hugetlb.c 56.1 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8 9 10 11 12
/*
 * Generic hugetlb support.
 * (C) William Irwin, April 2004
 */
#include <linux/gfp.h>
#include <linux/list.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/mm.h>
#include <linux/sysctl.h>
#include <linux/highmem.h>
#include <linux/nodemask.h>
David Gibson's avatar
David Gibson committed
13
#include <linux/pagemap.h>
14
#include <linux/mempolicy.h>
15
#include <linux/cpuset.h>
16
#include <linux/mutex.h>
17
#include <linux/bootmem.h>
18
#include <linux/sysfs.h>
19

David Gibson's avatar
David Gibson committed
20 21 22 23
#include <asm/page.h>
#include <asm/pgtable.h>

#include <linux/hugetlb.h>
24
#include "internal.h"
Linus Torvalds's avatar
Linus Torvalds committed
25 26

const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
27 28
static gfp_t htlb_alloc_mask = GFP_HIGHUSER;
unsigned long hugepages_treat_as_movable;
29

30 31 32 33
static int max_hstate;
unsigned int default_hstate_idx;
struct hstate hstates[HUGE_MAX_HSTATE];

34 35
__initdata LIST_HEAD(huge_boot_pages);

36 37 38
/* for command line parsing */
static struct hstate * __initdata parsed_hstate;
static unsigned long __initdata default_hstate_max_huge_pages;
39
static unsigned long __initdata default_hstate_size;
40 41 42

#define for_each_hstate(h) \
	for ((h) = hstates; (h) < &hstates[max_hstate]; (h)++)
43

44 45 46 47
/*
 * Protects updates to hugepage_freelists, nr_huge_pages, and free_huge_pages
 */
static DEFINE_SPINLOCK(hugetlb_lock);
48

49 50 51
/*
 * Region tracking -- allows tracking of reservations and instantiated pages
 *                    across the pages in a mapping.
52 53 54 55 56 57 58 59 60 61
 *
 * The region data structures are protected by a combination of the mmap_sem
 * and the hugetlb_instantion_mutex.  To access or modify a region the caller
 * must either hold the mmap_sem for write, or the mmap_sem for read and
 * the hugetlb_instantiation mutex:
 *
 * 	down_write(&mm->mmap_sem);
 * or
 * 	down_read(&mm->mmap_sem);
 * 	mutex_lock(&hugetlb_instantiation_mutex);
62 63 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183
 */
struct file_region {
	struct list_head link;
	long from;
	long to;
};

static long region_add(struct list_head *head, long f, long t)
{
	struct file_region *rg, *nrg, *trg;

	/* 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;
	return 0;
}

static long region_chg(struct list_head *head, long f, long t)
{
	struct file_region *rg, *nrg;
	long chg = 0;

	/* 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) {
		nrg = kmalloc(sizeof(*nrg), GFP_KERNEL);
		if (!nrg)
			return -ENOMEM;
		nrg->from = f;
		nrg->to   = f;
		INIT_LIST_HEAD(&nrg->link);
		list_add(&nrg->link, rg->link.prev);

		return t - f;
	}

	/* 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)
			return chg;

		/* We overlap with this area, if it extends futher than
		 * 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;
	}
	return chg;
}

static long region_truncate(struct list_head *head, long end)
{
	struct file_region *rg, *trg;
	long chg = 0;

	/* 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)
		return 0;

	/* 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);
	}
	return chg;
}

184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207
static long region_count(struct list_head *head, long f, long t)
{
	struct file_region *rg;
	long chg = 0;

	/* Locate each segment we overlap with, and count that overlap. */
	list_for_each_entry(rg, head, link) {
		int seg_from;
		int seg_to;

		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;
	}

	return chg;
}

208 209 210 211
/*
 * Convert the address within this vma to the page offset within
 * the mapping, in pagecache page units; huge pages here.
 */
212 213
static pgoff_t vma_hugecache_offset(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
214
{
215 216
	return ((address - vma->vm_start) >> huge_page_shift(h)) +
			(vma->vm_pgoff >> huge_page_order(h));
217 218
}

219 220 221 222 223 224 225
/*
 * 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)
226
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED)
227

228 229 230 231 232 233 234 235 236
/*
 * 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.
237 238 239 240 241 242 243 244 245
 *
 * 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.
246
 */
247 248 249 250 251 252 253 254 255 256 257
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;
}

258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
struct resv_map {
	struct kref refs;
	struct list_head regions;
};

struct resv_map *resv_map_alloc(void)
{
	struct resv_map *resv_map = kmalloc(sizeof(*resv_map), GFP_KERNEL);
	if (!resv_map)
		return NULL;

	kref_init(&resv_map->refs);
	INIT_LIST_HEAD(&resv_map->regions);

	return resv_map;
}

void resv_map_release(struct kref *ref)
{
	struct resv_map *resv_map = container_of(ref, struct resv_map, refs);

	/* Clear out any active regions before we release the map. */
	region_truncate(&resv_map->regions, 0);
	kfree(resv_map);
}

static struct resv_map *vma_resv_map(struct vm_area_struct *vma)
285 286 287
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
	if (!(vma->vm_flags & VM_SHARED))
288 289
		return (struct resv_map *)(get_vma_private_data(vma) &
							~HPAGE_RESV_MASK);
290 291 292
	return 0;
}

293
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map)
294 295 296 297
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
	VM_BUG_ON(vma->vm_flags & VM_SHARED);

298 299
	set_vma_private_data(vma, (get_vma_private_data(vma) &
				HPAGE_RESV_MASK) | (unsigned long)map);
300 301 302 303 304
}

static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
305 306 307
	VM_BUG_ON(vma->vm_flags & VM_SHARED);

	set_vma_private_data(vma, get_vma_private_data(vma) | flags);
308 309 310 311 312
}

static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
313 314

	return (get_vma_private_data(vma) & flag) != 0;
315 316 317
}

/* Decrement the reserved pages in the hugepage pool by one */
318 319
static void decrement_hugepage_resv_vma(struct hstate *h,
			struct vm_area_struct *vma)
320
{
321 322 323
	if (vma->vm_flags & VM_NORESERVE)
		return;

324 325
	if (vma->vm_flags & VM_SHARED) {
		/* Shared mappings always use reserves */
326
		h->resv_huge_pages--;
327
	} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
328 329 330 331
		/*
		 * Only the process that called mmap() has reserves for
		 * private mappings.
		 */
332
		h->resv_huge_pages--;
333 334 335
	}
}

336
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */
337 338 339 340 341 342 343 344
void reset_vma_resv_huge_pages(struct vm_area_struct *vma)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
	if (!(vma->vm_flags & VM_SHARED))
		vma->vm_private_data = (void *)0;
}

/* Returns true if the VMA has associated reserve pages */
345
static int vma_has_reserves(struct vm_area_struct *vma)
346 347
{
	if (vma->vm_flags & VM_SHARED)
348 349 350 351
		return 1;
	if (is_vma_resv_set(vma, HPAGE_RESV_OWNER))
		return 1;
	return 0;
352 353
}

354 355
static void clear_huge_page(struct page *page,
			unsigned long addr, unsigned long sz)
356 357 358 359
{
	int i;

	might_sleep();
360
	for (i = 0; i < sz/PAGE_SIZE; i++) {
361
		cond_resched();
362
		clear_user_highpage(page + i, addr + i * PAGE_SIZE);
363 364 365 366
	}
}

static void copy_huge_page(struct page *dst, struct page *src,
367
			   unsigned long addr, struct vm_area_struct *vma)
368 369
{
	int i;
370
	struct hstate *h = hstate_vma(vma);
371 372

	might_sleep();
373
	for (i = 0; i < pages_per_huge_page(h); i++) {
374
		cond_resched();
375
		copy_user_highpage(dst + i, src + i, addr + i*PAGE_SIZE, vma);
376 377 378
	}
}

379
static void enqueue_huge_page(struct hstate *h, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
380 381
{
	int nid = page_to_nid(page);
382 383 384
	list_add(&page->lru, &h->hugepage_freelists[nid]);
	h->free_huge_pages++;
	h->free_huge_pages_node[nid]++;
Linus Torvalds's avatar
Linus Torvalds committed
385 386
}

387
static struct page *dequeue_huge_page(struct hstate *h)
388 389 390 391 392
{
	int nid;
	struct page *page = NULL;

	for (nid = 0; nid < MAX_NUMNODES; ++nid) {
393 394
		if (!list_empty(&h->hugepage_freelists[nid])) {
			page = list_entry(h->hugepage_freelists[nid].next,
395 396
					  struct page, lru);
			list_del(&page->lru);
397 398
			h->free_huge_pages--;
			h->free_huge_pages_node[nid]--;
399 400 401 402 403 404
			break;
		}
	}
	return page;
}

405 406
static struct page *dequeue_huge_page_vma(struct hstate *h,
				struct vm_area_struct *vma,
407
				unsigned long address, int avoid_reserve)
Linus Torvalds's avatar
Linus Torvalds committed
408
{
409
	int nid;
Linus Torvalds's avatar
Linus Torvalds committed
410
	struct page *page = NULL;
411
	struct mempolicy *mpol;
412
	nodemask_t *nodemask;
413
	struct zonelist *zonelist = huge_zonelist(vma, address,
414
					htlb_alloc_mask, &mpol, &nodemask);
415 416
	struct zone *zone;
	struct zoneref *z;
Linus Torvalds's avatar
Linus Torvalds committed
417

418 419 420 421 422
	/*
	 * 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
	 */
423
	if (!vma_has_reserves(vma) &&
424
			h->free_huge_pages - h->resv_huge_pages == 0)
425 426
		return NULL;

427
	/* If reserves cannot be used, ensure enough pages are in the pool */
428
	if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0)
429 430
		return NULL;

431 432
	for_each_zone_zonelist_nodemask(zone, z, zonelist,
						MAX_NR_ZONES - 1, nodemask) {
433 434
		nid = zone_to_nid(zone);
		if (cpuset_zone_allowed_softwall(zone, htlb_alloc_mask) &&
435 436
		    !list_empty(&h->hugepage_freelists[nid])) {
			page = list_entry(h->hugepage_freelists[nid].next,
437 438
					  struct page, lru);
			list_del(&page->lru);
439 440
			h->free_huge_pages--;
			h->free_huge_pages_node[nid]--;
441 442

			if (!avoid_reserve)
443
				decrement_hugepage_resv_vma(h, vma);
444

Ken Chen's avatar
Ken Chen committed
445
			break;
446
		}
Linus Torvalds's avatar
Linus Torvalds committed
447
	}
448
	mpol_cond_put(mpol);
Linus Torvalds's avatar
Linus Torvalds committed
449 450 451
	return page;
}

452
static void update_and_free_page(struct hstate *h, struct page *page)
453 454
{
	int i;
455 456 457 458

	h->nr_huge_pages--;
	h->nr_huge_pages_node[page_to_nid(page)]--;
	for (i = 0; i < pages_per_huge_page(h); i++) {
459 460 461 462 463 464
		page[i].flags &= ~(1 << PG_locked | 1 << PG_error | 1 << PG_referenced |
				1 << PG_dirty | 1 << PG_active | 1 << PG_reserved |
				1 << PG_private | 1<< PG_writeback);
	}
	set_compound_page_dtor(page, NULL);
	set_page_refcounted(page);
465
	arch_release_hugepage(page);
466
	__free_pages(page, huge_page_order(h));
467 468
}

469 470 471 472 473 474 475 476 477 478 479
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;
}

480 481
static void free_huge_page(struct page *page)
{
482 483 484 485
	/*
	 * Can't pass hstate in here because it is called from the
	 * compound page destructor.
	 */
486
	struct hstate *h = page_hstate(page);
487
	int nid = page_to_nid(page);
488
	struct address_space *mapping;
489

490
	mapping = (struct address_space *) page_private(page);
491
	set_page_private(page, 0);
492
	BUG_ON(page_count(page));
493 494 495
	INIT_LIST_HEAD(&page->lru);

	spin_lock(&hugetlb_lock);
496
	if (h->surplus_huge_pages_node[nid] && huge_page_order(h) < MAX_ORDER) {
497 498 499
		update_and_free_page(h, page);
		h->surplus_huge_pages--;
		h->surplus_huge_pages_node[nid]--;
500
	} else {
501
		enqueue_huge_page(h, page);
502
	}
503
	spin_unlock(&hugetlb_lock);
504
	if (mapping)
505
		hugetlb_put_quota(mapping, 1);
506 507
}

508 509 510 511 512
/*
 * 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.
 */
513
static int adjust_pool_surplus(struct hstate *h, int delta)
514 515 516 517 518 519 520 521 522 523 524 525
{
	static int prev_nid;
	int nid = prev_nid;
	int ret = 0;

	VM_BUG_ON(delta != -1 && delta != 1);
	do {
		nid = next_node(nid, node_online_map);
		if (nid == MAX_NUMNODES)
			nid = first_node(node_online_map);

		/* To shrink on this node, there must be a surplus page */
526
		if (delta < 0 && !h->surplus_huge_pages_node[nid])
527 528
			continue;
		/* Surplus cannot exceed the total number of pages */
529 530
		if (delta > 0 && h->surplus_huge_pages_node[nid] >=
						h->nr_huge_pages_node[nid])
531 532
			continue;

533 534
		h->surplus_huge_pages += delta;
		h->surplus_huge_pages_node[nid] += delta;
535 536 537 538 539 540 541 542
		ret = 1;
		break;
	} while (nid != prev_nid);

	prev_nid = nid;
	return ret;
}

543
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid)
544 545 546
{
	set_compound_page_dtor(page, free_huge_page);
	spin_lock(&hugetlb_lock);
547 548
	h->nr_huge_pages++;
	h->nr_huge_pages_node[nid]++;
549 550 551 552
	spin_unlock(&hugetlb_lock);
	put_page(page); /* free it into the hugepage allocator */
}

553
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid)
Linus Torvalds's avatar
Linus Torvalds committed
554 555
{
	struct page *page;
556

557 558 559
	if (h->order >= MAX_ORDER)
		return NULL;

560
	page = alloc_pages_node(nid,
561 562
		htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE|
						__GFP_REPEAT|__GFP_NOWARN,
563
		huge_page_order(h));
Linus Torvalds's avatar
Linus Torvalds committed
564
	if (page) {
565 566
		if (arch_prepare_hugepage(page)) {
			__free_pages(page, HUGETLB_PAGE_ORDER);
567
			return NULL;
568
		}
569
		prep_new_huge_page(h, page, nid);
Linus Torvalds's avatar
Linus Torvalds committed
570
	}
571 572 573 574

	return page;
}

575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595
/*
 * Use a helper variable to find the next node and then
 * copy it back to hugetlb_next_nid afterwards:
 * otherwise there's a window in which a racer might
 * pass invalid nid MAX_NUMNODES to alloc_pages_node.
 * But we don't need to use a spin_lock here: it really
 * doesn't matter if occasionally a racer chooses the
 * same nid as we do.  Move nid forward in the mask even
 * if we just successfully allocated a hugepage so that
 * the next caller gets hugepages on the next node.
 */
static int hstate_next_node(struct hstate *h)
{
	int next_nid;
	next_nid = next_node(h->hugetlb_next_nid, node_online_map);
	if (next_nid == MAX_NUMNODES)
		next_nid = first_node(node_online_map);
	h->hugetlb_next_nid = next_nid;
	return next_nid;
}

596
static int alloc_fresh_huge_page(struct hstate *h)
597 598 599 600 601 602
{
	struct page *page;
	int start_nid;
	int next_nid;
	int ret = 0;

603
	start_nid = h->hugetlb_next_nid;
604 605

	do {
606
		page = alloc_fresh_huge_page_node(h, h->hugetlb_next_nid);
607 608
		if (page)
			ret = 1;
609
		next_nid = hstate_next_node(h);
610
	} while (!page && h->hugetlb_next_nid != start_nid);
611

612 613 614 615 616
	if (ret)
		count_vm_event(HTLB_BUDDY_PGALLOC);
	else
		count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);

617
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
618 619
}

620 621
static struct page *alloc_buddy_huge_page(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
622 623
{
	struct page *page;
624
	unsigned int nid;
625

626 627 628
	if (h->order >= MAX_ORDER)
		return NULL;

629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652
	/*
	 * 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);
653
	if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) {
654 655 656
		spin_unlock(&hugetlb_lock);
		return NULL;
	} else {
657 658
		h->nr_huge_pages++;
		h->surplus_huge_pages++;
659 660 661
	}
	spin_unlock(&hugetlb_lock);

662 663
	page = alloc_pages(htlb_alloc_mask|__GFP_COMP|
					__GFP_REPEAT|__GFP_NOWARN,
664
					huge_page_order(h));
665 666

	spin_lock(&hugetlb_lock);
667
	if (page) {
668 669 670 671 672 673
		/*
		 * This page is now managed by the hugetlb allocator and has
		 * no users -- drop the buddy allocator's reference.
		 */
		put_page_testzero(page);
		VM_BUG_ON(page_count(page));
674
		nid = page_to_nid(page);
675
		set_compound_page_dtor(page, free_huge_page);
676 677 678
		/*
		 * We incremented the global counters already
		 */
679 680
		h->nr_huge_pages_node[nid]++;
		h->surplus_huge_pages_node[nid]++;
681
		__count_vm_event(HTLB_BUDDY_PGALLOC);
682
	} else {
683 684
		h->nr_huge_pages--;
		h->surplus_huge_pages--;
685
		__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);
686
	}
687
	spin_unlock(&hugetlb_lock);
688 689 690 691

	return page;
}

692 693 694 695
/*
 * Increase the hugetlb pool such that it can accomodate a reservation
 * of size 'delta'.
 */
696
static int gather_surplus_pages(struct hstate *h, int delta)
697 698 699 700 701 702
{
	struct list_head surplus_list;
	struct page *page, *tmp;
	int ret, i;
	int needed, allocated;

703
	needed = (h->resv_huge_pages + delta) - h->free_huge_pages;
704
	if (needed <= 0) {
705
		h->resv_huge_pages += delta;
706
		return 0;
707
	}
708 709 710 711 712 713 714 715

	allocated = 0;
	INIT_LIST_HEAD(&surplus_list);

	ret = -ENOMEM;
retry:
	spin_unlock(&hugetlb_lock);
	for (i = 0; i < needed; i++) {
716
		page = alloc_buddy_huge_page(h, NULL, 0);
717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736
		if (!page) {
			/*
			 * We were not able to allocate enough pages to
			 * satisfy the entire reservation so we free what
			 * we've allocated so far.
			 */
			spin_lock(&hugetlb_lock);
			needed = 0;
			goto free;
		}

		list_add(&page->lru, &surplus_list);
	}
	allocated += needed;

	/*
	 * 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);
737 738
	needed = (h->resv_huge_pages + delta) -
			(h->free_huge_pages + allocated);
739 740 741 742 743 744 745
	if (needed > 0)
		goto retry;

	/*
	 * The surplus_list now contains _at_least_ the number of extra pages
	 * needed to accomodate the reservation.  Add the appropriate number
	 * of pages to the hugetlb pool and free the extras back to the buddy
746 747 748
	 * 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.
749 750
	 */
	needed += allocated;
751
	h->resv_huge_pages += delta;
752 753
	ret = 0;
free:
754
	/* Free the needed pages to the hugetlb pool */
755
	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
756 757
		if ((--needed) < 0)
			break;
758
		list_del(&page->lru);
759
		enqueue_huge_page(h, page);
760 761 762 763 764 765 766
	}

	/* Free unnecessary surplus pages to the buddy allocator */
	if (!list_empty(&surplus_list)) {
		spin_unlock(&hugetlb_lock);
		list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
			list_del(&page->lru);
767
			/*
768 769 770
			 * The page has a reference count of zero already, so
			 * call free_huge_page directly instead of using
			 * put_page.  This must be done with hugetlb_lock
771 772 773
			 * unlocked which is safe because free_huge_page takes
			 * hugetlb_lock before deciding how to free the page.
			 */
774
			free_huge_page(page);
775
		}
776
		spin_lock(&hugetlb_lock);
777 778 779 780 781 782 783 784 785 786
	}

	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.
 */
787 788
static void return_unused_surplus_pages(struct hstate *h,
					unsigned long unused_resv_pages)
789 790 791 792 793
{
	static int nid = -1;
	struct page *page;
	unsigned long nr_pages;

794 795 796 797 798 799 800 801
	/*
	 * We want to release as many surplus pages as possible, spread
	 * evenly across all nodes. Iterate across all nodes until we
	 * can no longer free unreserved surplus pages. This occurs when
	 * the nodes with surplus pages have no free pages.
	 */
	unsigned long remaining_iterations = num_online_nodes();

802
	/* Uncommit the reservation */
803
	h->resv_huge_pages -= unused_resv_pages;
804

805 806 807 808
	/* Cannot return gigantic pages currently */
	if (h->order >= MAX_ORDER)
		return;

809
	nr_pages = min(unused_resv_pages, h->surplus_huge_pages);
810

811
	while (remaining_iterations-- && nr_pages) {
812 813 814 815
		nid = next_node(nid, node_online_map);
		if (nid == MAX_NUMNODES)
			nid = first_node(node_online_map);

816
		if (!h->surplus_huge_pages_node[nid])
817 818
			continue;

819 820
		if (!list_empty(&h->hugepage_freelists[nid])) {
			page = list_entry(h->hugepage_freelists[nid].next,
821 822
					  struct page, lru);
			list_del(&page->lru);
823 824 825 826 827
			update_and_free_page(h, page);
			h->free_huge_pages--;
			h->free_huge_pages_node[nid]--;
			h->surplus_huge_pages--;
			h->surplus_huge_pages_node[nid]--;
828
			nr_pages--;
829
			remaining_iterations = num_online_nodes();
830 831 832 833
		}
	}
}

834 835 836 837 838 839 840 841 842
/*
 * Determine if the huge page at addr within the vma has an associated
 * reservation.  Where it does not we will need to logically increase
 * reservation and actually increase quota 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 quota'd allocated
 * an instantiated the change should be committed via vma_commit_reservation.
 * No action is required on failure.
 */
843 844
static int vma_needs_reservation(struct hstate *h,
			struct vm_area_struct *vma, unsigned long addr)
845 846 847 848 849
{
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;

	if (vma->vm_flags & VM_SHARED) {
850
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
851 852 853
		return region_chg(&inode->i_mapping->private_list,
							idx, idx + 1);

854 855
	} else if (!is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
		return 1;
856

857 858
	} else  {
		int err;
859
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
860 861 862 863 864 865 866
		struct resv_map *reservations = vma_resv_map(vma);

		err = region_chg(&reservations->regions, idx, idx + 1);
		if (err < 0)
			return err;
		return 0;
	}
867
}
868 869
static void vma_commit_reservation(struct hstate *h,
			struct vm_area_struct *vma, unsigned long addr)
870 871 872 873 874
{
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;

	if (vma->vm_flags & VM_SHARED) {
875
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
876
		region_add(&inode->i_mapping->private_list, idx, idx + 1);
877 878

	} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
879
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
880 881 882 883
		struct resv_map *reservations = vma_resv_map(vma);

		/* Mark this page used in the map. */
		region_add(&reservations->regions, idx, idx + 1);
884 885 886
	}
}

887
static struct page *alloc_huge_page(struct vm_area_struct *vma,
888
				    unsigned long addr, int avoid_reserve)
Linus Torvalds's avatar
Linus Torvalds committed
889
{
890
	struct hstate *h = hstate_vma(vma);
891
	struct page *page;
892 893
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;
894
	unsigned int chg;
895 896 897 898 899

	/*
	 * Processes that did not create the mapping will have no reserves and
	 * will not have accounted against quota. Check that the quota can be
	 * made before satisfying the allocation
900 901
	 * MAP_NORESERVE mappings may also need pages and quota allocated
	 * if no reserve mapping overlaps.
902
	 */
903
	chg = vma_needs_reservation(h, vma, addr);
904 905 906
	if (chg < 0)
		return ERR_PTR(chg);
	if (chg)
907 908
		if (hugetlb_get_quota(inode->i_mapping, chg))
			return ERR_PTR(-ENOSPC);
Linus Torvalds's avatar
Linus Torvalds committed
909 910

	spin_lock(&hugetlb_lock);
911
	page = dequeue_huge_page_vma(h, vma, addr, avoid_reserve);
Linus Torvalds's avatar
Linus Torvalds committed
912
	spin_unlock(&hugetlb_lock);
913

Ken Chen's avatar
Ken Chen committed
914
	if (!page) {
915
		page = alloc_buddy_huge_page(h, vma, addr);
Ken Chen's avatar
Ken Chen committed
916
		if (!page) {
917
			hugetlb_put_quota(inode->i_mapping, chg);
Ken Chen's avatar
Ken Chen committed
918 919 920
			return ERR_PTR(-VM_FAULT_OOM);
		}
	}
921

922 923
	set_page_refcounted(page);
	set_page_private(page, (unsigned long) mapping);
924

925
	vma_commit_reservation(h, vma, addr);
926

927
	return page;
928 929
}

930
__attribute__((weak)) int alloc_bootmem_huge_page(struct hstate *h)
931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979
{
	struct huge_bootmem_page *m;
	int nr_nodes = nodes_weight(node_online_map);

	while (nr_nodes) {
		void *addr;

		addr = __alloc_bootmem_node_nopanic(
				NODE_DATA(h->hugetlb_next_nid),
				huge_page_size(h), huge_page_size(h), 0);

		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;
			if (m)
				goto found;
		}
		hstate_next_node(h);
		nr_nodes--;
	}
	return 0;

found:
	BUG_ON((unsigned long)virt_to_phys(m) & (huge_page_size(h) - 1));
	/* 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;
}

/* 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 page *page = virt_to_page(m);
		struct hstate *h = m->hstate;
		__ClearPageReserved(page);
		WARN_ON(page_count(page) != 1);
		prep_compound_page(page, h->order);
		prep_new_huge_page(h, page, page_to_nid(page));
	}
}

980
static void __init hugetlb_hstate_alloc_pages(struct hstate *h)
Linus Torvalds's avatar
Linus Torvalds committed
981 982
{
	unsigned long i;
983

984
	for (i = 0; i < h->max_huge_pages; ++i) {
985 986 987 988
		if (h->order >= MAX_ORDER) {
			if (!alloc_bootmem_huge_page(h))
				break;
		} else if (!alloc_fresh_huge_page(h))
Linus Torvalds's avatar
Linus Torvalds committed
989 990
			break;
	}
991
	h->max_huge_pages = i;
992 993 994 995 996 997 998
}

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

	for_each_hstate(h) {
999 1000 1001
		/* oversize hugepages were init'ed in early boot */
		if (h->order < MAX_ORDER)
			hugetlb_hstate_alloc_pages(h);
1002 1003 1004
	}
}

Andi Kleen's avatar
Andi Kleen committed
1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015
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;
}

1016 1017 1018 1019 1020
static void __init report_hugepages(void)
{
	struct hstate *h;

	for_each_hstate(h) {
Andi Kleen's avatar
Andi Kleen committed
1021 1022 1023 1024 1025
		char buf[32];
		printk(KERN_INFO "HugeTLB registered %s page size, "
				 "pre-allocated %ld pages\n",
			memfmt(buf, huge_page_size(h)),
			h->free_huge_pages);
1026 1027 1028
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
1029 1030
#ifdef CONFIG_SYSCTL
#ifdef CONFIG_HIGHMEM
1031
static void try_to_free_low(struct hstate *h, unsigned long count)
Linus Torvalds's avatar
Linus Torvalds committed
1032
{
1033 1034
	int i;

1035 1036 1037
	if (h->order >= MAX_ORDER)
		return;

Linus Torvalds's avatar
Linus Torvalds committed
1038 1039
	for (i = 0; i < MAX_NUMNODES; ++i) {
		struct page *page, *next;
1040 1041 1042
		struct list_head *freel = &h->hugepage_freelists[i];
		list_for_each_entry_safe(page, next, freel, lru) {
			if (count >= h->nr_huge_pages)
1043
				return;
Linus Torvalds's avatar
Linus Torvalds committed
1044 1045 1046
			if (PageHighMem(page))
				continue;
			list_del(&page->lru);
1047
			update_and_free_page(h, page);
1048 1049
			h->free_huge_pages--;
			h->free_huge_pages_node[page_to_nid(page)]--;
Linus Torvalds's avatar
Linus Torvalds committed
1050 1051 1052 1053
		}
	}
}
#else
1054
static inline void try_to_free_low(struct hstate *h, unsigned long count)
Linus Torvalds's avatar
Linus Torvalds committed
1055 1056 1057 1058
{
}
#endif

1059
#define persistent_huge_pages(h) (h->nr_huge_pages - h->surplus_huge_pages)
1060
static unsigned long set_max_huge_pages(struct hstate *h, unsigned long count)
Linus Torvalds's avatar
Linus Torvalds committed
1061
{
1062
	unsigned long min_count, ret;
Linus Torvalds's avatar
Linus Torvalds committed
1063

1064 1065 1066
	if (h->order >= MAX_ORDER)
		return h->max_huge_pages;

1067 1068 1069 1070
	/*
	 * Increase the pool size
	 * First take pages out of surplus state.  Then make up the
	 * remaining difference by allocating fresh huge pages.
1071 1072 1073 1074 1075 1076
	 *
	 * 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.
1077
	 */
Linus Torvalds's avatar
Linus Torvalds committed
1078
	spin_lock(&hugetlb_lock);
1079 1080
	while (h->surplus_huge_pages && count > persistent_huge_pages(h)) {
		if (!adjust_pool_surplus(h, -1))
1081 1082 1083
			break;
	}

1084
	while (count > persistent_huge_pages(h)) {
1085 1086 1087 1088 1089 1090
		/*
		 * 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);
1091
		ret = alloc_fresh_huge_page(h);
1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
		spin_lock(&hugetlb_lock);
		if (!ret)
			goto out;

	}

	/*
	 * 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.
1104 1105 1106 1107 1108 1109 1110 1111
	 *
	 * 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.
1112
	 */
1113
	min_count = h->resv_huge_pages + h->nr_huge_pages - h->free_huge_pages;
1114
	min_count = max(count, min_count);
1115 1116 1117
	try_to_free_low(h, min_count);
	while (min_count < persistent_huge_pages(h)) {
		struct page *page = dequeue_huge_page(h);
Linus Torvalds's avatar
Linus Torvalds committed
1118 1119
		if (!page)
			break;
1120
		update_and_free_page(h, page);
Linus Torvalds's avatar
Linus Torvalds committed
1121
	}
1122 1123
	while (count < persistent_huge_pages(h)) {