
{
	struct hstate *h;

	for_each_hstate(h) {
A
Andi Kleen 已提交
1548
		char buf[32];
1549
		pr_info("HugeTLB registered %s page size, pre-allocated %ld pages\n",
A
Andi Kleen 已提交
1550 1551
			memfmt(buf, huge_page_size(h)),
			h->free_huge_pages);
1552 1553 1554
	}
}

L
Linus Torvalds 已提交
1555
#ifdef CONFIG_HIGHMEM
1556 1557
static void try_to_free_low(struct hstate *h, unsigned long count,
						nodemask_t *nodes_allowed)
L
Linus Torvalds 已提交
1558
{
1559 1560
	int i;

1561
	if (hstate_is_gigantic(h))
1562 1563
		return;

1564
	for_each_node_mask(i, *nodes_allowed) {
L
Linus Torvalds 已提交
1565
		struct page *page, *next;
1566 1567 1568
		struct list_head *freel = &h->hugepage_freelists[i];
		list_for_each_entry_safe(page, next, freel, lru) {
			if (count >= h->nr_huge_pages)
1569
				return;
L
Linus Torvalds 已提交
1570 1571 1572
			if (PageHighMem(page))
				continue;
			list_del(&page->lru);
1573
			update_and_free_page(h, page);
1574 1575
			h->free_huge_pages--;
			h->free_huge_pages_node[page_to_nid(page)]--;
L
Linus Torvalds 已提交
1576 1577 1578 1579
		}
	}
}
#else
1580 1581
static inline void try_to_free_low(struct hstate *h, unsigned long count,
						nodemask_t *nodes_allowed)
L
Linus Torvalds 已提交
1582 1583 1584 1585
{
}
#endif

1586 1587 1588 1589 1590
/*
 * Increment or decrement surplus_huge_pages.  Keep node-specific counters
 * balanced by operating on them in a round-robin fashion.
 * Returns 1 if an adjustment was made.
 */
1591 1592
static int adjust_pool_surplus(struct hstate *h, nodemask_t *nodes_allowed,
				int delta)
1593
{
1594
	int nr_nodes, node;
1595 1596 1597

	VM_BUG_ON(delta != -1 && delta != 1);

1598 1599 1600 1601
	if (delta < 0) {
		for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) {
			if (h->surplus_huge_pages_node[node])
				goto found;
1602
		}
1603 1604 1605 1606 1607
	} else {
		for_each_node_mask_to_free(h, nr_nodes, node, nodes_allowed) {
			if (h->surplus_huge_pages_node[node] <
					h->nr_huge_pages_node[node])
				goto found;
1608
		}
1609 1610
	}
	return 0;
1611

1612 1613 1614 1615
found:
	h->surplus_huge_pages += delta;
	h->surplus_huge_pages_node[node] += delta;
	return 1;
1616 1617
}

1618
#define persistent_huge_pages(h) (h->nr_huge_pages - h->surplus_huge_pages)
1619 1620
static unsigned long set_max_huge_pages(struct hstate *h, unsigned long count,
						nodemask_t *nodes_allowed)
L
Linus Torvalds 已提交
1621
{
1622
	unsigned long min_count, ret;
L
Linus Torvalds 已提交
1623

1624
	if (hstate_is_gigantic(h) && !gigantic_page_supported())
1625 1626
		return h->max_huge_pages;

1627 1628 1629 1630
	/*
	 * Increase the pool size
	 * First take pages out of surplus state.  Then make up the
	 * remaining difference by allocating fresh huge pages.
1631 1632 1633 1634 1635 1636
	 *
	 * We might race with alloc_buddy_huge_page() here and be unable
	 * to convert a surplus huge page to a normal huge page. That is
	 * not critical, though, it just means the overall size of the
	 * pool might be one hugepage larger than it needs to be, but
	 * within all the constraints specified by the sysctls.
1637
	 */
L
Linus Torvalds 已提交
1638
	spin_lock(&hugetlb_lock);
1639
	while (h->surplus_huge_pages && count > persistent_huge_pages(h)) {
1640
		if (!adjust_pool_surplus(h, nodes_allowed, -1))
1641 1642 1643
			break;
	}

1644
	while (count > persistent_huge_pages(h)) {
1645 1646 1647 1648 1649 1650
		/*
		 * If this allocation races such that we no longer need the
		 * page, free_huge_page will handle it by freeing the page
		 * and reducing the surplus.
		 */
		spin_unlock(&hugetlb_lock);
1651 1652 1653 1654
		if (hstate_is_gigantic(h))
			ret = alloc_fresh_gigantic_page(h, nodes_allowed);
		else
			ret = alloc_fresh_huge_page(h, nodes_allowed);
1655 1656 1657 1658
		spin_lock(&hugetlb_lock);
		if (!ret)
			goto out;

1659 1660 1661
		/* Bail for signals. Probably ctrl-c from user */
		if (signal_pending(current))
			goto out;
1662 1663 1664 1665 1666 1667 1668 1669
	}

	/*
	 * Decrease the pool size
	 * First return free pages to the buddy allocator (being careful
	 * to keep enough around to satisfy reservations).  Then place
	 * pages into surplus state as needed so the pool will shrink
	 * to the desired size as pages become free.
1670 1671 1672 1673 1674 1675 1676 1677
	 *
	 * By placing pages into the surplus state independent of the
	 * overcommit value, we are allowing the surplus pool size to
	 * exceed overcommit. There are few sane options here. Since
	 * alloc_buddy_huge_page() is checking the global counter,
	 * though, we'll note that we're not allowed to exceed surplus
	 * and won't grow the pool anywhere else. Not until one of the
	 * sysctls are changed, or the surplus pages go out of use.
1678
	 */
1679
	min_count = h->resv_huge_pages + h->nr_huge_pages - h->free_huge_pages;
1680
	min_count = max(count, min_count);
1681
	try_to_free_low(h, min_count, nodes_allowed);
1682
	while (min_count < persistent_huge_pages(h)) {
1683
		if (!free_pool_huge_page(h, nodes_allowed, 0))
L
Linus Torvalds 已提交
1684
			break;
1685
		cond_resched_lock(&hugetlb_lock);
L
Linus Torvalds 已提交
1686
	}
1687
	while (count < persistent_huge_pages(h)) {
1688
		if (!adjust_pool_surplus(h, nodes_allowed, 1))
1689 1690 1691
			break;
	}
out:
1692
	ret = persistent_huge_pages(h);
L
Linus Torvalds 已提交
1693
	spin_unlock(&hugetlb_lock);
1694
	return ret;
L
Linus Torvalds 已提交
1695 1696
}

1697 1698 1699 1700 1701 1702 1703 1704 1705 1706
#define HSTATE_ATTR_RO(_name) \
	static struct kobj_attribute _name##_attr = __ATTR_RO(_name)

#define HSTATE_ATTR(_name) \
	static struct kobj_attribute _name##_attr = \
		__ATTR(_name, 0644, _name##_show, _name##_store)

static struct kobject *hugepages_kobj;
static struct kobject *hstate_kobjs[HUGE_MAX_HSTATE];

1707 1708 1709
static struct hstate *kobj_to_node_hstate(struct kobject *kobj, int *nidp);

static struct hstate *kobj_to_hstate(struct kobject *kobj, int *nidp)
1710 1711
{
	int i;
1712

1713
	for (i = 0; i < HUGE_MAX_HSTATE; i++)
1714 1715 1716
		if (hstate_kobjs[i] == kobj) {
			if (nidp)
				*nidp = NUMA_NO_NODE;
1717
			return &hstates[i];
1718 1719 1720
		}

	return kobj_to_node_hstate(kobj, nidp);
1721 1722
}

1723
static ssize_t nr_hugepages_show_common(struct kobject *kobj,
1724 1725
					struct kobj_attribute *attr, char *buf)
{
1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736
	struct hstate *h;
	unsigned long nr_huge_pages;
	int nid;

	h = kobj_to_hstate(kobj, &nid);
	if (nid == NUMA_NO_NODE)
		nr_huge_pages = h->nr_huge_pages;
	else
		nr_huge_pages = h->nr_huge_pages_node[nid];

	return sprintf(buf, "%lu\n", nr_huge_pages);
1737
}
1738

1739 1740 1741
static ssize_t __nr_hugepages_store_common(bool obey_mempolicy,
					   struct hstate *h, int nid,
					   unsigned long count, size_t len)
1742 1743
{
	int err;
1744
	NODEMASK_ALLOC(nodemask_t, nodes_allowed, GFP_KERNEL | __GFP_NORETRY);
1745

1746
	if (hstate_is_gigantic(h) && !gigantic_page_supported()) {
1747 1748 1749 1750
		err = -EINVAL;
		goto out;
	}

1751 1752 1753 1754 1755 1756 1757
	if (nid == NUMA_NO_NODE) {
		/*
		 * global hstate attribute
		 */
		if (!(obey_mempolicy &&
				init_nodemask_of_mempolicy(nodes_allowed))) {
			NODEMASK_FREE(nodes_allowed);
1758
			nodes_allowed = &node_states[N_MEMORY];
1759 1760 1761 1762 1763 1764 1765 1766 1767
		}
	} else if (nodes_allowed) {
		/*
		 * per node hstate attribute: adjust count to global,
		 * but restrict alloc/free to the specified node.
		 */
		count += h->nr_huge_pages - h->nr_huge_pages_node[nid];
		init_nodemask_of_node(nodes_allowed, nid);
	} else
1768
		nodes_allowed = &node_states[N_MEMORY];
1769

1770
	h->max_huge_pages = set_max_huge_pages(h, count, nodes_allowed);
1771

1772
	if (nodes_allowed != &node_states[N_MEMORY])
1773 1774 1775
		NODEMASK_FREE(nodes_allowed);

	return len;
1776 1777 1778
out:
	NODEMASK_FREE(nodes_allowed);
	return err;
1779 1780
}

1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797
static ssize_t nr_hugepages_store_common(bool obey_mempolicy,
					 struct kobject *kobj, const char *buf,
					 size_t len)
{
	struct hstate *h;
	unsigned long count;
	int nid;
	int err;

	err = kstrtoul(buf, 10, &count);
	if (err)
		return err;

	h = kobj_to_hstate(kobj, &nid);
	return __nr_hugepages_store_common(obey_mempolicy, h, nid, count, len);
}

1798 1799 1800 1801 1802 1803 1804 1805 1806
static ssize_t nr_hugepages_show(struct kobject *kobj,
				       struct kobj_attribute *attr, char *buf)
{
	return nr_hugepages_show_common(kobj, attr, buf);
}

static ssize_t nr_hugepages_store(struct kobject *kobj,
	       struct kobj_attribute *attr, const char *buf, size_t len)
{
1807
	return nr_hugepages_store_common(false, kobj, buf, len);
1808 1809 1810
}
HSTATE_ATTR(nr_hugepages);

1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825
#ifdef CONFIG_NUMA

/*
 * hstate attribute for optionally mempolicy-based constraint on persistent
 * huge page alloc/free.
 */
static ssize_t nr_hugepages_mempolicy_show(struct kobject *kobj,
				       struct kobj_attribute *attr, char *buf)
{
	return nr_hugepages_show_common(kobj, attr, buf);
}

static ssize_t nr_hugepages_mempolicy_store(struct kobject *kobj,
	       struct kobj_attribute *attr, const char *buf, size_t len)
{
1826
	return nr_hugepages_store_common(true, kobj, buf, len);
1827 1828 1829 1830 1831
}
HSTATE_ATTR(nr_hugepages_mempolicy);
#endif


1832 1833 1834
static ssize_t nr_overcommit_hugepages_show(struct kobject *kobj,
					struct kobj_attribute *attr, char *buf)
{
1835
	struct hstate *h = kobj_to_hstate(kobj, NULL);
1836 1837
	return sprintf(buf, "%lu\n", h->nr_overcommit_huge_pages);
}
1838

1839 1840 1841 1842 1843
static ssize_t nr_overcommit_hugepages_store(struct kobject *kobj,
		struct kobj_attribute *attr, const char *buf, size_t count)
{
	int err;
	unsigned long input;
1844
	struct hstate *h = kobj_to_hstate(kobj, NULL);
1845

1846
	if (hstate_is_gigantic(h))
1847 1848
		return -EINVAL;

1849
	err = kstrtoul(buf, 10, &input);
1850
	if (err)
1851
		return err;
1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863

	spin_lock(&hugetlb_lock);
	h->nr_overcommit_huge_pages = input;
	spin_unlock(&hugetlb_lock);

	return count;
}
HSTATE_ATTR(nr_overcommit_hugepages);

static ssize_t free_hugepages_show(struct kobject *kobj,
					struct kobj_attribute *attr, char *buf)
{
1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874
	struct hstate *h;
	unsigned long free_huge_pages;
	int nid;

	h = kobj_to_hstate(kobj, &nid);
	if (nid == NUMA_NO_NODE)
		free_huge_pages = h->free_huge_pages;
	else
		free_huge_pages = h->free_huge_pages_node[nid];

	return sprintf(buf, "%lu\n", free_huge_pages);
1875 1876 1877 1878 1879 1880
}
HSTATE_ATTR_RO(free_hugepages);

static ssize_t resv_hugepages_show(struct kobject *kobj,
					struct kobj_attribute *attr, char *buf)
{
1881
	struct hstate *h = kobj_to_hstate(kobj, NULL);
1882 1883 1884 1885 1886 1887 1888
	return sprintf(buf, "%lu\n", h->resv_huge_pages);
}
HSTATE_ATTR_RO(resv_hugepages);

static ssize_t surplus_hugepages_show(struct kobject *kobj,
					struct kobj_attribute *attr, char *buf)
{
1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899
	struct hstate *h;
	unsigned long surplus_huge_pages;
	int nid;

	h = kobj_to_hstate(kobj, &nid);
	if (nid == NUMA_NO_NODE)
		surplus_huge_pages = h->surplus_huge_pages;
	else
		surplus_huge_pages = h->surplus_huge_pages_node[nid];

	return sprintf(buf, "%lu\n", surplus_huge_pages);
1900 1901 1902 1903 1904 1905 1906 1907 1908
}
HSTATE_ATTR_RO(surplus_hugepages);

static struct attribute *hstate_attrs[] = {
	&nr_hugepages_attr.attr,
	&nr_overcommit_hugepages_attr.attr,
	&free_hugepages_attr.attr,
	&resv_hugepages_attr.attr,
	&surplus_hugepages_attr.attr,
1909 1910 1911
#ifdef CONFIG_NUMA
	&nr_hugepages_mempolicy_attr.attr,
#endif
1912 1913 1914 1915 1916 1917 1918
	NULL,
};

static struct attribute_group hstate_attr_group = {
	.attrs = hstate_attrs,
};

J
Jeff Mahoney 已提交
1919 1920 1921
static int hugetlb_sysfs_add_hstate(struct hstate *h, struct kobject *parent,
				    struct kobject **hstate_kobjs,
				    struct attribute_group *hstate_attr_group)
1922 1923
{
	int retval;
1924
	int hi = hstate_index(h);
1925

1926 1927
	hstate_kobjs[hi] = kobject_create_and_add(h->name, parent);
	if (!hstate_kobjs[hi])
1928 1929
		return -ENOMEM;

1930
	retval = sysfs_create_group(hstate_kobjs[hi], hstate_attr_group);
1931
	if (retval)
1932
		kobject_put(hstate_kobjs[hi]);
1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946

	return retval;
}

static void __init hugetlb_sysfs_init(void)
{
	struct hstate *h;
	int err;

	hugepages_kobj = kobject_create_and_add("hugepages", mm_kobj);
	if (!hugepages_kobj)
		return;

	for_each_hstate(h) {
1947 1948
		err = hugetlb_sysfs_add_hstate(h, hugepages_kobj,
					 hstate_kobjs, &hstate_attr_group);
1949
		if (err)
1950
			pr_err("Hugetlb: Unable to add hstate %s", h->name);
1951 1952 1953
	}
}

1954 1955 1956 1957
#ifdef CONFIG_NUMA

/*
 * node_hstate/s - associate per node hstate attributes, via their kobjects,
1958 1959 1960
 * with node devices in node_devices[] using a parallel array.  The array
 * index of a node device or _hstate == node id.
 * This is here to avoid any static dependency of the node device driver, in
1961 1962 1963 1964 1965 1966 1967 1968 1969
 * the base kernel, on the hugetlb module.
 */
struct node_hstate {
	struct kobject		*hugepages_kobj;
	struct kobject		*hstate_kobjs[HUGE_MAX_HSTATE];
};
struct node_hstate node_hstates[MAX_NUMNODES];

/*
1970
 * A subset of global hstate attributes for node devices
1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983
 */
static struct attribute *per_node_hstate_attrs[] = {
	&nr_hugepages_attr.attr,
	&free_hugepages_attr.attr,
	&surplus_hugepages_attr.attr,
	NULL,
};

static struct attribute_group per_node_hstate_attr_group = {
	.attrs = per_node_hstate_attrs,
};

/*
1984
 * kobj_to_node_hstate - lookup global hstate for node device hstate attr kobj.
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
 * Returns node id via non-NULL nidp.
 */
static struct hstate *kobj_to_node_hstate(struct kobject *kobj, int *nidp)
{
	int nid;

	for (nid = 0; nid < nr_node_ids; nid++) {
		struct node_hstate *nhs = &node_hstates[nid];
		int i;
		for (i = 0; i < HUGE_MAX_HSTATE; i++)
			if (nhs->hstate_kobjs[i] == kobj) {
				if (nidp)
					*nidp = nid;
				return &hstates[i];
			}
	}

	BUG();
	return NULL;
}

/*
2007
 * Unregister hstate attributes from a single node device.
2008 2009
 * No-op if no hstate attributes attached.
 */
2010
static void hugetlb_unregister_node(struct node *node)
2011 2012
{
	struct hstate *h;
2013
	struct node_hstate *nhs = &node_hstates[node->dev.id];
2014 2015

	if (!nhs->hugepages_kobj)
2016
		return;		/* no hstate attributes */
2017

2018 2019 2020 2021 2022
	for_each_hstate(h) {
		int idx = hstate_index(h);
		if (nhs->hstate_kobjs[idx]) {
			kobject_put(nhs->hstate_kobjs[idx]);
			nhs->hstate_kobjs[idx] = NULL;
2023
		}
2024
	}
2025 2026 2027 2028 2029 2030

	kobject_put(nhs->hugepages_kobj);
	nhs->hugepages_kobj = NULL;
}

/*
2031
 * hugetlb module exit:  unregister hstate attributes from node devices
2032 2033 2034 2035 2036 2037 2038
 * that have them.
 */
static void hugetlb_unregister_all_nodes(void)
{
	int nid;

	/*
2039
	 * disable node device registrations.
2040 2041 2042 2043 2044 2045 2046
	 */
	register_hugetlbfs_with_node(NULL, NULL);

	/*
	 * remove hstate attributes from any nodes that have them.
	 */
	for (nid = 0; nid < nr_node_ids; nid++)
2047
		hugetlb_unregister_node(node_devices[nid]);
2048 2049 2050
}

/*
2051
 * Register hstate attributes for a single node device.
2052 2053
 * No-op if attributes already registered.
 */
2054
static void hugetlb_register_node(struct node *node)
2055 2056
{
	struct hstate *h;
2057
	struct node_hstate *nhs = &node_hstates[node->dev.id];
2058 2059 2060 2061 2062 2063
	int err;

	if (nhs->hugepages_kobj)
		return;		/* already allocated */

	nhs->hugepages_kobj = kobject_create_and_add("hugepages",
2064
							&node->dev.kobj);
2065 2066 2067 2068 2069 2070 2071 2072
	if (!nhs->hugepages_kobj)
		return;

	for_each_hstate(h) {
		err = hugetlb_sysfs_add_hstate(h, nhs->hugepages_kobj,
						nhs->hstate_kobjs,
						&per_node_hstate_attr_group);
		if (err) {
2073 2074
			pr_err("Hugetlb: Unable to add hstate %s for node %d\n",
				h->name, node->dev.id);
2075 2076 2077 2078 2079 2080 2081
			hugetlb_unregister_node(node);
			break;
		}
	}
}

/*
2082
 * hugetlb init time:  register hstate attributes for all registered node
2083 2084
 * devices of nodes that have memory.  All on-line nodes should have
 * registered their associated device by this time.
2085
 */
2086
static void __init hugetlb_register_all_nodes(void)
2087 2088 2089
{
	int nid;

2090
	for_each_node_state(nid, N_MEMORY) {
2091
		struct node *node = node_devices[nid];
2092
		if (node->dev.id == nid)
2093 2094 2095 2096
			hugetlb_register_node(node);
	}

	/*
2097
	 * Let the node device driver know we're here so it can
2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118
	 * [un]register hstate attributes on node hotplug.
	 */
	register_hugetlbfs_with_node(hugetlb_register_node,
				     hugetlb_unregister_node);
}
#else	/* !CONFIG_NUMA */

static struct hstate *kobj_to_node_hstate(struct kobject *kobj, int *nidp)
{
	BUG();
	if (nidp)
		*nidp = -1;
	return NULL;
}

static void hugetlb_unregister_all_nodes(void) { }

static void hugetlb_register_all_nodes(void) { }

#endif

2119 2120 2121 2122
static void __exit hugetlb_exit(void)
{
	struct hstate *h;

2123 2124
	hugetlb_unregister_all_nodes();

2125
	for_each_hstate(h) {
2126
		kobject_put(hstate_kobjs[hstate_index(h)]);
2127 2128 2129
	}

	kobject_put(hugepages_kobj);
2130
	kfree(htlb_fault_mutex_table);
2131 2132 2133 2134 2135
}
module_exit(hugetlb_exit);

static int __init hugetlb_init(void)
{
2136 2137
	int i;

2138
	if (!hugepages_supported())
2139
		return 0;
2140

2141 2142 2143 2144
	if (!size_to_hstate(default_hstate_size)) {
		default_hstate_size = HPAGE_SIZE;
		if (!size_to_hstate(default_hstate_size))
			hugetlb_add_hstate(HUGETLB_PAGE_ORDER);
2145
	}
2146
	default_hstate_idx = hstate_index(size_to_hstate(default_hstate_size));
2147 2148
	if (default_hstate_max_huge_pages)
		default_hstate.max_huge_pages = default_hstate_max_huge_pages;
2149 2150

	hugetlb_init_hstates();
2151
	gather_bootmem_prealloc();
2152 2153 2154
	report_hugepages();

	hugetlb_sysfs_init();
2155
	hugetlb_register_all_nodes();
2156
	hugetlb_cgroup_file_init();
2157

2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168
#ifdef CONFIG_SMP
	num_fault_mutexes = roundup_pow_of_two(8 * num_possible_cpus());
#else
	num_fault_mutexes = 1;
#endif
	htlb_fault_mutex_table =
		kmalloc(sizeof(struct mutex) * num_fault_mutexes, GFP_KERNEL);
	BUG_ON(!htlb_fault_mutex_table);

	for (i = 0; i < num_fault_mutexes; i++)
		mutex_init(&htlb_fault_mutex_table[i]);
2169 2170 2171 2172 2173 2174 2175 2176
	return 0;
}
module_init(hugetlb_init);

/* Should be called on processing a hugepagesz=... option */
void __init hugetlb_add_hstate(unsigned order)
{
	struct hstate *h;
2177 2178
	unsigned long i;

2179
	if (size_to_hstate(PAGE_SIZE << order)) {
2180
		pr_warning("hugepagesz= specified twice, ignoring\n");
2181 2182
		return;
	}
2183
	BUG_ON(hugetlb_max_hstate >= HUGE_MAX_HSTATE);
2184
	BUG_ON(order == 0);
2185
	h = &hstates[hugetlb_max_hstate++];
2186 2187
	h->order = order;
	h->mask = ~((1ULL << (order + PAGE_SHIFT)) - 1);
2188 2189 2190 2191
	h->nr_huge_pages = 0;
	h->free_huge_pages = 0;
	for (i = 0; i < MAX_NUMNODES; ++i)
		INIT_LIST_HEAD(&h->hugepage_freelists[i]);
2192
	INIT_LIST_HEAD(&h->hugepage_activelist);
2193 2194
	h->next_nid_to_alloc = first_node(node_states[N_MEMORY]);
	h->next_nid_to_free = first_node(node_states[N_MEMORY]);
2195 2196
	snprintf(h->name, HSTATE_NAME_LEN, "hugepages-%lukB",
					huge_page_size(h)/1024);
2197

2198 2199 2200
	parsed_hstate = h;
}

2201
static int __init hugetlb_nrpages_setup(char *s)
2202 2203
{
	unsigned long *mhp;
2204
	static unsigned long *last_mhp;
2205 2206

	/*
2207
	 * !hugetlb_max_hstate means we haven't parsed a hugepagesz= parameter yet,
2208 2209
	 * so this hugepages= parameter goes to the "default hstate".
	 */
2210
	if (!hugetlb_max_hstate)
2211 2212 2213 2214
		mhp = &default_hstate_max_huge_pages;
	else
		mhp = &parsed_hstate->max_huge_pages;

2215
	if (mhp == last_mhp) {
2216 2217
		pr_warning("hugepages= specified twice without "
			   "interleaving hugepagesz=, ignoring\n");
2218 2219 2220
		return 1;
	}

2221 2222 2223
	if (sscanf(s, "%lu", mhp) <= 0)
		*mhp = 0;

2224 2225 2226 2227 2228
	/*
	 * Global state is always initialized later in hugetlb_init.
	 * But we need to allocate >= MAX_ORDER hstates here early to still
	 * use the bootmem allocator.
	 */
2229
	if (hugetlb_max_hstate && parsed_hstate->order >= MAX_ORDER)
2230 2231 2232 2233
		hugetlb_hstate_alloc_pages(parsed_hstate);

	last_mhp = mhp;

2234 2235
	return 1;
}
2236 2237 2238 2239 2240 2241 2242 2243
__setup("hugepages=", hugetlb_nrpages_setup);

static int __init hugetlb_default_setup(char *s)
{
	default_hstate_size = memparse(s, &s);
	return 1;
}
__setup("default_hugepagesz=", hugetlb_default_setup);
2244

2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256
static unsigned int cpuset_mems_nr(unsigned int *array)
{
	int node;
	unsigned int nr = 0;

	for_each_node_mask(node, cpuset_current_mems_allowed)
		nr += array[node];

	return nr;
}

#ifdef CONFIG_SYSCTL
2257 2258 2259
static int hugetlb_sysctl_handler_common(bool obey_mempolicy,
			 struct ctl_table *table, int write,
			 void __user *buffer, size_t *length, loff_t *ppos)
L
Linus Torvalds 已提交
2260
{
2261
	struct hstate *h = &default_hstate;
2262
	unsigned long tmp = h->max_huge_pages;
2263
	int ret;
2264

2265 2266 2267
	if (!hugepages_supported())
		return -ENOTSUPP;

2268 2269
	table->data = &tmp;
	table->maxlen = sizeof(unsigned long);
2270 2271 2272
	ret = proc_doulongvec_minmax(table, write, buffer, length, ppos);
	if (ret)
		goto out;
2273

2274 2275 2276
	if (write)
		ret = __nr_hugepages_store_common(obey_mempolicy, h,
						  NUMA_NO_NODE, tmp, *length);
2277 2278
out:
	return ret;
L
Linus Torvalds 已提交
2279
}
2280

2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297
int hugetlb_sysctl_handler(struct ctl_table *table, int write,
			  void __user *buffer, size_t *length, loff_t *ppos)
{

	return hugetlb_sysctl_handler_common(false, table, write,
							buffer, length, ppos);
}

#ifdef CONFIG_NUMA
int hugetlb_mempolicy_sysctl_handler(struct ctl_table *table, int write,
			  void __user *buffer, size_t *length, loff_t *ppos)
{
	return hugetlb_sysctl_handler_common(true, table, write,
							buffer, length, ppos);
}
#endif /* CONFIG_NUMA */

2298
int hugetlb_overcommit_handler(struct ctl_table *table, int write,
2299
			void __user *buffer,
2300 2301
			size_t *length, loff_t *ppos)
{
2302
	struct hstate *h = &default_hstate;
2303
	unsigned long tmp;
2304
	int ret;
2305

2306 2307 2308
	if (!hugepages_supported())
		return -ENOTSUPP;

2309
	tmp = h->nr_overcommit_huge_pages;
2310

2311
	if (write && hstate_is_gigantic(h))
2312 2313
		return -EINVAL;

2314 2315
	table->data = &tmp;
	table->maxlen = sizeof(unsigned long);
2316 2317 2318
	ret = proc_doulongvec_minmax(table, write, buffer, length, ppos);
	if (ret)
		goto out;
2319 2320 2321 2322 2323 2324

	if (write) {
		spin_lock(&hugetlb_lock);
		h->nr_overcommit_huge_pages = tmp;
		spin_unlock(&hugetlb_lock);
	}
2325 2326
out:
	return ret;
2327 2328
}

L
Linus Torvalds 已提交
2329 2330
#endif /* CONFIG_SYSCTL */

2331
void hugetlb_report_meminfo(struct seq_file *m)
L
Linus Torvalds 已提交
2332
{
2333
	struct hstate *h = &default_hstate;
2334 2335
	if (!hugepages_supported())
		return;
2336
	seq_printf(m,
2337 2338 2339 2340 2341
			"HugePages_Total:   %5lu\n"
			"HugePages_Free:    %5lu\n"
			"HugePages_Rsvd:    %5lu\n"
			"HugePages_Surp:    %5lu\n"
			"Hugepagesize:   %8lu kB\n",
2342 2343 2344 2345 2346
			h->nr_huge_pages,
			h->free_huge_pages,
			h->resv_huge_pages,
			h->surplus_huge_pages,
			1UL << (huge_page_order(h) + PAGE_SHIFT - 10));
L
Linus Torvalds 已提交
2347 2348 2349 2350
}

int hugetlb_report_node_meminfo(int nid, char *buf)
{
2351
	struct hstate *h = &default_hstate;
2352 2353
	if (!hugepages_supported())
		return 0;
L
Linus Torvalds 已提交
2354 2355
	return sprintf(buf,
		"Node %d HugePages_Total: %5u\n"
2356 2357
		"Node %d HugePages_Free:  %5u\n"
		"Node %d HugePages_Surp:  %5u\n",
2358 2359 2360
		nid, h->nr_huge_pages_node[nid],
		nid, h->free_huge_pages_node[nid],
		nid, h->surplus_huge_pages_node[nid]);
L
Linus Torvalds 已提交
2361 2362
}

2363 2364 2365 2366 2367
void hugetlb_show_meminfo(void)
{
	struct hstate *h;
	int nid;

2368 2369 2370
	if (!hugepages_supported())
		return;

2371 2372 2373 2374 2375 2376 2377 2378 2379 2380
	for_each_node_state(nid, N_MEMORY)
		for_each_hstate(h)
			pr_info("Node %d hugepages_total=%u hugepages_free=%u hugepages_surp=%u hugepages_size=%lukB\n",
				nid,
				h->nr_huge_pages_node[nid],
				h->free_huge_pages_node[nid],
				h->surplus_huge_pages_node[nid],
				1UL << (huge_page_order(h) + PAGE_SHIFT - 10));
}

L
Linus Torvalds 已提交
2381 2382 2383
/* Return the number pages of memory we physically have, in PAGE_SIZE units. */
unsigned long hugetlb_total_pages(void)
{
2384 2385 2386 2387 2388 2389
	struct hstate *h;
	unsigned long nr_total_pages = 0;

	for_each_hstate(h)
		nr_total_pages += h->nr_huge_pages * pages_per_huge_page(h);
	return nr_total_pages;
L
Linus Torvalds 已提交
2390 2391
}

2392
static int hugetlb_acct_memory(struct hstate *h, long delta)
M
Mel Gorman 已提交
2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414
{
	int ret = -ENOMEM;

	spin_lock(&hugetlb_lock);
	/*
	 * When cpuset is configured, it breaks the strict hugetlb page
	 * reservation as the accounting is done on a global variable. Such
	 * reservation is completely rubbish in the presence of cpuset because
	 * the reservation is not checked against page availability for the
	 * current cpuset. Application can still potentially OOM'ed by kernel
	 * with lack of free htlb page in cpuset that the task is in.
	 * Attempt to enforce strict accounting with cpuset is almost
	 * impossible (or too ugly) because cpuset is too fluid that
	 * task or memory node can be dynamically moved between cpusets.
	 *
	 * The change of semantics for shared hugetlb mapping with cpuset is
	 * undesirable. However, in order to preserve some of the semantics,
	 * we fall back to check against current free page availability as
	 * a best attempt and hopefully to minimize the impact of changing
	 * semantics that cpuset has.
	 */
	if (delta > 0) {
2415
		if (gather_surplus_pages(h, delta) < 0)
M
Mel Gorman 已提交
2416 2417
			goto out;

2418 2419
		if (delta > cpuset_mems_nr(h->free_huge_pages_node)) {
			return_unused_surplus_pages(h, delta);
M
Mel Gorman 已提交
2420 2421 2422 2423 2424 2425
			goto out;
		}
	}

	ret = 0;
	if (delta < 0)
2426
		return_unused_surplus_pages(h, (unsigned long) -delta);
M
Mel Gorman 已提交
2427 2428 2429 2430 2431 2432

out:
	spin_unlock(&hugetlb_lock);
	return ret;
}

2433 2434
static void hugetlb_vm_op_open(struct vm_area_struct *vma)
{
2435
	struct resv_map *resv = vma_resv_map(vma);
2436 2437 2438 2439 2440

	/*
	 * This new VMA should share its siblings reservation map if present.
	 * The VMA will only ever have a valid reservation map pointer where
	 * it is being copied for another still existing VMA.  As that VMA
L
Lucas De Marchi 已提交
2441
	 * has a reference to the reservation map it cannot disappear until
2442 2443 2444
	 * after this open call completes.  It is therefore safe to take a
	 * new reference here without additional locking.
	 */
2445
	if (resv && is_vma_resv_set(vma, HPAGE_RESV_OWNER))
2446
		kref_get(&resv->refs);
2447 2448
}

2449 2450
static void hugetlb_vm_op_close(struct vm_area_struct *vma)
{
2451
	struct hstate *h = hstate_vma(vma);
2452
	struct resv_map *resv = vma_resv_map(vma);
2453
	struct hugepage_subpool *spool = subpool_vma(vma);
2454
	unsigned long reserve, start, end;
2455

2456 2457
	if (!resv || !is_vma_resv_set(vma, HPAGE_RESV_OWNER))
		return;
2458

2459 2460
	start = vma_hugecache_offset(h, vma, vma->vm_start);
	end = vma_hugecache_offset(h, vma, vma->vm_end);
2461

2462
	reserve = (end - start) - region_count(resv, start, end);
2463

2464 2465 2466 2467 2468
	kref_put(&resv->refs, resv_map_release);

	if (reserve) {
		hugetlb_acct_memory(h, -reserve);
		hugepage_subpool_put_pages(spool, reserve);
2469
	}
2470 2471
}

L
Linus Torvalds 已提交
2472 2473 2474 2475 2476 2477
/*
 * We cannot handle pagefaults against hugetlb pages at all.  They cause
 * handle_mm_fault() to try to instantiate regular-sized pages in the
 * hugegpage VMA.  do_page_fault() is supposed to trap this, so BUG is we get
 * this far.
 */
N
Nick Piggin 已提交
2478
static int hugetlb_vm_op_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
L
Linus Torvalds 已提交
2479 2480
{
	BUG();
N
Nick Piggin 已提交
2481
	return 0;
L
Linus Torvalds 已提交
2482 2483
}

2484
const struct vm_operations_struct hugetlb_vm_ops = {
N
Nick Piggin 已提交
2485
	.fault = hugetlb_vm_op_fault,
2486
	.open = hugetlb_vm_op_open,
2487
	.close = hugetlb_vm_op_close,
L
Linus Torvalds 已提交
2488 2489
};

2490 2491
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page,
				int writable)
D
David Gibson 已提交
2492 2493 2494
{
	pte_t entry;

2495
	if (writable) {
2496 2497
		entry = huge_pte_mkwrite(huge_pte_mkdirty(mk_huge_pte(page,
					 vma->vm_page_prot)));
D
David Gibson 已提交
2498
	} else {
2499 2500
		entry = huge_pte_wrprotect(mk_huge_pte(page,
					   vma->vm_page_prot));
D
David Gibson 已提交
2501 2502 2503
	}
	entry = pte_mkyoung(entry);
	entry = pte_mkhuge(entry);
2504
	entry = arch_make_huge_pte(entry, vma, page, writable);
D
David Gibson 已提交
2505 2506 2507 2508

	return entry;
}

2509 2510 2511 2512 2513
static void set_huge_ptep_writable(struct vm_area_struct *vma,
				   unsigned long address, pte_t *ptep)
{
	pte_t entry;

2514
	entry = huge_pte_mkwrite(huge_pte_mkdirty(huge_ptep_get(ptep)));
2515
	if (huge_ptep_set_access_flags(vma, address, ptep, entry, 1))
2516
		update_mmu_cache(vma, address, ptep);
2517 2518
}

2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543
static int is_hugetlb_entry_migration(pte_t pte)
{
	swp_entry_t swp;

	if (huge_pte_none(pte) || pte_present(pte))
		return 0;
	swp = pte_to_swp_entry(pte);
	if (non_swap_entry(swp) && is_migration_entry(swp))
		return 1;
	else
		return 0;
}

static int is_hugetlb_entry_hwpoisoned(pte_t pte)
{
	swp_entry_t swp;

	if (huge_pte_none(pte) || pte_present(pte))
		return 0;
	swp = pte_to_swp_entry(pte);
	if (non_swap_entry(swp) && is_hwpoison_entry(swp))
		return 1;
	else
		return 0;
}
2544

D
David Gibson 已提交
2545 2546 2547 2548 2549
int copy_hugetlb_page_range(struct mm_struct *dst, struct mm_struct *src,
			    struct vm_area_struct *vma)
{
	pte_t *src_pte, *dst_pte, entry;
	struct page *ptepage;
2550
	unsigned long addr;
2551
	int cow;
2552 2553
	struct hstate *h = hstate_vma(vma);
	unsigned long sz = huge_page_size(h);
2554 2555 2556
	unsigned long mmun_start;	/* For mmu_notifiers */
	unsigned long mmun_end;		/* For mmu_notifiers */
	int ret = 0;
2557 2558

	cow = (vma->vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
D
David Gibson 已提交
2559

2560 2561 2562 2563 2564
	mmun_start = vma->vm_start;
	mmun_end = vma->vm_end;
	if (cow)
		mmu_notifier_invalidate_range_start(src, mmun_start, mmun_end);

2565
	for (addr = vma->vm_start; addr < vma->vm_end; addr += sz) {
2566
		spinlock_t *src_ptl, *dst_ptl;
H
Hugh Dickins 已提交
2567 2568 2569
		src_pte = huge_pte_offset(src, addr);
		if (!src_pte)
			continue;
2570
		dst_pte = huge_pte_alloc(dst, addr, sz);
2571 2572 2573 2574
		if (!dst_pte) {
			ret = -ENOMEM;
			break;
		}
2575 2576 2577 2578 2579

		/* If the pagetables are shared don't copy or take references */
		if (dst_pte == src_pte)
			continue;

2580 2581 2582
		dst_ptl = huge_pte_lock(h, dst, dst_pte);
		src_ptl = huge_pte_lockptr(h, src, src_pte);
		spin_lock_nested(src_ptl, SINGLE_DEPTH_NESTING);
2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600
		entry = huge_ptep_get(src_pte);
		if (huge_pte_none(entry)) { /* skip none entry */
			;
		} else if (unlikely(is_hugetlb_entry_migration(entry) ||
				    is_hugetlb_entry_hwpoisoned(entry))) {
			swp_entry_t swp_entry = pte_to_swp_entry(entry);

			if (is_write_migration_entry(swp_entry) && cow) {
				/*
				 * COW mappings require pages in both
				 * parent and child to be set to read.
				 */
				make_migration_entry_read(&swp_entry);
				entry = swp_entry_to_pte(swp_entry);
				set_huge_pte_at(src, addr, src_pte, entry);
			}
			set_huge_pte_at(dst, addr, dst_pte, entry);
		} else {
2601
			if (cow) {
2602
				huge_ptep_set_wrprotect(src, addr, src_pte);
2603 2604 2605
				mmu_notifier_invalidate_range(src, mmun_start,
								   mmun_end);
			}
2606
			entry = huge_ptep_get(src_pte);
2607 2608
			ptepage = pte_page(entry);
			get_page(ptepage);
2609
			page_dup_rmap(ptepage);
2610 2611
			set_huge_pte_at(dst, addr, dst_pte, entry);
		}
2612 2613
		spin_unlock(src_ptl);
		spin_unlock(dst_ptl);
D
David Gibson 已提交
2614 2615
	}

2616 2617 2618 2619
	if (cow)
		mmu_notifier_invalidate_range_end(src, mmun_start, mmun_end);

	return ret;
D
David Gibson 已提交
2620 2621
}

2622 2623 2624
void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
			    unsigned long start, unsigned long end,
			    struct page *ref_page)
D
David Gibson 已提交
2625
{
2626
	int force_flush = 0;
D
David Gibson 已提交
2627 2628
	struct mm_struct *mm = vma->vm_mm;
	unsigned long address;
2629
	pte_t *ptep;
D
David Gibson 已提交
2630
	pte_t pte;
2631
	spinlock_t *ptl;
D
David Gibson 已提交
2632
	struct page *page;
2633 2634
	struct hstate *h = hstate_vma(vma);
	unsigned long sz = huge_page_size(h);
2635 2636
	const unsigned long mmun_start = start;	/* For mmu_notifiers */
	const unsigned long mmun_end   = end;	/* For mmu_notifiers */
2637

D
David Gibson 已提交
2638
	WARN_ON(!is_vm_hugetlb_page(vma));
2639 2640
	BUG_ON(start & ~huge_page_mask(h));
	BUG_ON(end & ~huge_page_mask(h));
D
David Gibson 已提交
2641

2642
	tlb_start_vma(tlb, vma);
2643
	mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);
2644
	address = start;
2645
again:
2646
	for (; address < end; address += sz) {
2647
		ptep = huge_pte_offset(mm, address);
A
Adam Litke 已提交
2648
		if (!ptep)
2649 2650
			continue;

2651
		ptl = huge_pte_lock(h, mm, ptep);
2652
		if (huge_pmd_unshare(mm, &address, ptep))
2653
			goto unlock;
2654

2655 2656
		pte = huge_ptep_get(ptep);
		if (huge_pte_none(pte))
2657
			goto unlock;
2658 2659

		/*
2660 2661
		 * Migrating hugepage or HWPoisoned hugepage is already
		 * unmapped and its refcount is dropped, so just clear pte here.
2662
		 */
2663
		if (unlikely(!pte_present(pte))) {
2664
			huge_pte_clear(mm, address, ptep);
2665
			goto unlock;
2666
		}
2667 2668

		page = pte_page(pte);
2669 2670 2671 2672 2673 2674 2675
		/*
		 * If a reference page is supplied, it is because a specific
		 * page is being unmapped, not a range. Ensure the page we
		 * are about to unmap is the actual page of interest.
		 */
		if (ref_page) {
			if (page != ref_page)
2676
				goto unlock;
2677 2678 2679 2680 2681 2682 2683 2684 2685

			/*
			 * Mark the VMA as having unmapped its page so that
			 * future faults in this VMA will fail rather than
			 * looking like data was lost
			 */
			set_vma_resv_flags(vma, HPAGE_RESV_UNMAPPED);
		}

2686
		pte = huge_ptep_get_and_clear(mm, address, ptep);
2687
		tlb_remove_tlb_entry(tlb, ptep, address);
2688
		if (huge_pte_dirty(pte))
2689
			set_page_dirty(page);
2690

2691 2692
		page_remove_rmap(page);
		force_flush = !__tlb_remove_page(tlb, page);
2693
		if (force_flush) {
2694
			address += sz;
2695
			spin_unlock(ptl);
2696
			break;
2697
		}
2698
		/* Bail out after unmapping reference page if supplied */
2699 2700
		if (ref_page) {
			spin_unlock(ptl);
2701
			break;
2702 2703 2704
		}
unlock:
		spin_unlock(ptl);
D
David Gibson 已提交
2705
	}
2706 2707 2708 2709 2710 2711 2712 2713 2714 2715
	/*
	 * mmu_gather ran out of room to batch pages, we break out of
	 * the PTE lock to avoid doing the potential expensive TLB invalidate
	 * and page-free while holding it.
	 */
	if (force_flush) {
		force_flush = 0;
		tlb_flush_mmu(tlb);
		if (address < end && !ref_page)
			goto again;
2716
	}
2717
	mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
2718
	tlb_end_vma(tlb, vma);
L
Linus Torvalds 已提交
2719
}
D
David Gibson 已提交
2720

2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732
void __unmap_hugepage_range_final(struct mmu_gather *tlb,
			  struct vm_area_struct *vma, unsigned long start,
			  unsigned long end, struct page *ref_page)
{
	__unmap_hugepage_range(tlb, vma, start, end, ref_page);

	/*
	 * Clear this flag so that x86's huge_pmd_share page_table_shareable
	 * test will fail on a vma being torn down, and not grab a page table
	 * on its way out.  We're lucky that the flag has such an appropriate
	 * name, and can in fact be safely cleared here. We could clear it
	 * before the __unmap_hugepage_range above, but all that's necessary
2733
	 * is to clear it before releasing the i_mmap_rwsem. This works
2734
	 * because in the context this is called, the VMA is about to be
2735
	 * destroyed and the i_mmap_rwsem is held.
2736 2737 2738 2739
	 */
	vma->vm_flags &= ~VM_MAYSHARE;
}

2740
void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
2741
			  unsigned long end, struct page *ref_page)
2742
{
2743 2744 2745 2746 2747
	struct mm_struct *mm;
	struct mmu_gather tlb;

	mm = vma->vm_mm;

2748
	tlb_gather_mmu(&tlb, mm, start, end);
2749 2750
	__unmap_hugepage_range(&tlb, vma, start, end, ref_page);
	tlb_finish_mmu(&tlb, start, end);
2751 2752
}

2753 2754 2755 2756 2757 2758
/*
 * This is called when the original mapper is failing to COW a MAP_PRIVATE
 * mappping it owns the reserve page for. The intention is to unmap the page
 * from other VMAs and let the children be SIGKILLed if they are faulting the
 * same region.
 */
2759 2760
static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma,
			      struct page *page, unsigned long address)
2761
{
2762
	struct hstate *h = hstate_vma(vma);
2763 2764 2765 2766 2767 2768 2769 2770
	struct vm_area_struct *iter_vma;
	struct address_space *mapping;
	pgoff_t pgoff;

	/*
	 * vm_pgoff is in PAGE_SIZE units, hence the different calculation
	 * from page cache lookup which is in HPAGE_SIZE units.
	 */
2771
	address = address & huge_page_mask(h);
2772 2773
	pgoff = ((address - vma->vm_start) >> PAGE_SHIFT) +
			vma->vm_pgoff;
A
Al Viro 已提交
2774
	mapping = file_inode(vma->vm_file)->i_mapping;
2775

2776 2777 2778 2779 2780
	/*
	 * Take the mapping lock for the duration of the table walk. As
	 * this mapping should be shared between all the VMAs,
	 * __unmap_hugepage_range() is called as the lock is already held
	 */
2781
	i_mmap_lock_write(mapping);
2782
	vma_interval_tree_foreach(iter_vma, &mapping->i_mmap, pgoff, pgoff) {
2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794
		/* Do not unmap the current VMA */
		if (iter_vma == vma)
			continue;

		/*
		 * Unmap the page from other VMAs without their own reserves.
		 * They get marked to be SIGKILLed if they fault in these
		 * areas. This is because a future no-page fault on this VMA
		 * could insert a zeroed page instead of the data existing
		 * from the time of fork. This would look like data corruption
		 */
		if (!is_vma_resv_set(iter_vma, HPAGE_RESV_OWNER))
2795 2796
			unmap_hugepage_range(iter_vma, address,
					     address + huge_page_size(h), page);
2797
	}
2798
	i_mmap_unlock_write(mapping);
2799 2800
}

2801 2802
/*
 * Hugetlb_cow() should be called with page lock of the original hugepage held.
2803 2804 2805
 * Called with hugetlb_instantiation_mutex held and pte_page locked so we
 * cannot race with other handlers or page migration.
 * Keep the pte_same checks anyway to make transition from the mutex easier.
2806
 */
2807
static int hugetlb_cow(struct mm_struct *mm, struct vm_area_struct *vma,
2808
			unsigned long address, pte_t *ptep, pte_t pte,
2809
			struct page *pagecache_page, spinlock_t *ptl)
2810
{
2811
	struct hstate *h = hstate_vma(vma);
2812
	struct page *old_page, *new_page;
2813
	int ret = 0, outside_reserve = 0;
2814 2815
	unsigned long mmun_start;	/* For mmu_notifiers */
	unsigned long mmun_end;		/* For mmu_notifiers */
2816 2817 2818

	old_page = pte_page(pte);

2819
retry_avoidcopy:
2820 2821
	/* If no-one else is actually using this page, avoid the copy
	 * and just make the page writable */
2822 2823
	if (page_mapcount(old_page) == 1 && PageAnon(old_page)) {
		page_move_anon_rmap(old_page, vma, address);
2824
		set_huge_ptep_writable(vma, address, ptep);
N
Nick Piggin 已提交
2825
		return 0;
2826 2827
	}

2828 2829 2830 2831 2832 2833 2834 2835 2836
	/*
	 * If the process that created a MAP_PRIVATE mapping is about to
	 * perform a COW due to a shared page count, attempt to satisfy
	 * the allocation without using the existing reserves. The pagecache
	 * page is used to determine if the reserve at this address was
	 * consumed or not. If reserves were used, a partial faulted mapping
	 * at the time of fork() could consume its reserves on COW instead
	 * of the full address range.
	 */
2837
	if (is_vma_resv_set(vma, HPAGE_RESV_OWNER) &&
2838 2839 2840
			old_page != pagecache_page)
		outside_reserve = 1;

2841
	page_cache_get(old_page);
2842

2843 2844 2845 2846
	/*
	 * Drop page table lock as buddy allocator may be called. It will
	 * be acquired again before returning to the caller, as expected.
	 */
2847
	spin_unlock(ptl);
2848
	new_page = alloc_huge_page(vma, address, outside_reserve);
2849

2850
	if (IS_ERR(new_page)) {
2851 2852 2853 2854 2855 2856 2857 2858
		/*
		 * If a process owning a MAP_PRIVATE mapping fails to COW,
		 * it is due to references held by a child and an insufficient
		 * huge page pool. To guarantee the original mappers
		 * reliability, unmap the page from child processes. The child
		 * may get SIGKILLed if it later faults.
		 */
		if (outside_reserve) {
2859
			page_cache_release(old_page);
2860
			BUG_ON(huge_pte_none(pte));
2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872
			unmap_ref_private(mm, vma, old_page, address);
			BUG_ON(huge_pte_none(pte));
			spin_lock(ptl);
			ptep = huge_pte_offset(mm, address & huge_page_mask(h));
			if (likely(ptep &&
				   pte_same(huge_ptep_get(ptep), pte)))
				goto retry_avoidcopy;
			/*
			 * race occurs while re-acquiring page table
			 * lock, and our job is done.
			 */
			return 0;
2873 2874
		}

2875 2876 2877
		ret = (PTR_ERR(new_page) == -ENOMEM) ?
			VM_FAULT_OOM : VM_FAULT_SIGBUS;
		goto out_release_old;
2878 2879
	}

2880 2881 2882 2883
	/*
	 * When the original hugepage is shared one, it does not have
	 * anon_vma prepared.
	 */
2884
	if (unlikely(anon_vma_prepare(vma))) {
2885 2886
		ret = VM_FAULT_OOM;
		goto out_release_all;
2887
	}
2888

A
Andrea Arcangeli 已提交
2889 2890
	copy_user_huge_page(new_page, old_page, address, vma,
			    pages_per_huge_page(h));
N
Nick Piggin 已提交
2891
	__SetPageUptodate(new_page);
2892

2893 2894 2895
	mmun_start = address & huge_page_mask(h);
	mmun_end = mmun_start + huge_page_size(h);
	mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);
2896

2897
	/*
2898
	 * Retake the page table lock to check for racing updates
2899 2900
	 * before the page tables are altered
	 */
2901
	spin_lock(ptl);
2902
	ptep = huge_pte_offset(mm, address & huge_page_mask(h));
2903
	if (likely(ptep && pte_same(huge_ptep_get(ptep), pte))) {
2904 2905
		ClearPagePrivate(new_page);

2906
		/* Break COW */
2907
		huge_ptep_clear_flush(vma, address, ptep);
2908
		mmu_notifier_invalidate_range(mm, mmun_start, mmun_end);
2909 2910
		set_huge_pte_at(mm, address, ptep,
				make_huge_pte(vma, new_page, 1));
2911
		page_remove_rmap(old_page);
2912
		hugepage_add_new_anon_rmap(new_page, vma, address);
2913 2914 2915
		/* Make the old page be freed below */
		new_page = old_page;
	}
2916
	spin_unlock(ptl);
2917
	mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
2918
out_release_all:
2919
	page_cache_release(new_page);
2920
out_release_old:
2921
	page_cache_release(old_page);
2922

2923 2924
	spin_lock(ptl); /* Caller expects lock to be held */
	return ret;
2925 2926
}

2927
/* Return the pagecache page at a given address within a VMA */
2928 2929
static struct page *hugetlbfs_pagecache_page(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
2930 2931
{
	struct address_space *mapping;
2932
	pgoff_t idx;
2933 2934

	mapping = vma->vm_file->f_mapping;
2935
	idx = vma_hugecache_offset(h, vma, address);
2936 2937 2938 2939

	return find_lock_page(mapping, idx);
}

H
Hugh Dickins 已提交
2940 2941 2942 2943 2944
/*
 * Return whether there is a pagecache page to back given address within VMA.
 * Caller follow_hugetlb_page() holds page_table_lock so we cannot lock_page.
 */
static bool hugetlbfs_pagecache_present(struct hstate *h,
H
Hugh Dickins 已提交
2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959
			struct vm_area_struct *vma, unsigned long address)
{
	struct address_space *mapping;
	pgoff_t idx;
	struct page *page;

	mapping = vma->vm_file->f_mapping;
	idx = vma_hugecache_offset(h, vma, address);

	page = find_get_page(mapping, idx);
	if (page)
		put_page(page);
	return page != NULL;
}

2960
static int hugetlb_no_page(struct mm_struct *mm, struct vm_area_struct *vma,
2961 2962
			   struct address_space *mapping, pgoff_t idx,
			   unsigned long address, pte_t *ptep, unsigned int flags)
2963
{
2964
	struct hstate *h = hstate_vma(vma);
2965
	int ret = VM_FAULT_SIGBUS;
2966
	int anon_rmap = 0;
A
Adam Litke 已提交
2967 2968
	unsigned long size;
	struct page *page;
2969
	pte_t new_pte;
2970
	spinlock_t *ptl;
A
Adam Litke 已提交
2971

2972 2973 2974
	/*
	 * Currently, we are forced to kill the process in the event the
	 * original mapper has unmapped pages from the child due to a failed
L
Lucas De Marchi 已提交
2975
	 * COW. Warn that such a situation has occurred as it may not be obvious
2976 2977
	 */
	if (is_vma_resv_set(vma, HPAGE_RESV_UNMAPPED)) {
2978 2979
		pr_warning("PID %d killed due to inadequate hugepage pool\n",
			   current->pid);
2980 2981 2982
		return ret;
	}

A
Adam Litke 已提交
2983 2984 2985 2986
	/*
	 * Use page lock to guard against racing truncation
	 * before we get page_table_lock.
	 */
2987 2988 2989
retry:
	page = find_lock_page(mapping, idx);
	if (!page) {
2990
		size = i_size_read(mapping->host) >> huge_page_shift(h);
2991 2992
		if (idx >= size)
			goto out;
2993
		page = alloc_huge_page(vma, address, 0);
2994
		if (IS_ERR(page)) {
2995 2996 2997 2998 2999
			ret = PTR_ERR(page);
			if (ret == -ENOMEM)
				ret = VM_FAULT_OOM;
			else
				ret = VM_FAULT_SIGBUS;
3000 3001
			goto out;
		}
A
Andrea Arcangeli 已提交
3002
		clear_huge_page(page, address, pages_per_huge_page(h));
N
Nick Piggin 已提交
3003
		__SetPageUptodate(page);
3004

3005
		if (vma->vm_flags & VM_MAYSHARE) {
3006
			int err;
K
Ken Chen 已提交
3007
			struct inode *inode = mapping->host;
3008 3009 3010 3011 3012 3013 3014 3015

			err = add_to_page_cache(page, mapping, idx, GFP_KERNEL);
			if (err) {
				put_page(page);
				if (err == -EEXIST)
					goto retry;
				goto out;
			}
3016
			ClearPagePrivate(page);
K
Ken Chen 已提交
3017 3018

			spin_lock(&inode->i_lock);
3019
			inode->i_blocks += blocks_per_huge_page(h);
K
Ken Chen 已提交
3020
			spin_unlock(&inode->i_lock);
3021
		} else {
3022
			lock_page(page);
3023 3024 3025 3026
			if (unlikely(anon_vma_prepare(vma))) {
				ret = VM_FAULT_OOM;
				goto backout_unlocked;
			}
3027
			anon_rmap = 1;
3028
		}
3029
	} else {
3030 3031 3032 3033 3034 3035
		/*
		 * If memory error occurs between mmap() and fault, some process
		 * don't have hwpoisoned swap entry for errored virtual address.
		 * So we need to block hugepage fault by PG_hwpoison bit check.
		 */
		if (unlikely(PageHWPoison(page))) {
3036
			ret = VM_FAULT_HWPOISON |
3037
				VM_FAULT_SET_HINDEX(hstate_index(h));
3038 3039
			goto backout_unlocked;
		}
3040
	}
3041

3042 3043 3044 3045 3046 3047
	/*
	 * If we are going to COW a private mapping later, we examine the
	 * pending reservations for this page now. This will ensure that
	 * any allocations necessary to record that reservation occur outside
	 * the spinlock.
	 */
3048
	if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED))
3049 3050 3051 3052
		if (vma_needs_reservation(h, vma, address) < 0) {
			ret = VM_FAULT_OOM;
			goto backout_unlocked;
		}
3053

3054 3055
	ptl = huge_pte_lockptr(h, mm, ptep);
	spin_lock(ptl);
3056
	size = i_size_read(mapping->host) >> huge_page_shift(h);
A
Adam Litke 已提交
3057 3058 3059
	if (idx >= size)
		goto backout;

N
Nick Piggin 已提交
3060
	ret = 0;
3061
	if (!huge_pte_none(huge_ptep_get(ptep)))
A
Adam Litke 已提交
3062 3063
		goto backout;

3064 3065
	if (anon_rmap) {
		ClearPagePrivate(page);
3066
		hugepage_add_new_anon_rmap(page, vma, address);
3067
	} else
3068
		page_dup_rmap(page);
3069 3070 3071 3072
	new_pte = make_huge_pte(vma, page, ((vma->vm_flags & VM_WRITE)
				&& (vma->vm_flags & VM_SHARED)));
	set_huge_pte_at(mm, address, ptep, new_pte);

3073
	if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) {
3074
		/* Optimization, do the COW without a second fault */
3075
		ret = hugetlb_cow(mm, vma, address, ptep, new_pte, page, ptl);
3076 3077
	}

3078
	spin_unlock(ptl);
A
Adam Litke 已提交
3079 3080
	unlock_page(page);
out:
3081
	return ret;
A
Adam Litke 已提交
3082 3083

backout:
3084
	spin_unlock(ptl);
3085
backout_unlocked:
A
Adam Litke 已提交
3086 3087 3088
	unlock_page(page);
	put_page(page);
	goto out;
3089 3090
}

3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125
#ifdef CONFIG_SMP
static u32 fault_mutex_hash(struct hstate *h, struct mm_struct *mm,
			    struct vm_area_struct *vma,
			    struct address_space *mapping,
			    pgoff_t idx, unsigned long address)
{
	unsigned long key[2];
	u32 hash;

	if (vma->vm_flags & VM_SHARED) {
		key[0] = (unsigned long) mapping;
		key[1] = idx;
	} else {
		key[0] = (unsigned long) mm;
		key[1] = address >> huge_page_shift(h);
	}

	hash = jhash2((u32 *)&key, sizeof(key)/sizeof(u32), 0);

	return hash & (num_fault_mutexes - 1);
}
#else
/*
 * For uniprocesor systems we always use a single mutex, so just
 * return 0 and avoid the hashing overhead.
 */
static u32 fault_mutex_hash(struct hstate *h, struct mm_struct *mm,
			    struct vm_area_struct *vma,
			    struct address_space *mapping,
			    pgoff_t idx, unsigned long address)
{
	return 0;
}
#endif

3126
int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3127
			unsigned long address, unsigned int flags)
3128
{
3129
	pte_t *ptep, entry;
3130
	spinlock_t *ptl;
3131
	int ret;
3132 3133
	u32 hash;
	pgoff_t idx;
3134
	struct page *page = NULL;
3135
	struct page *pagecache_page = NULL;
3136
	struct hstate *h = hstate_vma(vma);
3137
	struct address_space *mapping;
3138
	int need_wait_lock = 0;
3139

3140 3141
	address &= huge_page_mask(h);

3142 3143 3144
	ptep = huge_pte_offset(mm, address);
	if (ptep) {
		entry = huge_ptep_get(ptep);
N
Naoya Horiguchi 已提交
3145
		if (unlikely(is_hugetlb_entry_migration(entry))) {
3146
			migration_entry_wait_huge(vma, mm, ptep);
N
Naoya Horiguchi 已提交
3147 3148
			return 0;
		} else if (unlikely(is_hugetlb_entry_hwpoisoned(entry)))
3149
			return VM_FAULT_HWPOISON_LARGE |
3150
				VM_FAULT_SET_HINDEX(hstate_index(h));
3151 3152
	}

3153
	ptep = huge_pte_alloc(mm, address, huge_page_size(h));
3154 3155 3156
	if (!ptep)
		return VM_FAULT_OOM;

3157 3158 3159
	mapping = vma->vm_file->f_mapping;
	idx = vma_hugecache_offset(h, vma, address);

3160 3161 3162 3163 3164
	/*
	 * Serialize hugepage allocation and instantiation, so that we don't
	 * get spurious allocation failures if two CPUs race to instantiate
	 * the same page in the page cache.
	 */
3165 3166 3167
	hash = fault_mutex_hash(h, mm, vma, mapping, idx, address);
	mutex_lock(&htlb_fault_mutex_table[hash]);

3168 3169
	entry = huge_ptep_get(ptep);
	if (huge_pte_none(entry)) {
3170
		ret = hugetlb_no_page(mm, vma, mapping, idx, address, ptep, flags);
3171
		goto out_mutex;
3172
	}
3173

N
Nick Piggin 已提交
3174
	ret = 0;
3175

3176 3177 3178 3179 3180 3181 3182 3183 3184 3185
	/*
	 * entry could be a migration/hwpoison entry at this point, so this
	 * check prevents the kernel from going below assuming that we have
	 * a active hugepage in pagecache. This goto expects the 2nd page fault,
	 * and is_hugetlb_entry_(migration|hwpoisoned) check will properly
	 * handle it.
	 */
	if (!pte_present(entry))
		goto out_mutex;

3186 3187 3188 3189 3190 3191 3192 3193
	/*
	 * If we are going to COW the mapping later, we examine the pending
	 * reservations for this page now. This will ensure that any
	 * allocations necessary to record that reservation occur outside the
	 * spinlock. For private mappings, we also lookup the pagecache
	 * page now as it is used to determine if a reservation has been
	 * consumed.
	 */
3194
	if ((flags & FAULT_FLAG_WRITE) && !huge_pte_write(entry)) {
3195 3196
		if (vma_needs_reservation(h, vma, address) < 0) {
			ret = VM_FAULT_OOM;
3197
			goto out_mutex;
3198
		}
3199

3200
		if (!(vma->vm_flags & VM_MAYSHARE))
3201 3202 3203 3204
			pagecache_page = hugetlbfs_pagecache_page(h,
								vma, address);
	}

3205 3206 3207 3208 3209 3210
	ptl = huge_pte_lock(h, mm, ptep);

	/* Check for a racing update before calling hugetlb_cow */
	if (unlikely(!pte_same(entry, huge_ptep_get(ptep))))
		goto out_ptl;

3211 3212 3213 3214 3215 3216 3217
	/*
	 * hugetlb_cow() requires page locks of pte_page(entry) and
	 * pagecache_page, so here we need take the former one
	 * when page != pagecache_page or !pagecache_page.
	 */
	page = pte_page(entry);
	if (page != pagecache_page)
3218 3219 3220 3221
		if (!trylock_page(page)) {
			need_wait_lock = 1;
			goto out_ptl;
		}
3222

3223
	get_page(page);
3224

3225
	if (flags & FAULT_FLAG_WRITE) {
3226
		if (!huge_pte_write(entry)) {
3227
			ret = hugetlb_cow(mm, vma, address, ptep, entry,
3228
					pagecache_page, ptl);
3229
			goto out_put_page;
3230
		}
3231
		entry = huge_pte_mkdirty(entry);
3232 3233
	}
	entry = pte_mkyoung(entry);
3234 3235
	if (huge_ptep_set_access_flags(vma, address, ptep, entry,
						flags & FAULT_FLAG_WRITE))
3236
		update_mmu_cache(vma, address, ptep);
3237 3238 3239 3240
out_put_page:
	if (page != pagecache_page)
		unlock_page(page);
	put_page(page);
3241 3242
out_ptl:
	spin_unlock(ptl);
3243 3244 3245 3246 3247

	if (pagecache_page) {
		unlock_page(pagecache_page);
		put_page(pagecache_page);
	}
3248
out_mutex:
3249
	mutex_unlock(&htlb_fault_mutex_table[hash]);
3250 3251 3252 3253 3254 3255 3256 3257 3258
	/*
	 * Generally it's safe to hold refcount during waiting page lock. But
	 * here we just wait to defer the next page fault to avoid busy loop and
	 * the page is not used after unlocked before returning from the current
	 * page fault. So we are safe from accessing freed page, even if we wait
	 * here without taking refcount.
	 */
	if (need_wait_lock)
		wait_on_page_locked(page);
3259
	return ret;
3260 3261
}

3262 3263 3264 3265
long follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma,
			 struct page **pages, struct vm_area_struct **vmas,
			 unsigned long *position, unsigned long *nr_pages,
			 long i, unsigned int flags)
D
David Gibson 已提交
3266
{
3267 3268
	unsigned long pfn_offset;
	unsigned long vaddr = *position;
3269
	unsigned long remainder = *nr_pages;
3270
	struct hstate *h = hstate_vma(vma);
D
David Gibson 已提交
3271 3272

	while (vaddr < vma->vm_end && remainder) {
A
Adam Litke 已提交
3273
		pte_t *pte;
3274
		spinlock_t *ptl = NULL;
H
Hugh Dickins 已提交
3275
		int absent;
A
Adam Litke 已提交
3276
		struct page *page;
D
David Gibson 已提交
3277

A
Adam Litke 已提交
3278 3279
		/*
		 * Some archs (sparc64, sh*) have multiple pte_ts to
H
Hugh Dickins 已提交
3280
		 * each hugepage.  We have to make sure we get the
A
Adam Litke 已提交
3281
		 * first, for the page indexing below to work.
3282 3283
		 *
		 * Note that page table lock is not held when pte is null.
A
Adam Litke 已提交
3284
		 */
3285
		pte = huge_pte_offset(mm, vaddr & huge_page_mask(h));
3286 3287
		if (pte)
			ptl = huge_pte_lock(h, mm, pte);
H
Hugh Dickins 已提交
3288 3289 3290 3291
		absent = !pte || huge_pte_none(huge_ptep_get(pte));

		/*
		 * When coredumping, it suits get_dump_page if we just return
H
Hugh Dickins 已提交
3292 3293 3294 3295
		 * an error where there's an empty slot with no huge pagecache
		 * to back it.  This way, we avoid allocating a hugepage, and
		 * the sparse dumpfile avoids allocating disk blocks, but its
		 * huge holes still show up with zeroes where they need to be.
H
Hugh Dickins 已提交
3296
		 */
H
Hugh Dickins 已提交
3297 3298
		if (absent && (flags & FOLL_DUMP) &&
		    !hugetlbfs_pagecache_present(h, vma, vaddr)) {
3299 3300
			if (pte)
				spin_unlock(ptl);
H
Hugh Dickins 已提交
3301 3302 3303
			remainder = 0;
			break;
		}
D
David Gibson 已提交
3304

3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315
		/*
		 * We need call hugetlb_fault for both hugepages under migration
		 * (in which case hugetlb_fault waits for the migration,) and
		 * hwpoisoned hugepages (in which case we need to prevent the
		 * caller from accessing to them.) In order to do this, we use
		 * here is_swap_pte instead of is_hugetlb_entry_migration and
		 * is_hugetlb_entry_hwpoisoned. This is because it simply covers
		 * both cases, and because we can't follow correct pages
		 * directly from any kind of swap entries.
		 */
		if (absent || is_swap_pte(huge_ptep_get(pte)) ||
3316 3317
		    ((flags & FOLL_WRITE) &&
		      !huge_pte_write(huge_ptep_get(pte)))) {
A
Adam Litke 已提交
3318
			int ret;
D
David Gibson 已提交
3319

3320 3321
			if (pte)
				spin_unlock(ptl);
H
Hugh Dickins 已提交
3322 3323
			ret = hugetlb_fault(mm, vma, vaddr,
				(flags & FOLL_WRITE) ? FAULT_FLAG_WRITE : 0);
3324
			if (!(ret & VM_FAULT_ERROR))
A
Adam Litke 已提交
3325
				continue;
D
David Gibson 已提交
3326

A
Adam Litke 已提交
3327 3328 3329 3330
			remainder = 0;
			break;
		}

3331
		pfn_offset = (vaddr & ~huge_page_mask(h)) >> PAGE_SHIFT;
3332
		page = pte_page(huge_ptep_get(pte));
3333
same_page:
3334
		if (pages) {
H
Hugh Dickins 已提交
3335
			pages[i] = mem_map_offset(page, pfn_offset);
3336
			get_page_foll(pages[i]);
3337
		}
D
David Gibson 已提交
3338 3339 3340 3341 3342

		if (vmas)
			vmas[i] = vma;

		vaddr += PAGE_SIZE;
3343
		++pfn_offset;
D
David Gibson 已提交
3344 3345
		--remainder;
		++i;
3346
		if (vaddr < vma->vm_end && remainder &&
3347
				pfn_offset < pages_per_huge_page(h)) {
3348 3349 3350 3351 3352 3353
			/*
			 * We use pfn_offset to avoid touching the pageframes
			 * of this compound page.
			 */
			goto same_page;
		}
3354
		spin_unlock(ptl);
D
David Gibson 已提交
3355
	}
3356
	*nr_pages = remainder;
D
David Gibson 已提交
3357 3358
	*position = vaddr;

H
Hugh Dickins 已提交
3359
	return i ? i : -EFAULT;
D
David Gibson 已提交
3360
}
3361

3362
unsigned long hugetlb_change_protection(struct vm_area_struct *vma,
3363 3364 3365 3366 3367 3368
		unsigned long address, unsigned long end, pgprot_t newprot)
{
	struct mm_struct *mm = vma->vm_mm;
	unsigned long start = address;
	pte_t *ptep;
	pte_t pte;
3369
	struct hstate *h = hstate_vma(vma);
3370
	unsigned long pages = 0;
3371 3372 3373 3374

	BUG_ON(address >= end);
	flush_cache_range(vma, address, end);

3375
	mmu_notifier_invalidate_range_start(mm, start, end);
3376
	i_mmap_lock_write(vma->vm_file->f_mapping);
3377
	for (; address < end; address += huge_page_size(h)) {
3378
		spinlock_t *ptl;
3379 3380 3381
		ptep = huge_pte_offset(mm, address);
		if (!ptep)
			continue;
3382
		ptl = huge_pte_lock(h, mm, ptep);
3383 3384
		if (huge_pmd_unshare(mm, &address, ptep)) {
			pages++;
3385
			spin_unlock(ptl);
3386
			continue;
3387
		}
3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407
		pte = huge_ptep_get(ptep);
		if (unlikely(is_hugetlb_entry_hwpoisoned(pte))) {
			spin_unlock(ptl);
			continue;
		}
		if (unlikely(is_hugetlb_entry_migration(pte))) {
			swp_entry_t entry = pte_to_swp_entry(pte);

			if (is_write_migration_entry(entry)) {
				pte_t newpte;

				make_migration_entry_read(&entry);
				newpte = swp_entry_to_pte(entry);
				set_huge_pte_at(mm, address, ptep, newpte);
				pages++;
			}
			spin_unlock(ptl);
			continue;
		}
		if (!huge_pte_none(pte)) {
3408
			pte = huge_ptep_get_and_clear(mm, address, ptep);
3409
			pte = pte_mkhuge(huge_pte_modify(pte, newprot));
3410
			pte = arch_make_huge_pte(pte, vma, NULL, 0);
3411
			set_huge_pte_at(mm, address, ptep, pte);
3412
			pages++;
3413
		}
3414
		spin_unlock(ptl);
3415
	}
3416
	/*
3417
	 * Must flush TLB before releasing i_mmap_rwsem: x86's huge_pmd_unshare
3418
	 * may have cleared our pud entry and done put_page on the page table:
3419
	 * once we release i_mmap_rwsem, another task can do the final put_page
3420 3421
	 * and that page table be reused and filled with junk.
	 */
3422
	flush_tlb_range(vma, start, end);
3423
	mmu_notifier_invalidate_range(mm, start, end);
3424
	i_mmap_unlock_write(vma->vm_file->f_mapping);
3425
	mmu_notifier_invalidate_range_end(mm, start, end);
3426 3427

	return pages << h->order;
3428 3429
}

3430 3431
int hugetlb_reserve_pages(struct inode *inode,
					long from, long to,
3432
					struct vm_area_struct *vma,
3433
					vm_flags_t vm_flags)
3434
{
3435
	long ret, chg;
3436
	struct hstate *h = hstate_inode(inode);
3437
	struct hugepage_subpool *spool = subpool_inode(inode);
3438
	struct resv_map *resv_map;
3439

3440 3441 3442
	/*
	 * Only apply hugepage reservation if asked. At fault time, an
	 * attempt will be made for VM_NORESERVE to allocate a page
3443
	 * without using reserves
3444
	 */
3445
	if (vm_flags & VM_NORESERVE)
3446 3447
		return 0;

3448 3449 3450 3451 3452 3453
	/*
	 * Shared mappings base their reservation on the number of pages that
	 * are already allocated on behalf of the file. Private mappings need
	 * to reserve the full area even if read-only as mprotect() may be
	 * called to make the mapping read-write. Assume !vma is a shm mapping
	 */
3454
	if (!vma || vma->vm_flags & VM_MAYSHARE) {
3455
		resv_map = inode_resv_map(inode);
3456

3457
		chg = region_chg(resv_map, from, to);
3458 3459 3460

	} else {
		resv_map = resv_map_alloc();
3461 3462 3463
		if (!resv_map)
			return -ENOMEM;

3464
		chg = to - from;
3465

3466 3467 3468 3469
		set_vma_resv_map(vma, resv_map);
		set_vma_resv_flags(vma, HPAGE_RESV_OWNER);
	}

3470 3471 3472 3473
	if (chg < 0) {
		ret = chg;
		goto out_err;
	}
3474

3475
	/* There must be enough pages in the subpool for the mapping */
3476 3477 3478 3479
	if (hugepage_subpool_get_pages(spool, chg)) {
		ret = -ENOSPC;
		goto out_err;
	}
3480 3481

	/*
3482
	 * Check enough hugepages are available for the reservation.
3483
	 * Hand the pages back to the subpool if there are not
3484
	 */
3485
	ret = hugetlb_acct_memory(h, chg);
K
Ken Chen 已提交
3486
	if (ret < 0) {
3487
		hugepage_subpool_put_pages(spool, chg);
3488
		goto out_err;
K
Ken Chen 已提交
3489
	}
3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501

	/*
	 * Account for the reservations made. Shared mappings record regions
	 * that have reservations as they are shared by multiple VMAs.
	 * When the last VMA disappears, the region map says how much
	 * the reservation was and the page cache tells how much of
	 * the reservation was consumed. Private mappings are per-VMA and
	 * only the consumed reservations are tracked. When the VMA
	 * disappears, the original reservation is the VMA size and the
	 * consumed reservations are stored in the map. Hence, nothing
	 * else has to be done for private mappings here
	 */
3502
	if (!vma || vma->vm_flags & VM_MAYSHARE)
3503
		region_add(resv_map, from, to);
3504
	return 0;
3505
out_err:
J
Joonsoo Kim 已提交
3506 3507
	if (vma && is_vma_resv_set(vma, HPAGE_RESV_OWNER))
		kref_put(&resv_map->refs, resv_map_release);
3508
	return ret;
3509 3510 3511 3512
}

void hugetlb_unreserve_pages(struct inode *inode, long offset, long freed)
{
3513
	struct hstate *h = hstate_inode(inode);
3514
	struct resv_map *resv_map = inode_resv_map(inode);
3515
	long chg = 0;
3516
	struct hugepage_subpool *spool = subpool_inode(inode);
K
Ken Chen 已提交
3517

3518
	if (resv_map)
3519
		chg = region_truncate(resv_map, offset);
K
Ken Chen 已提交
3520
	spin_lock(&inode->i_lock);
3521
	inode->i_blocks -= (blocks_per_huge_page(h) * freed);
K
Ken Chen 已提交
3522 3523
	spin_unlock(&inode->i_lock);

3524
	hugepage_subpool_put_pages(spool, (chg - freed));
3525
	hugetlb_acct_memory(h, -(chg - freed));
3526
}
3527

3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572
#ifdef CONFIG_ARCH_WANT_HUGE_PMD_SHARE
static unsigned long page_table_shareable(struct vm_area_struct *svma,
				struct vm_area_struct *vma,
				unsigned long addr, pgoff_t idx)
{
	unsigned long saddr = ((idx - svma->vm_pgoff) << PAGE_SHIFT) +
				svma->vm_start;
	unsigned long sbase = saddr & PUD_MASK;
	unsigned long s_end = sbase + PUD_SIZE;

	/* Allow segments to share if only one is marked locked */
	unsigned long vm_flags = vma->vm_flags & ~VM_LOCKED;
	unsigned long svm_flags = svma->vm_flags & ~VM_LOCKED;

	/*
	 * match the virtual addresses, permission and the alignment of the
	 * page table page.
	 */
	if (pmd_index(addr) != pmd_index(saddr) ||
	    vm_flags != svm_flags ||
	    sbase < svma->vm_start || svma->vm_end < s_end)
		return 0;

	return saddr;
}

static int vma_shareable(struct vm_area_struct *vma, unsigned long addr)
{
	unsigned long base = addr & PUD_MASK;
	unsigned long end = base + PUD_SIZE;

	/*
	 * check on proper vm_flags and page table alignment
	 */
	if (vma->vm_flags & VM_MAYSHARE &&
	    vma->vm_start <= base && end <= vma->vm_end)
		return 1;
	return 0;
}

/*
 * Search for a shareable pmd page for hugetlb. In any case calls pmd_alloc()
 * and returns the corresponding pte. While this is not necessary for the
 * !shared pmd case because we can allocate the pmd later as well, it makes the
 * code much cleaner. pmd allocation is essential for the shared case because
3573
 * pud has to be populated inside the same i_mmap_rwsem section - otherwise
3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586
 * racing tasks could either miss the sharing (see huge_pte_offset) or select a
 * bad pmd for sharing.
 */
pte_t *huge_pmd_share(struct mm_struct *mm, unsigned long addr, pud_t *pud)
{
	struct vm_area_struct *vma = find_vma(mm, addr);
	struct address_space *mapping = vma->vm_file->f_mapping;
	pgoff_t idx = ((addr - vma->vm_start) >> PAGE_SHIFT) +
			vma->vm_pgoff;
	struct vm_area_struct *svma;
	unsigned long saddr;
	pte_t *spte = NULL;
	pte_t *pte;
3587
	spinlock_t *ptl;
3588 3589 3590 3591

	if (!vma_shareable(vma, addr))
		return (pte_t *)pmd_alloc(mm, pud, addr);

3592
	i_mmap_lock_write(mapping);
3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609
	vma_interval_tree_foreach(svma, &mapping->i_mmap, idx, idx) {
		if (svma == vma)
			continue;

		saddr = page_table_shareable(svma, vma, addr, idx);
		if (saddr) {
			spte = huge_pte_offset(svma->vm_mm, saddr);
			if (spte) {
				get_page(virt_to_page(spte));
				break;
			}
		}
	}

	if (!spte)
		goto out;

3610 3611
	ptl = huge_pte_lockptr(hstate_vma(vma), mm, spte);
	spin_lock(ptl);
3612 3613 3614 3615 3616
	if (pud_none(*pud))
		pud_populate(mm, pud,
				(pmd_t *)((unsigned long)spte & PAGE_MASK));
	else
		put_page(virt_to_page(spte));
3617
	spin_unlock(ptl);
3618 3619
out:
	pte = (pte_t *)pmd_alloc(mm, pud, addr);
3620
	i_mmap_unlock_write(mapping);
3621 3622 3623 3624 3625 3626 3627 3628 3629 3630
	return pte;
}

/*
 * unmap huge page backed by shared pte.
 *
 * Hugetlb pte page is ref counted at the time of mapping.  If pte is shared
 * indicated by page_count > 1, unmap is achieved by clearing pud and
 * decrementing the ref count. If count == 1, the pte page is not shared.
 *
3631
 * called with page table lock held.
3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649
 *
 * returns: 1 successfully unmapped a shared pte page
 *	    0 the underlying pte page is not shared, or it is the last user
 */
int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep)
{
	pgd_t *pgd = pgd_offset(mm, *addr);
	pud_t *pud = pud_offset(pgd, *addr);

	BUG_ON(page_count(virt_to_page(ptep)) == 0);
	if (page_count(virt_to_page(ptep)) == 1)
		return 0;

	pud_clear(pud);
	put_page(virt_to_page(ptep));
	*addr = ALIGN(*addr, HPAGE_SIZE * PTRS_PER_PTE) - HPAGE_SIZE;
	return 1;
}
3650 3651 3652 3653 3654 3655 3656
#define want_pmd_share()	(1)
#else /* !CONFIG_ARCH_WANT_HUGE_PMD_SHARE */
pte_t *huge_pmd_share(struct mm_struct *mm, unsigned long addr, pud_t *pud)
{
	return NULL;
}
#define want_pmd_share()	(0)
3657 3658
#endif /* CONFIG_ARCH_WANT_HUGE_PMD_SHARE */

3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702
#ifdef CONFIG_ARCH_WANT_GENERAL_HUGETLB
pte_t *huge_pte_alloc(struct mm_struct *mm,
			unsigned long addr, unsigned long sz)
{
	pgd_t *pgd;
	pud_t *pud;
	pte_t *pte = NULL;

	pgd = pgd_offset(mm, addr);
	pud = pud_alloc(mm, pgd, addr);
	if (pud) {
		if (sz == PUD_SIZE) {
			pte = (pte_t *)pud;
		} else {
			BUG_ON(sz != PMD_SIZE);
			if (want_pmd_share() && pud_none(*pud))
				pte = huge_pmd_share(mm, addr, pud);
			else
				pte = (pte_t *)pmd_alloc(mm, pud, addr);
		}
	}
	BUG_ON(pte && !pte_none(*pte) && !pte_huge(*pte));

	return pte;
}

pte_t *huge_pte_offset(struct mm_struct *mm, unsigned long addr)
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd = NULL;

	pgd = pgd_offset(mm, addr);
	if (pgd_present(*pgd)) {
		pud = pud_offset(pgd, addr);
		if (pud_present(*pud)) {
			if (pud_huge(*pud))
				return (pte_t *)pud;
			pmd = pmd_offset(pud, addr);
		}
	}
	return (pte_t *) pmd;
}

3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716
#endif /* CONFIG_ARCH_WANT_GENERAL_HUGETLB */

/*
 * These functions are overwritable if your architecture needs its own
 * behavior.
 */
struct page * __weak
follow_huge_addr(struct mm_struct *mm, unsigned long address,
			      int write)
{
	return ERR_PTR(-EINVAL);
}

struct page * __weak
3717
follow_huge_pmd(struct mm_struct *mm, unsigned long address,
3718
		pmd_t *pmd, int flags)
3719
{
3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748
	struct page *page = NULL;
	spinlock_t *ptl;
retry:
	ptl = pmd_lockptr(mm, pmd);
	spin_lock(ptl);
	/*
	 * make sure that the address range covered by this pmd is not
	 * unmapped from other threads.
	 */
	if (!pmd_huge(*pmd))
		goto out;
	if (pmd_present(*pmd)) {
		page = pte_page(*(pte_t *)pmd) +
			((address & ~PMD_MASK) >> PAGE_SHIFT);
		if (flags & FOLL_GET)
			get_page(page);
	} else {
		if (is_hugetlb_entry_migration(huge_ptep_get((pte_t *)pmd))) {
			spin_unlock(ptl);
			__migration_entry_wait(mm, (pte_t *)pmd, ptl);
			goto retry;
		}
		/*
		 * hwpoisoned entry is treated as no_page_table in
		 * follow_page_mask().
		 */
	}
out:
	spin_unlock(ptl);
3749 3750 3751
	return page;
}

3752
struct page * __weak
3753
follow_huge_pud(struct mm_struct *mm, unsigned long address,
3754
		pud_t *pud, int flags)
3755
{
3756 3757
	if (flags & FOLL_GET)
		return NULL;
3758

3759
	return pte_page(*(pte_t *)pud) + ((address & ~PUD_MASK) >> PAGE_SHIFT);
3760 3761
}

3762 3763
#ifdef CONFIG_MEMORY_FAILURE

3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777
/* Should be called in hugetlb_lock */
static int is_hugepage_on_freelist(struct page *hpage)
{
	struct page *page;
	struct page *tmp;
	struct hstate *h = page_hstate(hpage);
	int nid = page_to_nid(hpage);

	list_for_each_entry_safe(page, tmp, &h->hugepage_freelists[nid], lru)
		if (page == hpage)
			return 1;
	return 0;
}

3778 3779 3780 3781
/*
 * This function is called from memory failure code.
 * Assume the caller holds page lock of the head page.
 */
3782
int dequeue_hwpoisoned_huge_page(struct page *hpage)
3783 3784 3785
{
	struct hstate *h = page_hstate(hpage);
	int nid = page_to_nid(hpage);
3786
	int ret = -EBUSY;
3787 3788

	spin_lock(&hugetlb_lock);
3789
	if (is_hugepage_on_freelist(hpage)) {
3790 3791 3792 3793 3794 3795 3796
		/*
		 * Hwpoisoned hugepage isn't linked to activelist or freelist,
		 * but dangling hpage->lru can trigger list-debug warnings
		 * (this happens when we call unpoison_memory() on it),
		 * so let it point to itself with list_del_init().
		 */
		list_del_init(&hpage->lru);
3797
		set_page_refcounted(hpage);
3798 3799 3800 3801
		h->free_huge_pages--;
		h->free_huge_pages_node[nid]--;
		ret = 0;
	}
3802
	spin_unlock(&hugetlb_lock);
3803
	return ret;
3804
}
3805
#endif
3806 3807 3808

bool isolate_huge_page(struct page *page, struct list_head *list)
{
3809
	VM_BUG_ON_PAGE(!PageHead(page), page);
3810 3811 3812 3813 3814 3815 3816 3817 3818 3819
	if (!get_page_unless_zero(page))
		return false;
	spin_lock(&hugetlb_lock);
	list_move_tail(&page->lru, list);
	spin_unlock(&hugetlb_lock);
	return true;
}

void putback_active_hugepage(struct page *page)
{
3820
	VM_BUG_ON_PAGE(!PageHead(page), page);
3821 3822 3823 3824 3825
	spin_lock(&hugetlb_lock);
	list_move_tail(&page->lru, &(page_hstate(page))->hugepage_activelist);
	spin_unlock(&hugetlb_lock);
	put_page(page);
}
3826 3827 3828

bool is_hugepage_active(struct page *page)
{
3829
	VM_BUG_ON_PAGE(!PageHuge(page), page);
3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847
	/*
	 * This function can be called for a tail page because the caller,
	 * scan_movable_pages, scans through a given pfn-range which typically
	 * covers one memory block. In systems using gigantic hugepage (1GB
	 * for x86_64,) a hugepage is larger than a memory block, and we don't
	 * support migrating such large hugepages for now, so return false
	 * when called for tail pages.
	 */
	if (PageTail(page))
		return false;
	/*
	 * Refcount of a hwpoisoned hugepages is 1, but they are not active,
	 * so we should return false for them.
	 */
	if (unlikely(PageHWPoison(page)))
		return false;
	return page_count(page) > 0;
}