memory.c 113.4 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
/*
 *  linux/mm/memory.c
 *
 *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 */

/*
 * demand-loading started 01.12.91 - seems it is high on the list of
 * things wanted, and it should be easy to implement. - Linus
 */

/*
 * Ok, demand-loading was easy, shared pages a little bit tricker. Shared
 * pages started 02.12.91, seems to work. - Linus.
 *
 * Tested sharing by executing about 30 /bin/sh: under the old kernel it
 * would have taken more than the 6M I have free, but it worked well as
 * far as I could see.
 *
 * Also corrected some "invalidate()"s - I wasn't doing enough of them.
 */

/*
 * Real VM (paging to/from disk) started 18.12.91. Much more work and
 * thought has to go into this. Oh, well..
 * 19.12.91  -  works, somewhat. Sometimes I get faults, don't know why.
 *		Found it. Everything seems to work now.
 * 20.12.91  -  Ok, making the swap-device changeable like the root.
 */

/*
 * 05.04.94  -  Multi-page memory management added for v1.1.
 * 		Idea by Alex Bligh (alex@cconcepts.co.uk)
 *
 * 16.07.99  -  Support of BIGMEM added by Gerhard Wichert, Siemens AG
 *		(Gerhard.Wichert@pdb.siemens.de)
 *
 * Aug/Sep 2004 Changed to four level page tables (Andi Kleen)
 */

#include <linux/kernel_stat.h>
#include <linux/mm.h>
#include <linux/hugetlb.h>
#include <linux/mman.h>
#include <linux/swap.h>
#include <linux/highmem.h>
#include <linux/pagemap.h>
H
Hugh Dickins 已提交
48
#include <linux/ksm.h>
L
Linus Torvalds 已提交
49
#include <linux/rmap.h>
50
#include <linux/export.h>
51
#include <linux/delayacct.h>
L
Linus Torvalds 已提交
52
#include <linux/init.h>
P
Peter Zijlstra 已提交
53
#include <linux/writeback.h>
54
#include <linux/memcontrol.h>
A
Andrea Arcangeli 已提交
55
#include <linux/mmu_notifier.h>
56 57 58
#include <linux/kallsyms.h>
#include <linux/swapops.h>
#include <linux/elf.h>
59
#include <linux/gfp.h>
60
#include <linux/migrate.h>
A
Andy Shevchenko 已提交
61
#include <linux/string.h>
L
Linus Torvalds 已提交
62

A
Alexey Dobriyan 已提交
63
#include <asm/io.h>
L
Linus Torvalds 已提交
64 65 66 67 68 69
#include <asm/pgalloc.h>
#include <asm/uaccess.h>
#include <asm/tlb.h>
#include <asm/tlbflush.h>
#include <asm/pgtable.h>

70 71
#include "internal.h"

A
Andy Whitcroft 已提交
72
#ifndef CONFIG_NEED_MULTIPLE_NODES
L
Linus Torvalds 已提交
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
/* use the per-pgdat data instead for discontigmem - mbligh */
unsigned long max_mapnr;
struct page *mem_map;

EXPORT_SYMBOL(max_mapnr);
EXPORT_SYMBOL(mem_map);
#endif

unsigned long num_physpages;
/*
 * A number of key systems in x86 including ioremap() rely on the assumption
 * that high_memory defines the upper bound on direct map memory, then end
 * of ZONE_NORMAL.  Under CONFIG_DISCONTIG this means that max_low_pfn and
 * highstart_pfn must be the same; there must be no gap between ZONE_NORMAL
 * and ZONE_HIGHMEM.
 */
void * high_memory;

EXPORT_SYMBOL(num_physpages);
EXPORT_SYMBOL(high_memory);

94 95 96 97 98 99 100 101 102 103 104 105
/*
 * Randomize the address space (stacks, mmaps, brk, etc.).
 *
 * ( When CONFIG_COMPAT_BRK=y we exclude brk from randomization,
 *   as ancient (libc5 based) binaries can segfault. )
 */
int randomize_va_space __read_mostly =
#ifdef CONFIG_COMPAT_BRK
					1;
#else
					2;
#endif
106 107 108 109

static int __init disable_randmaps(char *s)
{
	randomize_va_space = 0;
110
	return 1;
111 112 113
}
__setup("norandmaps", disable_randmaps);

H
Hugh Dickins 已提交
114
unsigned long zero_pfn __read_mostly;
H
Hugh Dickins 已提交
115
unsigned long highest_memmap_pfn __read_mostly;
H
Hugh Dickins 已提交
116 117 118 119 120 121 122 123 124 125

/*
 * CONFIG_MMU architectures set up ZERO_PAGE in their paging_init()
 */
static int __init init_zero_pfn(void)
{
	zero_pfn = page_to_pfn(ZERO_PAGE(0));
	return 0;
}
core_initcall(init_zero_pfn);
126

K
KAMEZAWA Hiroyuki 已提交
127

128 129
#if defined(SPLIT_RSS_COUNTING)

130
void sync_mm_rss(struct mm_struct *mm)
131 132 133 134
{
	int i;

	for (i = 0; i < NR_MM_COUNTERS; i++) {
135 136 137
		if (current->rss_stat.count[i]) {
			add_mm_counter(mm, i, current->rss_stat.count[i]);
			current->rss_stat.count[i] = 0;
138 139
		}
	}
140
	current->rss_stat.events = 0;
141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161
}

static void add_mm_counter_fast(struct mm_struct *mm, int member, int val)
{
	struct task_struct *task = current;

	if (likely(task->mm == mm))
		task->rss_stat.count[member] += val;
	else
		add_mm_counter(mm, member, val);
}
#define inc_mm_counter_fast(mm, member) add_mm_counter_fast(mm, member, 1)
#define dec_mm_counter_fast(mm, member) add_mm_counter_fast(mm, member, -1)

/* sync counter once per 64 page faults */
#define TASK_RSS_EVENTS_THRESH	(64)
static void check_sync_rss_stat(struct task_struct *task)
{
	if (unlikely(task != current))
		return;
	if (unlikely(task->rss_stat.events++ > TASK_RSS_EVENTS_THRESH))
162
		sync_mm_rss(task->mm);
163
}
164
#else /* SPLIT_RSS_COUNTING */
165 166 167 168 169 170 171 172

#define inc_mm_counter_fast(mm, member) inc_mm_counter(mm, member)
#define dec_mm_counter_fast(mm, member) dec_mm_counter(mm, member)

static void check_sync_rss_stat(struct task_struct *task)
{
}

173 174 175 176 177 178 179 180 181 182 183 184 185 186
#endif /* SPLIT_RSS_COUNTING */

#ifdef HAVE_GENERIC_MMU_GATHER

static int tlb_next_batch(struct mmu_gather *tlb)
{
	struct mmu_gather_batch *batch;

	batch = tlb->active;
	if (batch->next) {
		tlb->active = batch->next;
		return 1;
	}

187 188 189
	if (tlb->batch_count == MAX_GATHER_BATCH_COUNT)
		return 0;

190 191 192 193
	batch = (void *)__get_free_pages(GFP_NOWAIT | __GFP_NOWARN, 0);
	if (!batch)
		return 0;

194
	tlb->batch_count++;
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214
	batch->next = NULL;
	batch->nr   = 0;
	batch->max  = MAX_GATHER_BATCH;

	tlb->active->next = batch;
	tlb->active = batch;

	return 1;
}

/* tlb_gather_mmu
 *	Called to initialize an (on-stack) mmu_gather structure for page-table
 *	tear-down from @mm. The @fullmm argument is used when @mm is without
 *	users and we're going to destroy the full address space (exit/execve).
 */
void tlb_gather_mmu(struct mmu_gather *tlb, struct mm_struct *mm, bool fullmm)
{
	tlb->mm = mm;

	tlb->fullmm     = fullmm;
215 216
	tlb->start	= -1UL;
	tlb->end	= 0;
217 218 219 220 221 222
	tlb->need_flush = 0;
	tlb->fast_mode  = (num_possible_cpus() == 1);
	tlb->local.next = NULL;
	tlb->local.nr   = 0;
	tlb->local.max  = ARRAY_SIZE(tlb->__pages);
	tlb->active     = &tlb->local;
223
	tlb->batch_count = 0;
224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239

#ifdef CONFIG_HAVE_RCU_TABLE_FREE
	tlb->batch = NULL;
#endif
}

void tlb_flush_mmu(struct mmu_gather *tlb)
{
	struct mmu_gather_batch *batch;

	if (!tlb->need_flush)
		return;
	tlb->need_flush = 0;
	tlb_flush(tlb);
#ifdef CONFIG_HAVE_RCU_TABLE_FREE
	tlb_table_flush(tlb);
240 241
#endif

242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259
	if (tlb_fast_mode(tlb))
		return;

	for (batch = &tlb->local; batch; batch = batch->next) {
		free_pages_and_swap_cache(batch->pages, batch->nr);
		batch->nr = 0;
	}
	tlb->active = &tlb->local;
}

/* tlb_finish_mmu
 *	Called at the end of the shootdown operation to free up any resources
 *	that were required.
 */
void tlb_finish_mmu(struct mmu_gather *tlb, unsigned long start, unsigned long end)
{
	struct mmu_gather_batch *batch, *next;

260 261
	tlb->start = start;
	tlb->end   = end;
262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283
	tlb_flush_mmu(tlb);

	/* keep the page table cache within bounds */
	check_pgt_cache();

	for (batch = tlb->local.next; batch; batch = next) {
		next = batch->next;
		free_pages((unsigned long)batch, 0);
	}
	tlb->local.next = NULL;
}

/* __tlb_remove_page
 *	Must perform the equivalent to __free_pte(pte_get_and_clear(ptep)), while
 *	handling the additional races in SMP caused by other CPUs caching valid
 *	mappings in their TLBs. Returns the number of free page slots left.
 *	When out of page slots we must call tlb_flush_mmu().
 */
int __tlb_remove_page(struct mmu_gather *tlb, struct page *page)
{
	struct mmu_gather_batch *batch;

S
Shaohua Li 已提交
284
	VM_BUG_ON(!tlb->need_flush);
285 286 287 288 289 290 291 292 293 294 295

	if (tlb_fast_mode(tlb)) {
		free_page_and_swap_cache(page);
		return 1; /* avoid calling tlb_flush_mmu() */
	}

	batch = tlb->active;
	batch->pages[batch->nr++] = page;
	if (batch->nr == batch->max) {
		if (!tlb_next_batch(tlb))
			return 0;
296
		batch = tlb->active;
297 298 299 300 301 302 303 304
	}
	VM_BUG_ON(batch->nr > batch->max);

	return batch->max - batch->nr;
}

#endif /* HAVE_GENERIC_MMU_GATHER */

305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379
#ifdef CONFIG_HAVE_RCU_TABLE_FREE

/*
 * See the comment near struct mmu_table_batch.
 */

static void tlb_remove_table_smp_sync(void *arg)
{
	/* Simply deliver the interrupt */
}

static void tlb_remove_table_one(void *table)
{
	/*
	 * This isn't an RCU grace period and hence the page-tables cannot be
	 * assumed to be actually RCU-freed.
	 *
	 * It is however sufficient for software page-table walkers that rely on
	 * IRQ disabling. See the comment near struct mmu_table_batch.
	 */
	smp_call_function(tlb_remove_table_smp_sync, NULL, 1);
	__tlb_remove_table(table);
}

static void tlb_remove_table_rcu(struct rcu_head *head)
{
	struct mmu_table_batch *batch;
	int i;

	batch = container_of(head, struct mmu_table_batch, rcu);

	for (i = 0; i < batch->nr; i++)
		__tlb_remove_table(batch->tables[i]);

	free_page((unsigned long)batch);
}

void tlb_table_flush(struct mmu_gather *tlb)
{
	struct mmu_table_batch **batch = &tlb->batch;

	if (*batch) {
		call_rcu_sched(&(*batch)->rcu, tlb_remove_table_rcu);
		*batch = NULL;
	}
}

void tlb_remove_table(struct mmu_gather *tlb, void *table)
{
	struct mmu_table_batch **batch = &tlb->batch;

	tlb->need_flush = 1;

	/*
	 * When there's less then two users of this mm there cannot be a
	 * concurrent page-table walk.
	 */
	if (atomic_read(&tlb->mm->mm_users) < 2) {
		__tlb_remove_table(table);
		return;
	}

	if (*batch == NULL) {
		*batch = (struct mmu_table_batch *)__get_free_page(GFP_NOWAIT | __GFP_NOWARN);
		if (*batch == NULL) {
			tlb_remove_table_one(table);
			return;
		}
		(*batch)->nr = 0;
	}
	(*batch)->tables[(*batch)->nr++] = table;
	if ((*batch)->nr == MAX_TABLE_BATCH)
		tlb_table_flush(tlb);
}

380
#endif /* CONFIG_HAVE_RCU_TABLE_FREE */
381

L
Linus Torvalds 已提交
382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409
/*
 * If a p?d_bad entry is found while walking page tables, report
 * the error, before resetting entry to p?d_none.  Usually (but
 * very seldom) called out from the p?d_none_or_clear_bad macros.
 */

void pgd_clear_bad(pgd_t *pgd)
{
	pgd_ERROR(*pgd);
	pgd_clear(pgd);
}

void pud_clear_bad(pud_t *pud)
{
	pud_ERROR(*pud);
	pud_clear(pud);
}

void pmd_clear_bad(pmd_t *pmd)
{
	pmd_ERROR(*pmd);
	pmd_clear(pmd);
}

/*
 * Note: this doesn't free the actual pages themselves. That
 * has been handled earlier when unmapping all the memory regions.
 */
410 411
static void free_pte_range(struct mmu_gather *tlb, pmd_t *pmd,
			   unsigned long addr)
L
Linus Torvalds 已提交
412
{
413
	pgtable_t token = pmd_pgtable(*pmd);
414
	pmd_clear(pmd);
415
	pte_free_tlb(tlb, token, addr);
416
	tlb->mm->nr_ptes--;
L
Linus Torvalds 已提交
417 418
}

419 420 421
static inline void free_pmd_range(struct mmu_gather *tlb, pud_t *pud,
				unsigned long addr, unsigned long end,
				unsigned long floor, unsigned long ceiling)
L
Linus Torvalds 已提交
422 423 424
{
	pmd_t *pmd;
	unsigned long next;
425
	unsigned long start;
L
Linus Torvalds 已提交
426

427
	start = addr;
L
Linus Torvalds 已提交
428 429 430 431 432
	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
433
		free_pte_range(tlb, pmd, addr);
L
Linus Torvalds 已提交
434 435
	} while (pmd++, addr = next, addr != end);

436 437 438 439 440 441 442
	start &= PUD_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PUD_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
443
	}
444 445 446 447 448
	if (end - 1 > ceiling - 1)
		return;

	pmd = pmd_offset(pud, start);
	pud_clear(pud);
449
	pmd_free_tlb(tlb, pmd, start);
L
Linus Torvalds 已提交
450 451
}

452 453 454
static inline void free_pud_range(struct mmu_gather *tlb, pgd_t *pgd,
				unsigned long addr, unsigned long end,
				unsigned long floor, unsigned long ceiling)
L
Linus Torvalds 已提交
455 456 457
{
	pud_t *pud;
	unsigned long next;
458
	unsigned long start;
L
Linus Torvalds 已提交
459

460
	start = addr;
L
Linus Torvalds 已提交
461 462 463 464 465
	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
466
		free_pmd_range(tlb, pud, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
467 468
	} while (pud++, addr = next, addr != end);

469 470 471 472 473 474 475
	start &= PGDIR_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PGDIR_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
476
	}
477 478 479 480 481
	if (end - 1 > ceiling - 1)
		return;

	pud = pud_offset(pgd, start);
	pgd_clear(pgd);
482
	pud_free_tlb(tlb, pud, start);
L
Linus Torvalds 已提交
483 484 485
}

/*
486 487
 * This function frees user-level page tables of a process.
 *
L
Linus Torvalds 已提交
488 489
 * Must be called with pagetable lock held.
 */
490
void free_pgd_range(struct mmu_gather *tlb,
491 492
			unsigned long addr, unsigned long end,
			unsigned long floor, unsigned long ceiling)
L
Linus Torvalds 已提交
493 494 495
{
	pgd_t *pgd;
	unsigned long next;
496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521

	/*
	 * The next few lines have given us lots of grief...
	 *
	 * Why are we testing PMD* at this top level?  Because often
	 * there will be no work to do at all, and we'd prefer not to
	 * go all the way down to the bottom just to discover that.
	 *
	 * Why all these "- 1"s?  Because 0 represents both the bottom
	 * of the address space and the top of it (using -1 for the
	 * top wouldn't help much: the masks would do the wrong thing).
	 * The rule is that addr 0 and floor 0 refer to the bottom of
	 * the address space, but end 0 and ceiling 0 refer to the top
	 * Comparisons need to use "end - 1" and "ceiling - 1" (though
	 * that end 0 case should be mythical).
	 *
	 * Wherever addr is brought up or ceiling brought down, we must
	 * be careful to reject "the opposite 0" before it confuses the
	 * subsequent tests.  But what about where end is brought down
	 * by PMD_SIZE below? no, end can't go down to 0 there.
	 *
	 * Whereas we round start (addr) and ceiling down, by different
	 * masks at different levels, in order to test whether a table
	 * now has no other vmas using it, so can be freed, we don't
	 * bother to round floor or end up - the tests don't need that.
	 */
L
Linus Torvalds 已提交
522

523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538
	addr &= PMD_MASK;
	if (addr < floor) {
		addr += PMD_SIZE;
		if (!addr)
			return;
	}
	if (ceiling) {
		ceiling &= PMD_MASK;
		if (!ceiling)
			return;
	}
	if (end - 1 > ceiling - 1)
		end -= PMD_SIZE;
	if (addr > end - 1)
		return;

539
	pgd = pgd_offset(tlb->mm, addr);
L
Linus Torvalds 已提交
540 541 542 543
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
544
		free_pud_range(tlb, pgd, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
545
	} while (pgd++, addr = next, addr != end);
546 547
}

548
void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *vma,
549
		unsigned long floor, unsigned long ceiling)
550 551 552 553 554
{
	while (vma) {
		struct vm_area_struct *next = vma->vm_next;
		unsigned long addr = vma->vm_start;

555
		/*
N
npiggin@suse.de 已提交
556 557
		 * Hide vma from rmap and truncate_pagecache before freeing
		 * pgtables
558
		 */
559
		unlink_anon_vmas(vma);
560 561
		unlink_file_vma(vma);

562
		if (is_vm_hugetlb_page(vma)) {
563
			hugetlb_free_pgd_range(tlb, addr, vma->vm_end,
564
				floor, next? next->vm_start: ceiling);
565 566 567 568 569
		} else {
			/*
			 * Optimization: gather nearby vmas into one call down
			 */
			while (next && next->vm_start <= vma->vm_end + PMD_SIZE
570
			       && !is_vm_hugetlb_page(next)) {
571 572
				vma = next;
				next = vma->vm_next;
573
				unlink_anon_vmas(vma);
574
				unlink_file_vma(vma);
575 576 577 578
			}
			free_pgd_range(tlb, addr, vma->vm_end,
				floor, next? next->vm_start: ceiling);
		}
579 580
		vma = next;
	}
L
Linus Torvalds 已提交
581 582
}

583 584
int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma,
		pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
585
{
586
	pgtable_t new = pte_alloc_one(mm, address);
587
	int wait_split_huge_page;
588 589 590
	if (!new)
		return -ENOMEM;

591 592 593 594 595 596 597 598 599 600 601 602 603 604 605
	/*
	 * Ensure all pte setup (eg. pte page lock and page clearing) are
	 * visible before the pte is made visible to other CPUs by being
	 * put into page tables.
	 *
	 * The other side of the story is the pointer chasing in the page
	 * table walking code (when walking the page table without locking;
	 * ie. most of the time). Fortunately, these data accesses consist
	 * of a chain of data-dependent loads, meaning most CPUs (alpha
	 * being the notable exception) will already guarantee loads are
	 * seen in-order. See the alpha page table accessors for the
	 * smp_read_barrier_depends() barriers in page table walking code.
	 */
	smp_wmb(); /* Could be smp_wmb__xxx(before|after)_spin_lock */

H
Hugh Dickins 已提交
606
	spin_lock(&mm->page_table_lock);
607 608
	wait_split_huge_page = 0;
	if (likely(pmd_none(*pmd))) {	/* Has another populated it ? */
L
Linus Torvalds 已提交
609 610
		mm->nr_ptes++;
		pmd_populate(mm, pmd, new);
611
		new = NULL;
612 613
	} else if (unlikely(pmd_trans_splitting(*pmd)))
		wait_split_huge_page = 1;
H
Hugh Dickins 已提交
614
	spin_unlock(&mm->page_table_lock);
615 616
	if (new)
		pte_free(mm, new);
617 618
	if (wait_split_huge_page)
		wait_split_huge_page(vma->anon_vma, pmd);
619
	return 0;
L
Linus Torvalds 已提交
620 621
}

622
int __pte_alloc_kernel(pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
623
{
624 625 626 627
	pte_t *new = pte_alloc_one_kernel(&init_mm, address);
	if (!new)
		return -ENOMEM;

628 629
	smp_wmb(); /* See comment in __pte_alloc */

630
	spin_lock(&init_mm.page_table_lock);
631
	if (likely(pmd_none(*pmd))) {	/* Has another populated it ? */
632
		pmd_populate_kernel(&init_mm, pmd, new);
633
		new = NULL;
634 635
	} else
		VM_BUG_ON(pmd_trans_splitting(*pmd));
636
	spin_unlock(&init_mm.page_table_lock);
637 638
	if (new)
		pte_free_kernel(&init_mm, new);
639
	return 0;
L
Linus Torvalds 已提交
640 641
}

K
KAMEZAWA Hiroyuki 已提交
642 643 644 645 646 647
static inline void init_rss_vec(int *rss)
{
	memset(rss, 0, sizeof(int) * NR_MM_COUNTERS);
}

static inline void add_mm_rss_vec(struct mm_struct *mm, int *rss)
648
{
K
KAMEZAWA Hiroyuki 已提交
649 650
	int i;

651
	if (current->mm == mm)
652
		sync_mm_rss(mm);
K
KAMEZAWA Hiroyuki 已提交
653 654 655
	for (i = 0; i < NR_MM_COUNTERS; i++)
		if (rss[i])
			add_mm_counter(mm, i, rss[i]);
656 657
}

N
Nick Piggin 已提交
658
/*
659 660 661
 * This function is called to print an error when a bad pte
 * is found. For example, we might have a PFN-mapped pte in
 * a region that doesn't allow it.
N
Nick Piggin 已提交
662 663 664
 *
 * The calling function must still handle the error.
 */
665 666
static void print_bad_pte(struct vm_area_struct *vma, unsigned long addr,
			  pte_t pte, struct page *page)
N
Nick Piggin 已提交
667
{
668 669 670 671 672
	pgd_t *pgd = pgd_offset(vma->vm_mm, addr);
	pud_t *pud = pud_offset(pgd, addr);
	pmd_t *pmd = pmd_offset(pud, addr);
	struct address_space *mapping;
	pgoff_t index;
673 674 675 676 677 678 679 680 681 682 683 684 685 686
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

	/*
	 * Allow a burst of 60 reports, then keep quiet for that minute;
	 * or allow a steady drip of one report per second.
	 */
	if (nr_shown == 60) {
		if (time_before(jiffies, resume)) {
			nr_unshown++;
			return;
		}
		if (nr_unshown) {
687 688
			printk(KERN_ALERT
				"BUG: Bad page map: %lu messages suppressed\n",
689 690 691 692 693 694 695
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;
696 697 698 699

	mapping = vma->vm_file ? vma->vm_file->f_mapping : NULL;
	index = linear_page_index(vma, addr);

700 701
	printk(KERN_ALERT
		"BUG: Bad page map in process %s  pte:%08llx pmd:%08llx\n",
702 703
		current->comm,
		(long long)pte_val(pte), (long long)pmd_val(*pmd));
704 705
	if (page)
		dump_page(page);
706
	printk(KERN_ALERT
707 708 709 710 711 712
		"addr:%p vm_flags:%08lx anon_vma:%p mapping:%p index:%lx\n",
		(void *)addr, vma->vm_flags, vma->anon_vma, mapping, index);
	/*
	 * Choose text because data symbols depend on CONFIG_KALLSYMS_ALL=y
	 */
	if (vma->vm_ops)
713
		print_symbol(KERN_ALERT "vma->vm_ops->fault: %s\n",
714 715
				(unsigned long)vma->vm_ops->fault);
	if (vma->vm_file && vma->vm_file->f_op)
716
		print_symbol(KERN_ALERT "vma->vm_file->f_op->mmap: %s\n",
717
				(unsigned long)vma->vm_file->f_op->mmap);
N
Nick Piggin 已提交
718
	dump_stack();
719
	add_taint(TAINT_BAD_PAGE);
N
Nick Piggin 已提交
720 721
}

722
static inline bool is_cow_mapping(vm_flags_t flags)
723 724 725 726
{
	return (flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
}

H
Hugh Dickins 已提交
727
/*
N
Nick Piggin 已提交
728
 * vm_normal_page -- This function gets the "struct page" associated with a pte.
729
 *
N
Nick Piggin 已提交
730 731 732
 * "Special" mappings do not wish to be associated with a "struct page" (either
 * it doesn't exist, or it exists but they don't want to touch it). In this
 * case, NULL is returned here. "Normal" mappings do have a struct page.
J
Jared Hulbert 已提交
733
 *
N
Nick Piggin 已提交
734 735 736 737 738 739 740 741
 * There are 2 broad cases. Firstly, an architecture may define a pte_special()
 * pte bit, in which case this function is trivial. Secondly, an architecture
 * may not have a spare pte bit, which requires a more complicated scheme,
 * described below.
 *
 * A raw VM_PFNMAP mapping (ie. one that is not COWed) is always considered a
 * special mapping (even if there are underlying and valid "struct pages").
 * COWed pages of a VM_PFNMAP are always normal.
742
 *
J
Jared Hulbert 已提交
743 744
 * The way we recognize COWed pages within VM_PFNMAP mappings is through the
 * rules set up by "remap_pfn_range()": the vma will have the VM_PFNMAP bit
N
Nick Piggin 已提交
745 746
 * set, and the vm_pgoff will point to the first PFN mapped: thus every special
 * mapping will always honor the rule
747 748 749
 *
 *	pfn_of_page == vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT)
 *
N
Nick Piggin 已提交
750 751 752 753 754 755
 * And for normal mappings this is false.
 *
 * This restricts such mappings to be a linear translation from virtual address
 * to pfn. To get around this restriction, we allow arbitrary mappings so long
 * as the vma is not a COW mapping; in that case, we know that all ptes are
 * special (because none can have been COWed).
J
Jared Hulbert 已提交
756 757
 *
 *
N
Nick Piggin 已提交
758
 * In order to support COW of arbitrary special mappings, we have VM_MIXEDMAP.
J
Jared Hulbert 已提交
759 760 761 762 763 764 765 766 767
 *
 * VM_MIXEDMAP mappings can likewise contain memory with or without "struct
 * page" backing, however the difference is that _all_ pages with a struct
 * page (that is, those where pfn_valid is true) are refcounted and considered
 * normal pages by the VM. The disadvantage is that pages are refcounted
 * (which can be slower and simply not an option for some PFNMAP users). The
 * advantage is that we don't have to follow the strict linearity rule of
 * PFNMAP mappings in order to support COWable mappings.
 *
H
Hugh Dickins 已提交
768
 */
N
Nick Piggin 已提交
769 770 771 772 773 774 775
#ifdef __HAVE_ARCH_PTE_SPECIAL
# define HAVE_PTE_SPECIAL 1
#else
# define HAVE_PTE_SPECIAL 0
#endif
struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
				pte_t pte)
H
Hugh Dickins 已提交
776
{
777
	unsigned long pfn = pte_pfn(pte);
N
Nick Piggin 已提交
778 779

	if (HAVE_PTE_SPECIAL) {
780 781
		if (likely(!pte_special(pte)))
			goto check_pfn;
H
Hugh Dickins 已提交
782 783
		if (vma->vm_flags & (VM_PFNMAP | VM_MIXEDMAP))
			return NULL;
H
Hugh Dickins 已提交
784
		if (!is_zero_pfn(pfn))
785
			print_bad_pte(vma, addr, pte, NULL);
N
Nick Piggin 已提交
786 787 788 789 790
		return NULL;
	}

	/* !HAVE_PTE_SPECIAL case follows: */

J
Jared Hulbert 已提交
791 792 793 794 795 796
	if (unlikely(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP))) {
		if (vma->vm_flags & VM_MIXEDMAP) {
			if (!pfn_valid(pfn))
				return NULL;
			goto out;
		} else {
N
Nick Piggin 已提交
797 798
			unsigned long off;
			off = (addr - vma->vm_start) >> PAGE_SHIFT;
J
Jared Hulbert 已提交
799 800 801 802 803
			if (pfn == vma->vm_pgoff + off)
				return NULL;
			if (!is_cow_mapping(vma->vm_flags))
				return NULL;
		}
804 805
	}

H
Hugh Dickins 已提交
806 807
	if (is_zero_pfn(pfn))
		return NULL;
808 809 810 811 812
check_pfn:
	if (unlikely(pfn > highest_memmap_pfn)) {
		print_bad_pte(vma, addr, pte, NULL);
		return NULL;
	}
813 814

	/*
N
Nick Piggin 已提交
815 816
	 * NOTE! We still have PageReserved() pages in the page tables.
	 * eg. VDSO mappings can cause them to exist.
817
	 */
J
Jared Hulbert 已提交
818
out:
819
	return pfn_to_page(pfn);
H
Hugh Dickins 已提交
820 821
}

L
Linus Torvalds 已提交
822 823 824 825 826 827
/*
 * copy one vm_area from one task to the other. Assumes the page tables
 * already present in the new task to be cleared in the whole range
 * covered by this vma.
 */

H
Hugh Dickins 已提交
828
static inline unsigned long
L
Linus Torvalds 已提交
829
copy_one_pte(struct mm_struct *dst_mm, struct mm_struct *src_mm,
N
Nick Piggin 已提交
830
		pte_t *dst_pte, pte_t *src_pte, struct vm_area_struct *vma,
H
Hugh Dickins 已提交
831
		unsigned long addr, int *rss)
L
Linus Torvalds 已提交
832
{
N
Nick Piggin 已提交
833
	unsigned long vm_flags = vma->vm_flags;
L
Linus Torvalds 已提交
834 835 836 837 838 839
	pte_t pte = *src_pte;
	struct page *page;

	/* pte contains position in swap or file, so copy. */
	if (unlikely(!pte_present(pte))) {
		if (!pte_file(pte)) {
840 841
			swp_entry_t entry = pte_to_swp_entry(pte);

H
Hugh Dickins 已提交
842 843 844
			if (swap_duplicate(entry) < 0)
				return entry.val;

L
Linus Torvalds 已提交
845 846 847
			/* make sure dst_mm is on swapoff's mmlist. */
			if (unlikely(list_empty(&dst_mm->mmlist))) {
				spin_lock(&mmlist_lock);
848 849 850
				if (list_empty(&dst_mm->mmlist))
					list_add(&dst_mm->mmlist,
						 &src_mm->mmlist);
L
Linus Torvalds 已提交
851 852
				spin_unlock(&mmlist_lock);
			}
K
KAMEZAWA Hiroyuki 已提交
853 854
			if (likely(!non_swap_entry(entry)))
				rss[MM_SWAPENTS]++;
855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872
			else if (is_migration_entry(entry)) {
				page = migration_entry_to_page(entry);

				if (PageAnon(page))
					rss[MM_ANONPAGES]++;
				else
					rss[MM_FILEPAGES]++;

				if (is_write_migration_entry(entry) &&
				    is_cow_mapping(vm_flags)) {
					/*
					 * COW mappings require pages in both
					 * parent and child to be set to read.
					 */
					make_migration_entry_read(&entry);
					pte = swp_entry_to_pte(entry);
					set_pte_at(src_mm, addr, src_pte, pte);
				}
873
			}
L
Linus Torvalds 已提交
874
		}
875
		goto out_set_pte;
L
Linus Torvalds 已提交
876 877 878 879 880 881
	}

	/*
	 * If it's a COW mapping, write protect it both
	 * in the parent and the child
	 */
882
	if (is_cow_mapping(vm_flags)) {
L
Linus Torvalds 已提交
883
		ptep_set_wrprotect(src_mm, addr, src_pte);
884
		pte = pte_wrprotect(pte);
L
Linus Torvalds 已提交
885 886 887 888 889 890 891 892 893
	}

	/*
	 * If it's a shared mapping, mark it clean in
	 * the child
	 */
	if (vm_flags & VM_SHARED)
		pte = pte_mkclean(pte);
	pte = pte_mkold(pte);
894 895 896 897

	page = vm_normal_page(vma, addr, pte);
	if (page) {
		get_page(page);
H
Hugh Dickins 已提交
898
		page_dup_rmap(page);
K
KAMEZAWA Hiroyuki 已提交
899 900 901 902
		if (PageAnon(page))
			rss[MM_ANONPAGES]++;
		else
			rss[MM_FILEPAGES]++;
903
	}
904 905 906

out_set_pte:
	set_pte_at(dst_mm, addr, dst_pte, pte);
H
Hugh Dickins 已提交
907
	return 0;
L
Linus Torvalds 已提交
908 909
}

910 911 912
int copy_pte_range(struct mm_struct *dst_mm, struct mm_struct *src_mm,
		   pmd_t *dst_pmd, pmd_t *src_pmd, struct vm_area_struct *vma,
		   unsigned long addr, unsigned long end)
L
Linus Torvalds 已提交
913
{
914
	pte_t *orig_src_pte, *orig_dst_pte;
L
Linus Torvalds 已提交
915
	pte_t *src_pte, *dst_pte;
H
Hugh Dickins 已提交
916
	spinlock_t *src_ptl, *dst_ptl;
917
	int progress = 0;
K
KAMEZAWA Hiroyuki 已提交
918
	int rss[NR_MM_COUNTERS];
H
Hugh Dickins 已提交
919
	swp_entry_t entry = (swp_entry_t){0};
L
Linus Torvalds 已提交
920 921

again:
K
KAMEZAWA Hiroyuki 已提交
922 923
	init_rss_vec(rss);

H
Hugh Dickins 已提交
924
	dst_pte = pte_alloc_map_lock(dst_mm, dst_pmd, addr, &dst_ptl);
L
Linus Torvalds 已提交
925 926
	if (!dst_pte)
		return -ENOMEM;
P
Peter Zijlstra 已提交
927
	src_pte = pte_offset_map(src_pmd, addr);
H
Hugh Dickins 已提交
928
	src_ptl = pte_lockptr(src_mm, src_pmd);
I
Ingo Molnar 已提交
929
	spin_lock_nested(src_ptl, SINGLE_DEPTH_NESTING);
930 931
	orig_src_pte = src_pte;
	orig_dst_pte = dst_pte;
932
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
933 934 935 936 937 938

	do {
		/*
		 * We are holding two locks at this point - either of them
		 * could generate latencies in another task on another CPU.
		 */
939 940 941
		if (progress >= 32) {
			progress = 0;
			if (need_resched() ||
N
Nick Piggin 已提交
942
			    spin_needbreak(src_ptl) || spin_needbreak(dst_ptl))
943 944
				break;
		}
L
Linus Torvalds 已提交
945 946 947 948
		if (pte_none(*src_pte)) {
			progress++;
			continue;
		}
H
Hugh Dickins 已提交
949 950 951 952
		entry.val = copy_one_pte(dst_mm, src_mm, dst_pte, src_pte,
							vma, addr, rss);
		if (entry.val)
			break;
L
Linus Torvalds 已提交
953 954 955
		progress += 8;
	} while (dst_pte++, src_pte++, addr += PAGE_SIZE, addr != end);

956
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
957
	spin_unlock(src_ptl);
P
Peter Zijlstra 已提交
958
	pte_unmap(orig_src_pte);
K
KAMEZAWA Hiroyuki 已提交
959
	add_mm_rss_vec(dst_mm, rss);
960
	pte_unmap_unlock(orig_dst_pte, dst_ptl);
H
Hugh Dickins 已提交
961
	cond_resched();
H
Hugh Dickins 已提交
962 963 964 965 966 967

	if (entry.val) {
		if (add_swap_count_continuation(entry, GFP_KERNEL) < 0)
			return -ENOMEM;
		progress = 0;
	}
L
Linus Torvalds 已提交
968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985
	if (addr != end)
		goto again;
	return 0;
}

static inline int copy_pmd_range(struct mm_struct *dst_mm, struct mm_struct *src_mm,
		pud_t *dst_pud, pud_t *src_pud, struct vm_area_struct *vma,
		unsigned long addr, unsigned long end)
{
	pmd_t *src_pmd, *dst_pmd;
	unsigned long next;

	dst_pmd = pmd_alloc(dst_mm, dst_pud, addr);
	if (!dst_pmd)
		return -ENOMEM;
	src_pmd = pmd_offset(src_pud, addr);
	do {
		next = pmd_addr_end(addr, end);
986 987
		if (pmd_trans_huge(*src_pmd)) {
			int err;
988
			VM_BUG_ON(next-addr != HPAGE_PMD_SIZE);
989 990 991 992 993 994 995 996
			err = copy_huge_pmd(dst_mm, src_mm,
					    dst_pmd, src_pmd, addr, vma);
			if (err == -ENOMEM)
				return -ENOMEM;
			if (!err)
				continue;
			/* fall through */
		}
L
Linus Torvalds 已提交
997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034
		if (pmd_none_or_clear_bad(src_pmd))
			continue;
		if (copy_pte_range(dst_mm, src_mm, dst_pmd, src_pmd,
						vma, addr, next))
			return -ENOMEM;
	} while (dst_pmd++, src_pmd++, addr = next, addr != end);
	return 0;
}

static inline int copy_pud_range(struct mm_struct *dst_mm, struct mm_struct *src_mm,
		pgd_t *dst_pgd, pgd_t *src_pgd, struct vm_area_struct *vma,
		unsigned long addr, unsigned long end)
{
	pud_t *src_pud, *dst_pud;
	unsigned long next;

	dst_pud = pud_alloc(dst_mm, dst_pgd, addr);
	if (!dst_pud)
		return -ENOMEM;
	src_pud = pud_offset(src_pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(src_pud))
			continue;
		if (copy_pmd_range(dst_mm, src_mm, dst_pud, src_pud,
						vma, addr, next))
			return -ENOMEM;
	} while (dst_pud++, src_pud++, addr = next, addr != end);
	return 0;
}

int copy_page_range(struct mm_struct *dst_mm, struct mm_struct *src_mm,
		struct vm_area_struct *vma)
{
	pgd_t *src_pgd, *dst_pgd;
	unsigned long next;
	unsigned long addr = vma->vm_start;
	unsigned long end = vma->vm_end;
1035 1036 1037
	unsigned long mmun_start;	/* For mmu_notifiers */
	unsigned long mmun_end;		/* For mmu_notifiers */
	bool is_cow;
A
Andrea Arcangeli 已提交
1038
	int ret;
L
Linus Torvalds 已提交
1039

1040 1041 1042 1043 1044 1045
	/*
	 * Don't copy ptes where a page fault will fill them correctly.
	 * Fork becomes much lighter when there are big shared or private
	 * readonly mappings. The tradeoff is that copy_page_range is more
	 * efficient than faulting.
	 */
1046 1047
	if (!(vma->vm_flags & (VM_HUGETLB | VM_NONLINEAR |
			       VM_PFNMAP | VM_MIXEDMAP))) {
1048 1049 1050 1051
		if (!vma->anon_vma)
			return 0;
	}

L
Linus Torvalds 已提交
1052 1053 1054
	if (is_vm_hugetlb_page(vma))
		return copy_hugetlb_page_range(dst_mm, src_mm, vma);

1055
	if (unlikely(vma->vm_flags & VM_PFNMAP)) {
1056 1057 1058 1059
		/*
		 * We do not free on error cases below as remove_vma
		 * gets called on error from higher level routine
		 */
1060
		ret = track_pfn_copy(vma);
1061 1062 1063 1064
		if (ret)
			return ret;
	}

A
Andrea Arcangeli 已提交
1065 1066 1067 1068 1069 1070
	/*
	 * We need to invalidate the secondary MMU mappings only when
	 * there could be a permission downgrade on the ptes of the
	 * parent mm. And a permission downgrade will only happen if
	 * is_cow_mapping() returns true.
	 */
1071 1072 1073 1074 1075 1076
	is_cow = is_cow_mapping(vma->vm_flags);
	mmun_start = addr;
	mmun_end   = end;
	if (is_cow)
		mmu_notifier_invalidate_range_start(src_mm, mmun_start,
						    mmun_end);
A
Andrea Arcangeli 已提交
1077 1078

	ret = 0;
L
Linus Torvalds 已提交
1079 1080 1081 1082 1083 1084
	dst_pgd = pgd_offset(dst_mm, addr);
	src_pgd = pgd_offset(src_mm, addr);
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(src_pgd))
			continue;
A
Andrea Arcangeli 已提交
1085 1086 1087 1088 1089
		if (unlikely(copy_pud_range(dst_mm, src_mm, dst_pgd, src_pgd,
					    vma, addr, next))) {
			ret = -ENOMEM;
			break;
		}
L
Linus Torvalds 已提交
1090
	} while (dst_pgd++, src_pgd++, addr = next, addr != end);
A
Andrea Arcangeli 已提交
1091

1092 1093
	if (is_cow)
		mmu_notifier_invalidate_range_end(src_mm, mmun_start, mmun_end);
A
Andrea Arcangeli 已提交
1094
	return ret;
L
Linus Torvalds 已提交
1095 1096
}

1097
static unsigned long zap_pte_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
1098
				struct vm_area_struct *vma, pmd_t *pmd,
L
Linus Torvalds 已提交
1099
				unsigned long addr, unsigned long end,
1100
				struct zap_details *details)
L
Linus Torvalds 已提交
1101
{
N
Nick Piggin 已提交
1102
	struct mm_struct *mm = tlb->mm;
P
Peter Zijlstra 已提交
1103
	int force_flush = 0;
K
KAMEZAWA Hiroyuki 已提交
1104
	int rss[NR_MM_COUNTERS];
1105
	spinlock_t *ptl;
1106
	pte_t *start_pte;
1107
	pte_t *pte;
K
KAMEZAWA Hiroyuki 已提交
1108

P
Peter Zijlstra 已提交
1109
again:
1110
	init_rss_vec(rss);
1111 1112
	start_pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
	pte = start_pte;
1113
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
1114 1115
	do {
		pte_t ptent = *pte;
1116
		if (pte_none(ptent)) {
L
Linus Torvalds 已提交
1117
			continue;
1118
		}
1119

L
Linus Torvalds 已提交
1120
		if (pte_present(ptent)) {
H
Hugh Dickins 已提交
1121
			struct page *page;
1122

1123
			page = vm_normal_page(vma, addr, ptent);
L
Linus Torvalds 已提交
1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141
			if (unlikely(details) && page) {
				/*
				 * unmap_shared_mapping_pages() wants to
				 * invalidate cache without truncating:
				 * unmap shared but keep private pages.
				 */
				if (details->check_mapping &&
				    details->check_mapping != page->mapping)
					continue;
				/*
				 * Each page->index must be checked when
				 * invalidating or truncating nonlinear.
				 */
				if (details->nonlinear_vma &&
				    (page->index < details->first_index ||
				     page->index > details->last_index))
					continue;
			}
N
Nick Piggin 已提交
1142
			ptent = ptep_get_and_clear_full(mm, addr, pte,
1143
							tlb->fullmm);
L
Linus Torvalds 已提交
1144 1145 1146 1147 1148 1149
			tlb_remove_tlb_entry(tlb, pte, addr);
			if (unlikely(!page))
				continue;
			if (unlikely(details) && details->nonlinear_vma
			    && linear_page_index(details->nonlinear_vma,
						addr) != page->index)
N
Nick Piggin 已提交
1150
				set_pte_at(mm, addr, pte,
L
Linus Torvalds 已提交
1151 1152
					   pgoff_to_pte(page->index));
			if (PageAnon(page))
K
KAMEZAWA Hiroyuki 已提交
1153
				rss[MM_ANONPAGES]--;
1154 1155 1156
			else {
				if (pte_dirty(ptent))
					set_page_dirty(page);
1157 1158
				if (pte_young(ptent) &&
				    likely(!VM_SequentialReadHint(vma)))
1159
					mark_page_accessed(page);
K
KAMEZAWA Hiroyuki 已提交
1160
				rss[MM_FILEPAGES]--;
1161
			}
1162
			page_remove_rmap(page);
1163 1164
			if (unlikely(page_mapcount(page) < 0))
				print_bad_pte(vma, addr, ptent, page);
P
Peter Zijlstra 已提交
1165 1166 1167
			force_flush = !__tlb_remove_page(tlb, page);
			if (force_flush)
				break;
L
Linus Torvalds 已提交
1168 1169 1170 1171 1172 1173 1174 1175
			continue;
		}
		/*
		 * If details->check_mapping, we leave swap entries;
		 * if details->nonlinear_vma, we leave file entries.
		 */
		if (unlikely(details))
			continue;
1176 1177 1178
		if (pte_file(ptent)) {
			if (unlikely(!(vma->vm_flags & VM_NONLINEAR)))
				print_bad_pte(vma, addr, ptent, NULL);
K
KAMEZAWA Hiroyuki 已提交
1179 1180 1181 1182 1183
		} else {
			swp_entry_t entry = pte_to_swp_entry(ptent);

			if (!non_swap_entry(entry))
				rss[MM_SWAPENTS]--;
1184 1185 1186 1187 1188 1189 1190 1191 1192 1193
			else if (is_migration_entry(entry)) {
				struct page *page;

				page = migration_entry_to_page(entry);

				if (PageAnon(page))
					rss[MM_ANONPAGES]--;
				else
					rss[MM_FILEPAGES]--;
			}
K
KAMEZAWA Hiroyuki 已提交
1194 1195 1196
			if (unlikely(!free_swap_and_cache(entry)))
				print_bad_pte(vma, addr, ptent, NULL);
		}
1197
		pte_clear_not_present_full(mm, addr, pte, tlb->fullmm);
1198
	} while (pte++, addr += PAGE_SIZE, addr != end);
1199

K
KAMEZAWA Hiroyuki 已提交
1200
	add_mm_rss_vec(mm, rss);
1201
	arch_leave_lazy_mmu_mode();
1202
	pte_unmap_unlock(start_pte, ptl);
1203

P
Peter Zijlstra 已提交
1204 1205 1206 1207 1208 1209 1210
	/*
	 * 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;
1211 1212 1213 1214 1215

#ifdef HAVE_GENERIC_MMU_GATHER
		tlb->start = addr;
		tlb->end = end;
#endif
P
Peter Zijlstra 已提交
1216 1217 1218 1219 1220
		tlb_flush_mmu(tlb);
		if (addr != end)
			goto again;
	}

1221
	return addr;
L
Linus Torvalds 已提交
1222 1223
}

1224
static inline unsigned long zap_pmd_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
1225
				struct vm_area_struct *vma, pud_t *pud,
L
Linus Torvalds 已提交
1226
				unsigned long addr, unsigned long end,
1227
				struct zap_details *details)
L
Linus Torvalds 已提交
1228 1229 1230 1231 1232 1233 1234
{
	pmd_t *pmd;
	unsigned long next;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
1235
		if (pmd_trans_huge(*pmd)) {
1236
			if (next - addr != HPAGE_PMD_SIZE) {
1237 1238 1239 1240 1241 1242 1243 1244 1245
#ifdef CONFIG_DEBUG_VM
				if (!rwsem_is_locked(&tlb->mm->mmap_sem)) {
					pr_err("%s: mmap_sem is unlocked! addr=0x%lx end=0x%lx vma->vm_start=0x%lx vma->vm_end=0x%lx\n",
						__func__, addr, end,
						vma->vm_start,
						vma->vm_end);
					BUG();
				}
#endif
1246
				split_huge_page_pmd(vma, addr, pmd);
S
Shaohua Li 已提交
1247
			} else if (zap_huge_pmd(tlb, vma, pmd, addr))
1248
				goto next;
1249 1250
			/* fall through */
		}
1251 1252 1253 1254 1255 1256 1257 1258 1259
		/*
		 * Here there can be other concurrent MADV_DONTNEED or
		 * trans huge page faults running, and if the pmd is
		 * none or trans huge it can change under us. This is
		 * because MADV_DONTNEED holds the mmap_sem in read
		 * mode.
		 */
		if (pmd_none_or_trans_huge_or_clear_bad(pmd))
			goto next;
1260
		next = zap_pte_range(tlb, vma, pmd, addr, next, details);
1261
next:
1262 1263
		cond_resched();
	} while (pmd++, addr = next, addr != end);
1264 1265

	return addr;
L
Linus Torvalds 已提交
1266 1267
}

1268
static inline unsigned long zap_pud_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
1269
				struct vm_area_struct *vma, pgd_t *pgd,
L
Linus Torvalds 已提交
1270
				unsigned long addr, unsigned long end,
1271
				struct zap_details *details)
L
Linus Torvalds 已提交
1272 1273 1274 1275 1276 1277 1278
{
	pud_t *pud;
	unsigned long next;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
1279
		if (pud_none_or_clear_bad(pud))
L
Linus Torvalds 已提交
1280
			continue;
1281 1282
		next = zap_pmd_range(tlb, vma, pud, addr, next, details);
	} while (pud++, addr = next, addr != end);
1283 1284

	return addr;
L
Linus Torvalds 已提交
1285 1286
}

A
Al Viro 已提交
1287 1288 1289 1290
static void unmap_page_range(struct mmu_gather *tlb,
			     struct vm_area_struct *vma,
			     unsigned long addr, unsigned long end,
			     struct zap_details *details)
L
Linus Torvalds 已提交
1291 1292 1293 1294 1295 1296 1297 1298
{
	pgd_t *pgd;
	unsigned long next;

	if (details && !details->check_mapping && !details->nonlinear_vma)
		details = NULL;

	BUG_ON(addr >= end);
1299
	mem_cgroup_uncharge_start();
L
Linus Torvalds 已提交
1300 1301 1302 1303
	tlb_start_vma(tlb, vma);
	pgd = pgd_offset(vma->vm_mm, addr);
	do {
		next = pgd_addr_end(addr, end);
1304
		if (pgd_none_or_clear_bad(pgd))
L
Linus Torvalds 已提交
1305
			continue;
1306 1307
		next = zap_pud_range(tlb, vma, pgd, addr, next, details);
	} while (pgd++, addr = next, addr != end);
L
Linus Torvalds 已提交
1308
	tlb_end_vma(tlb, vma);
1309
	mem_cgroup_uncharge_end();
L
Linus Torvalds 已提交
1310
}
1311

1312 1313 1314

static void unmap_single_vma(struct mmu_gather *tlb,
		struct vm_area_struct *vma, unsigned long start_addr,
1315
		unsigned long end_addr,
1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326
		struct zap_details *details)
{
	unsigned long start = max(vma->vm_start, start_addr);
	unsigned long end;

	if (start >= vma->vm_end)
		return;
	end = min(vma->vm_end, end_addr);
	if (end <= vma->vm_start)
		return;

1327 1328 1329
	if (vma->vm_file)
		uprobe_munmap(vma, start, end);

1330
	if (unlikely(vma->vm_flags & VM_PFNMAP))
1331
		untrack_pfn(vma, 0, 0);
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345

	if (start != end) {
		if (unlikely(is_vm_hugetlb_page(vma))) {
			/*
			 * It is undesirable to test vma->vm_file as it
			 * should be non-null for valid hugetlb area.
			 * However, vm_file will be NULL in the error
			 * cleanup path of do_mmap_pgoff. When
			 * hugetlbfs ->mmap method fails,
			 * do_mmap_pgoff() nullifies vma->vm_file
			 * before calling this function to clean up.
			 * Since no pte has actually been setup, it is
			 * safe to do nothing in this case.
			 */
1346 1347
			if (vma->vm_file) {
				mutex_lock(&vma->vm_file->f_mapping->i_mmap_mutex);
1348
				__unmap_hugepage_range_final(tlb, vma, start, end, NULL);
1349 1350
				mutex_unlock(&vma->vm_file->f_mapping->i_mmap_mutex);
			}
1351 1352 1353
		} else
			unmap_page_range(tlb, vma, start, end, details);
	}
L
Linus Torvalds 已提交
1354 1355 1356 1357
}

/**
 * unmap_vmas - unmap a range of memory covered by a list of vma's
1358
 * @tlb: address of the caller's struct mmu_gather
L
Linus Torvalds 已提交
1359 1360 1361 1362
 * @vma: the starting vma
 * @start_addr: virtual address at which to start unmapping
 * @end_addr: virtual address at which to end unmapping
 *
1363
 * Unmap all pages in the vma list.
L
Linus Torvalds 已提交
1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
 *
 * Only addresses between `start' and `end' will be unmapped.
 *
 * The VMA list must be sorted in ascending virtual address order.
 *
 * unmap_vmas() assumes that the caller will flush the whole unmapped address
 * range after unmap_vmas() returns.  So the only responsibility here is to
 * ensure that any thus-far unmapped pages are flushed before unmap_vmas()
 * drops the lock and schedules.
 */
A
Al Viro 已提交
1374
void unmap_vmas(struct mmu_gather *tlb,
L
Linus Torvalds 已提交
1375
		struct vm_area_struct *vma, unsigned long start_addr,
1376
		unsigned long end_addr)
L
Linus Torvalds 已提交
1377
{
A
Andrea Arcangeli 已提交
1378
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
1379

A
Andrea Arcangeli 已提交
1380
	mmu_notifier_invalidate_range_start(mm, start_addr, end_addr);
1381
	for ( ; vma && vma->vm_start < end_addr; vma = vma->vm_next)
1382
		unmap_single_vma(tlb, vma, start_addr, end_addr, NULL);
A
Andrea Arcangeli 已提交
1383
	mmu_notifier_invalidate_range_end(mm, start_addr, end_addr);
L
Linus Torvalds 已提交
1384 1385 1386 1387 1388
}

/**
 * zap_page_range - remove user pages in a given range
 * @vma: vm_area_struct holding the applicable pages
1389
 * @start: starting address of pages to zap
L
Linus Torvalds 已提交
1390 1391
 * @size: number of bytes to zap
 * @details: details of nonlinear truncation or shared cache invalidation
1392 1393
 *
 * Caller must protect the VMA list
L
Linus Torvalds 已提交
1394
 */
1395
void zap_page_range(struct vm_area_struct *vma, unsigned long start,
L
Linus Torvalds 已提交
1396 1397 1398
		unsigned long size, struct zap_details *details)
{
	struct mm_struct *mm = vma->vm_mm;
P
Peter Zijlstra 已提交
1399
	struct mmu_gather tlb;
1400
	unsigned long end = start + size;
L
Linus Torvalds 已提交
1401 1402

	lru_add_drain();
P
Peter Zijlstra 已提交
1403
	tlb_gather_mmu(&tlb, mm, 0);
1404
	update_hiwater_rss(mm);
1405 1406
	mmu_notifier_invalidate_range_start(mm, start, end);
	for ( ; vma && vma->vm_start < end; vma = vma->vm_next)
1407
		unmap_single_vma(&tlb, vma, start, end, details);
1408 1409
	mmu_notifier_invalidate_range_end(mm, start, end);
	tlb_finish_mmu(&tlb, start, end);
L
Linus Torvalds 已提交
1410 1411
}

1412 1413 1414 1415 1416 1417 1418 1419
/**
 * zap_page_range_single - remove user pages in a given range
 * @vma: vm_area_struct holding the applicable pages
 * @address: starting address of pages to zap
 * @size: number of bytes to zap
 * @details: details of nonlinear truncation or shared cache invalidation
 *
 * The range must fit into one VMA.
L
Linus Torvalds 已提交
1420
 */
1421
static void zap_page_range_single(struct vm_area_struct *vma, unsigned long address,
L
Linus Torvalds 已提交
1422 1423 1424
		unsigned long size, struct zap_details *details)
{
	struct mm_struct *mm = vma->vm_mm;
P
Peter Zijlstra 已提交
1425
	struct mmu_gather tlb;
L
Linus Torvalds 已提交
1426 1427 1428
	unsigned long end = address + size;

	lru_add_drain();
P
Peter Zijlstra 已提交
1429
	tlb_gather_mmu(&tlb, mm, 0);
1430
	update_hiwater_rss(mm);
1431
	mmu_notifier_invalidate_range_start(mm, address, end);
1432
	unmap_single_vma(&tlb, vma, address, end, details);
1433
	mmu_notifier_invalidate_range_end(mm, address, end);
P
Peter Zijlstra 已提交
1434
	tlb_finish_mmu(&tlb, address, end);
L
Linus Torvalds 已提交
1435 1436
}

1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454
/**
 * zap_vma_ptes - remove ptes mapping the vma
 * @vma: vm_area_struct holding ptes to be zapped
 * @address: starting address of pages to zap
 * @size: number of bytes to zap
 *
 * This function only unmaps ptes assigned to VM_PFNMAP vmas.
 *
 * The entire address range must be fully contained within the vma.
 *
 * Returns 0 if successful.
 */
int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
		unsigned long size)
{
	if (address < vma->vm_start || address + size > vma->vm_end ||
	    		!(vma->vm_flags & VM_PFNMAP))
		return -1;
1455
	zap_page_range_single(vma, address, size, NULL);
1456 1457 1458 1459
	return 0;
}
EXPORT_SYMBOL_GPL(zap_vma_ptes);

J
Johannes Weiner 已提交
1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470
/**
 * follow_page - look up a page descriptor from a user-virtual address
 * @vma: vm_area_struct mapping @address
 * @address: virtual address to look up
 * @flags: flags modifying lookup behaviour
 *
 * @flags can have FOLL_ flags set, defined in <linux/mm.h>
 *
 * Returns the mapped (struct page *), %NULL if no mapping exists, or
 * an error pointer if there is a mapping to something not represented
 * by a page descriptor (see also vm_normal_page()).
L
Linus Torvalds 已提交
1471
 */
1472
struct page *follow_page(struct vm_area_struct *vma, unsigned long address,
1473
			unsigned int flags)
L
Linus Torvalds 已提交
1474 1475 1476 1477 1478
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *ptep, pte;
1479
	spinlock_t *ptl;
L
Linus Torvalds 已提交
1480
	struct page *page;
1481
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
1482

1483 1484 1485 1486 1487
	page = follow_huge_addr(mm, address, flags & FOLL_WRITE);
	if (!IS_ERR(page)) {
		BUG_ON(flags & FOLL_GET);
		goto out;
	}
L
Linus Torvalds 已提交
1488

1489
	page = NULL;
L
Linus Torvalds 已提交
1490 1491
	pgd = pgd_offset(mm, address);
	if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
1492
		goto no_page_table;
L
Linus Torvalds 已提交
1493 1494

	pud = pud_offset(pgd, address);
A
Andi Kleen 已提交
1495
	if (pud_none(*pud))
1496
		goto no_page_table;
1497
	if (pud_huge(*pud) && vma->vm_flags & VM_HUGETLB) {
A
Andi Kleen 已提交
1498 1499 1500 1501 1502 1503 1504
		BUG_ON(flags & FOLL_GET);
		page = follow_huge_pud(mm, address, pud, flags & FOLL_WRITE);
		goto out;
	}
	if (unlikely(pud_bad(*pud)))
		goto no_page_table;

L
Linus Torvalds 已提交
1505
	pmd = pmd_offset(pud, address);
1506
	if (pmd_none(*pmd))
1507
		goto no_page_table;
1508
	if (pmd_huge(*pmd) && vma->vm_flags & VM_HUGETLB) {
1509 1510
		BUG_ON(flags & FOLL_GET);
		page = follow_huge_pmd(mm, address, pmd, flags & FOLL_WRITE);
L
Linus Torvalds 已提交
1511
		goto out;
1512
	}
1513 1514
	if ((flags & FOLL_NUMA) && pmd_numa(*pmd))
		goto no_page_table;
1515
	if (pmd_trans_huge(*pmd)) {
1516
		if (flags & FOLL_SPLIT) {
1517
			split_huge_page_pmd(vma, address, pmd);
1518 1519
			goto split_fallthrough;
		}
1520 1521 1522 1523 1524 1525
		spin_lock(&mm->page_table_lock);
		if (likely(pmd_trans_huge(*pmd))) {
			if (unlikely(pmd_trans_splitting(*pmd))) {
				spin_unlock(&mm->page_table_lock);
				wait_split_huge_page(vma->anon_vma, pmd);
			} else {
1526
				page = follow_trans_huge_pmd(vma, address,
1527 1528 1529 1530 1531 1532 1533 1534
							     pmd, flags);
				spin_unlock(&mm->page_table_lock);
				goto out;
			}
		} else
			spin_unlock(&mm->page_table_lock);
		/* fall through */
	}
1535
split_fallthrough:
1536 1537 1538
	if (unlikely(pmd_bad(*pmd)))
		goto no_page_table;

1539
	ptep = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
1540 1541

	pte = *ptep;
1542
	if (!pte_present(pte))
1543
		goto no_page;
1544 1545
	if ((flags & FOLL_NUMA) && pte_numa(pte))
		goto no_page;
1546 1547
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;
H
Hugh Dickins 已提交
1548

1549
	page = vm_normal_page(vma, address, pte);
H
Hugh Dickins 已提交
1550 1551
	if (unlikely(!page)) {
		if ((flags & FOLL_DUMP) ||
H
Hugh Dickins 已提交
1552
		    !is_zero_pfn(pte_pfn(pte)))
H
Hugh Dickins 已提交
1553 1554 1555
			goto bad_page;
		page = pte_page(pte);
	}
L
Linus Torvalds 已提交
1556

1557
	if (flags & FOLL_GET)
1558
		get_page_foll(page);
1559 1560 1561 1562
	if (flags & FOLL_TOUCH) {
		if ((flags & FOLL_WRITE) &&
		    !pte_dirty(pte) && !PageDirty(page))
			set_page_dirty(page);
1563 1564 1565 1566 1567
		/*
		 * pte_mkyoung() would be more correct here, but atomic care
		 * is needed to avoid losing the dirty bit: it is easier to use
		 * mark_page_accessed().
		 */
1568 1569
		mark_page_accessed(page);
	}
1570
	if ((flags & FOLL_MLOCK) && (vma->vm_flags & VM_LOCKED)) {
1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582
		/*
		 * The preliminary mapping check is mainly to avoid the
		 * pointless overhead of lock_page on the ZERO_PAGE
		 * which might bounce very badly if there is contention.
		 *
		 * If the page is already locked, we don't need to
		 * handle it now - vmscan will handle it later if and
		 * when it attempts to reclaim the page.
		 */
		if (page->mapping && trylock_page(page)) {
			lru_add_drain();  /* push cached pages to LRU */
			/*
1583 1584 1585 1586
			 * Because we lock page here, and migration is
			 * blocked by the pte's page reference, and we
			 * know the page is still mapped, we don't even
			 * need to check for file-cache page truncation.
1587
			 */
1588
			mlock_vma_page(page);
1589 1590 1591
			unlock_page(page);
		}
	}
1592 1593
unlock:
	pte_unmap_unlock(ptep, ptl);
L
Linus Torvalds 已提交
1594
out:
1595
	return page;
L
Linus Torvalds 已提交
1596

1597 1598 1599 1600 1601 1602 1603 1604
bad_page:
	pte_unmap_unlock(ptep, ptl);
	return ERR_PTR(-EFAULT);

no_page:
	pte_unmap_unlock(ptep, ptl);
	if (!pte_none(pte))
		return page;
H
Hugh Dickins 已提交
1605

1606 1607 1608
no_page_table:
	/*
	 * When core dumping an enormous anonymous area that nobody
H
Hugh Dickins 已提交
1609 1610 1611 1612 1613
	 * has touched so far, we don't want to allocate unnecessary pages or
	 * page tables.  Return error instead of NULL to skip handle_mm_fault,
	 * then get_dump_page() will return NULL to leave a hole in the dump.
	 * But we can only make this optimization where a hole would surely
	 * be zero-filled if handle_mm_fault() actually did handle it.
1614
	 */
H
Hugh Dickins 已提交
1615 1616 1617
	if ((flags & FOLL_DUMP) &&
	    (!vma->vm_ops || !vma->vm_ops->fault))
		return ERR_PTR(-EFAULT);
1618
	return page;
L
Linus Torvalds 已提交
1619 1620
}

1621 1622
static inline int stack_guard_page(struct vm_area_struct *vma, unsigned long addr)
{
1623 1624
	return stack_guard_page_start(vma, addr) ||
	       stack_guard_page_end(vma, addr+PAGE_SIZE);
1625 1626
}

H
Huang Ying 已提交
1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675
/**
 * __get_user_pages() - pin user pages in memory
 * @tsk:	task_struct of target task
 * @mm:		mm_struct of target mm
 * @start:	starting user address
 * @nr_pages:	number of pages from start to pin
 * @gup_flags:	flags modifying pin behaviour
 * @pages:	array that receives pointers to the pages pinned.
 *		Should be at least nr_pages long. Or NULL, if caller
 *		only intends to ensure the pages are faulted in.
 * @vmas:	array of pointers to vmas corresponding to each page.
 *		Or NULL if the caller does not require them.
 * @nonblocking: whether waiting for disk IO or mmap_sem contention
 *
 * Returns number of pages pinned. This may be fewer than the number
 * requested. If nr_pages is 0 or negative, returns 0. If no pages
 * were pinned, returns -errno. Each page returned must be released
 * with a put_page() call when it is finished with. vmas will only
 * remain valid while mmap_sem is held.
 *
 * Must be called with mmap_sem held for read or write.
 *
 * __get_user_pages walks a process's page tables and takes a reference to
 * each struct page that each user address corresponds to at a given
 * instant. That is, it takes the page that would be accessed if a user
 * thread accesses the given user virtual address at that instant.
 *
 * This does not guarantee that the page exists in the user mappings when
 * __get_user_pages returns, and there may even be a completely different
 * page there in some cases (eg. if mmapped pagecache has been invalidated
 * and subsequently re faulted). However it does guarantee that the page
 * won't be freed completely. And mostly callers simply care that the page
 * contains data that was valid *at some point in time*. Typically, an IO
 * or similar operation cannot guarantee anything stronger anyway because
 * locks can't be held over the syscall boundary.
 *
 * If @gup_flags & FOLL_WRITE == 0, the page must not be written to. If
 * the page is written to, set_page_dirty (or set_page_dirty_lock, as
 * appropriate) must be called after the page is finished with, and
 * before put_page is called.
 *
 * If @nonblocking != NULL, __get_user_pages will not wait for disk IO
 * or mmap_sem contention, and if waiting is needed to pin all pages,
 * *@nonblocking will be set to 0.
 *
 * In most cases, get_user_pages or get_user_pages_fast should be used
 * instead of __get_user_pages. __get_user_pages should be used only if
 * you need some special @gup_flags.
 */
N
Nick Piggin 已提交
1676
int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
H
Hugh Dickins 已提交
1677
		     unsigned long start, int nr_pages, unsigned int gup_flags,
1678 1679
		     struct page **pages, struct vm_area_struct **vmas,
		     int *nonblocking)
L
Linus Torvalds 已提交
1680 1681
{
	int i;
H
Hugh Dickins 已提交
1682
	unsigned long vm_flags;
L
Linus Torvalds 已提交
1683

1684
	if (nr_pages <= 0)
1685
		return 0;
H
Hugh Dickins 已提交
1686 1687 1688

	VM_BUG_ON(!!pages != !!(gup_flags & FOLL_GET));

L
Linus Torvalds 已提交
1689 1690
	/* 
	 * Require read or write permissions.
H
Hugh Dickins 已提交
1691
	 * If FOLL_FORCE is set, we only require the "MAY" flags.
L
Linus Torvalds 已提交
1692
	 */
H
Hugh Dickins 已提交
1693 1694 1695 1696
	vm_flags  = (gup_flags & FOLL_WRITE) ?
			(VM_WRITE | VM_MAYWRITE) : (VM_READ | VM_MAYREAD);
	vm_flags &= (gup_flags & FOLL_FORCE) ?
			(VM_MAYREAD | VM_MAYWRITE) : (VM_READ | VM_WRITE);
1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709

	/*
	 * If FOLL_FORCE and FOLL_NUMA are both set, handle_mm_fault
	 * would be called on PROT_NONE ranges. We must never invoke
	 * handle_mm_fault on PROT_NONE ranges or the NUMA hinting
	 * page faults would unprotect the PROT_NONE ranges if
	 * _PAGE_NUMA and _PAGE_PROTNONE are sharing the same pte/pmd
	 * bitflag. So to avoid that, don't set FOLL_NUMA if
	 * FOLL_FORCE is set.
	 */
	if (!(gup_flags & FOLL_FORCE))
		gup_flags |= FOLL_NUMA;

L
Linus Torvalds 已提交
1710 1711 1712
	i = 0;

	do {
1713
		struct vm_area_struct *vma;
L
Linus Torvalds 已提交
1714 1715

		vma = find_extend_vma(mm, start);
1716
		if (!vma && in_gate_area(mm, start)) {
L
Linus Torvalds 已提交
1717 1718 1719 1720 1721
			unsigned long pg = start & PAGE_MASK;
			pgd_t *pgd;
			pud_t *pud;
			pmd_t *pmd;
			pte_t *pte;
N
Nick Piggin 已提交
1722 1723

			/* user gate pages are read-only */
H
Hugh Dickins 已提交
1724
			if (gup_flags & FOLL_WRITE)
L
Linus Torvalds 已提交
1725 1726 1727 1728 1729 1730 1731 1732 1733
				return i ? : -EFAULT;
			if (pg > TASK_SIZE)
				pgd = pgd_offset_k(pg);
			else
				pgd = pgd_offset_gate(mm, pg);
			BUG_ON(pgd_none(*pgd));
			pud = pud_offset(pgd, pg);
			BUG_ON(pud_none(*pud));
			pmd = pmd_offset(pud, pg);
1734 1735
			if (pmd_none(*pmd))
				return i ? : -EFAULT;
1736
			VM_BUG_ON(pmd_trans_huge(*pmd));
L
Linus Torvalds 已提交
1737
			pte = pte_offset_map(pmd, pg);
1738 1739 1740 1741
			if (pte_none(*pte)) {
				pte_unmap(pte);
				return i ? : -EFAULT;
			}
1742
			vma = get_gate_vma(mm);
L
Linus Torvalds 已提交
1743
			if (pages) {
1744 1745
				struct page *page;

1746
				page = vm_normal_page(vma, start, *pte);
1747 1748 1749 1750 1751 1752 1753 1754 1755
				if (!page) {
					if (!(gup_flags & FOLL_DUMP) &&
					     is_zero_pfn(pte_pfn(*pte)))
						page = pte_page(*pte);
					else {
						pte_unmap(pte);
						return i ? : -EFAULT;
					}
				}
1756
				pages[i] = page;
1757
				get_page(page);
L
Linus Torvalds 已提交
1758 1759
			}
			pte_unmap(pte);
1760
			goto next_page;
L
Linus Torvalds 已提交
1761 1762
		}

N
Nick Piggin 已提交
1763 1764
		if (!vma ||
		    (vma->vm_flags & (VM_IO | VM_PFNMAP)) ||
H
Hugh Dickins 已提交
1765
		    !(vm_flags & vma->vm_flags))
L
Linus Torvalds 已提交
1766 1767
			return i ? : -EFAULT;

H
Hugh Dickins 已提交
1768 1769
		if (is_vm_hugetlb_page(vma)) {
			i = follow_hugetlb_page(mm, vma, pages, vmas,
H
Hugh Dickins 已提交
1770
					&start, &nr_pages, i, gup_flags);
H
Hugh Dickins 已提交
1771 1772
			continue;
		}
1773

L
Linus Torvalds 已提交
1774
		do {
1775
			struct page *page;
H
Hugh Dickins 已提交
1776
			unsigned int foll_flags = gup_flags;
L
Linus Torvalds 已提交
1777

1778
			/*
1779
			 * If we have a pending SIGKILL, don't keep faulting
H
Hugh Dickins 已提交
1780
			 * pages and potentially allocating memory.
1781
			 */
H
Hugh Dickins 已提交
1782
			if (unlikely(fatal_signal_pending(current)))
1783
				return i ? i : -ERESTARTSYS;
1784

1785
			cond_resched();
1786
			while (!(page = follow_page(vma, start, foll_flags))) {
1787
				int ret;
1788 1789
				unsigned int fault_flags = 0;

1790 1791 1792 1793 1794
				/* For mlock, just skip the stack guard page. */
				if (foll_flags & FOLL_MLOCK) {
					if (stack_guard_page(vma, start))
						goto next_page;
				}
1795 1796 1797 1798
				if (foll_flags & FOLL_WRITE)
					fault_flags |= FAULT_FLAG_WRITE;
				if (nonblocking)
					fault_flags |= FAULT_FLAG_ALLOW_RETRY;
1799 1800
				if (foll_flags & FOLL_NOWAIT)
					fault_flags |= (FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_RETRY_NOWAIT);
1801

1802
				ret = handle_mm_fault(mm, vma, start,
1803
							fault_flags);
1804

N
Nick Piggin 已提交
1805 1806 1807
				if (ret & VM_FAULT_ERROR) {
					if (ret & VM_FAULT_OOM)
						return i ? i : -ENOMEM;
1808 1809 1810 1811 1812 1813 1814 1815 1816 1817
					if (ret & (VM_FAULT_HWPOISON |
						   VM_FAULT_HWPOISON_LARGE)) {
						if (i)
							return i;
						else if (gup_flags & FOLL_HWPOISON)
							return -EHWPOISON;
						else
							return -EFAULT;
					}
					if (ret & VM_FAULT_SIGBUS)
N
Nick Piggin 已提交
1818 1819 1820
						return i ? i : -EFAULT;
					BUG();
				}
1821 1822 1823 1824 1825 1826 1827

				if (tsk) {
					if (ret & VM_FAULT_MAJOR)
						tsk->maj_flt++;
					else
						tsk->min_flt++;
				}
N
Nick Piggin 已提交
1828

1829
				if (ret & VM_FAULT_RETRY) {
1830 1831
					if (nonblocking)
						*nonblocking = 0;
1832 1833 1834
					return i;
				}

1835
				/*
N
Nick Piggin 已提交
1836 1837 1838 1839
				 * The VM_FAULT_WRITE bit tells us that
				 * do_wp_page has broken COW when necessary,
				 * even if maybe_mkwrite decided not to set
				 * pte_write. We can thus safely do subsequent
1840 1841 1842 1843 1844 1845
				 * page lookups as if they were reads. But only
				 * do so when looping for pte_write is futile:
				 * in some cases userspace may also be wanting
				 * to write to the gotten user page, which a
				 * read fault here might prevent (a readonly
				 * page might get reCOWed by userspace write).
1846
				 */
1847 1848
				if ((ret & VM_FAULT_WRITE) &&
				    !(vma->vm_flags & VM_WRITE))
1849
					foll_flags &= ~FOLL_WRITE;
N
Nick Piggin 已提交
1850

1851
				cond_resched();
L
Linus Torvalds 已提交
1852
			}
1853 1854
			if (IS_ERR(page))
				return i ? i : PTR_ERR(page);
L
Linus Torvalds 已提交
1855
			if (pages) {
1856
				pages[i] = page;
1857

1858
				flush_anon_page(vma, page, start);
1859
				flush_dcache_page(page);
L
Linus Torvalds 已提交
1860
			}
1861
next_page:
L
Linus Torvalds 已提交
1862 1863 1864 1865
			if (vmas)
				vmas[i] = vma;
			i++;
			start += PAGE_SIZE;
1866 1867 1868
			nr_pages--;
		} while (nr_pages && start < vma->vm_end);
	} while (nr_pages);
L
Linus Torvalds 已提交
1869 1870
	return i;
}
H
Huang Ying 已提交
1871
EXPORT_SYMBOL(__get_user_pages);
N
Nick Piggin 已提交
1872

1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929
/*
 * fixup_user_fault() - manually resolve a user page fault
 * @tsk:	the task_struct to use for page fault accounting, or
 *		NULL if faults are not to be recorded.
 * @mm:		mm_struct of target mm
 * @address:	user address
 * @fault_flags:flags to pass down to handle_mm_fault()
 *
 * This is meant to be called in the specific scenario where for locking reasons
 * we try to access user memory in atomic context (within a pagefault_disable()
 * section), this returns -EFAULT, and we want to resolve the user fault before
 * trying again.
 *
 * Typically this is meant to be used by the futex code.
 *
 * The main difference with get_user_pages() is that this function will
 * unconditionally call handle_mm_fault() which will in turn perform all the
 * necessary SW fixup of the dirty and young bits in the PTE, while
 * handle_mm_fault() only guarantees to update these in the struct page.
 *
 * This is important for some architectures where those bits also gate the
 * access permission to the page because they are maintained in software.  On
 * such architectures, gup() will not be enough to make a subsequent access
 * succeed.
 *
 * This should be called with the mm_sem held for read.
 */
int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm,
		     unsigned long address, unsigned int fault_flags)
{
	struct vm_area_struct *vma;
	int ret;

	vma = find_extend_vma(mm, address);
	if (!vma || address < vma->vm_start)
		return -EFAULT;

	ret = handle_mm_fault(mm, vma, address, fault_flags);
	if (ret & VM_FAULT_ERROR) {
		if (ret & VM_FAULT_OOM)
			return -ENOMEM;
		if (ret & (VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE))
			return -EHWPOISON;
		if (ret & VM_FAULT_SIGBUS)
			return -EFAULT;
		BUG();
	}
	if (tsk) {
		if (ret & VM_FAULT_MAJOR)
			tsk->maj_flt++;
		else
			tsk->min_flt++;
	}
	return 0;
}

/*
1930
 * get_user_pages() - pin user pages in memory
1931 1932
 * @tsk:	the task_struct to use for page fault accounting, or
 *		NULL if faults are not to be recorded.
1933 1934
 * @mm:		mm_struct of target mm
 * @start:	starting user address
1935
 * @nr_pages:	number of pages from start to pin
1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946
 * @write:	whether pages will be written to by the caller
 * @force:	whether to force write access even if user mapping is
 *		readonly. This will result in the page being COWed even
 *		in MAP_SHARED mappings. You do not want this.
 * @pages:	array that receives pointers to the pages pinned.
 *		Should be at least nr_pages long. Or NULL, if caller
 *		only intends to ensure the pages are faulted in.
 * @vmas:	array of pointers to vmas corresponding to each page.
 *		Or NULL if the caller does not require them.
 *
 * Returns number of pages pinned. This may be fewer than the number
1947
 * requested. If nr_pages is 0 or negative, returns 0. If no pages
1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
 * were pinned, returns -errno. Each page returned must be released
 * with a put_page() call when it is finished with. vmas will only
 * remain valid while mmap_sem is held.
 *
 * Must be called with mmap_sem held for read or write.
 *
 * get_user_pages walks a process's page tables and takes a reference to
 * each struct page that each user address corresponds to at a given
 * instant. That is, it takes the page that would be accessed if a user
 * thread accesses the given user virtual address at that instant.
 *
 * This does not guarantee that the page exists in the user mappings when
 * get_user_pages returns, and there may even be a completely different
 * page there in some cases (eg. if mmapped pagecache has been invalidated
 * and subsequently re faulted). However it does guarantee that the page
 * won't be freed completely. And mostly callers simply care that the page
 * contains data that was valid *at some point in time*. Typically, an IO
 * or similar operation cannot guarantee anything stronger anyway because
 * locks can't be held over the syscall boundary.
 *
 * If write=0, the page must not be written to. If the page is written to,
 * set_page_dirty (or set_page_dirty_lock, as appropriate) must be called
 * after the page is finished with, and before put_page is called.
 *
 * get_user_pages is typically used for fewer-copy IO operations, to get a
 * handle on the memory by some means other than accesses via the user virtual
 * addresses. The pages may be submitted for DMA to devices or accessed via
 * their kernel linear mapping (via the kmap APIs). Care should be taken to
 * use the correct cache flushing APIs.
 *
 * See also get_user_pages_fast, for performance critical applications.
 */
N
Nick Piggin 已提交
1980
int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
1981
		unsigned long start, int nr_pages, int write, int force,
N
Nick Piggin 已提交
1982 1983
		struct page **pages, struct vm_area_struct **vmas)
{
H
Hugh Dickins 已提交
1984
	int flags = FOLL_TOUCH;
N
Nick Piggin 已提交
1985

H
Hugh Dickins 已提交
1986 1987
	if (pages)
		flags |= FOLL_GET;
N
Nick Piggin 已提交
1988
	if (write)
H
Hugh Dickins 已提交
1989
		flags |= FOLL_WRITE;
N
Nick Piggin 已提交
1990
	if (force)
H
Hugh Dickins 已提交
1991
		flags |= FOLL_FORCE;
N
Nick Piggin 已提交
1992

1993 1994
	return __get_user_pages(tsk, mm, start, nr_pages, flags, pages, vmas,
				NULL);
N
Nick Piggin 已提交
1995
}
L
Linus Torvalds 已提交
1996 1997
EXPORT_SYMBOL(get_user_pages);

H
Hugh Dickins 已提交
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
/**
 * get_dump_page() - pin user page in memory while writing it to core dump
 * @addr: user address
 *
 * Returns struct page pointer of user page pinned for dump,
 * to be freed afterwards by page_cache_release() or put_page().
 *
 * Returns NULL on any kind of failure - a hole must then be inserted into
 * the corefile, to preserve alignment with its headers; and also returns
 * NULL wherever the ZERO_PAGE, or an anonymous pte_none, has been found -
 * allowing a hole to be left in the corefile to save diskspace.
 *
 * Called without mmap_sem, but after all other threads have been killed.
 */
#ifdef CONFIG_ELF_CORE
struct page *get_dump_page(unsigned long addr)
{
	struct vm_area_struct *vma;
	struct page *page;

	if (__get_user_pages(current, current->mm, addr, 1,
2019 2020
			     FOLL_FORCE | FOLL_DUMP | FOLL_GET, &page, &vma,
			     NULL) < 1)
H
Hugh Dickins 已提交
2021 2022 2023 2024 2025 2026
		return NULL;
	flush_cache_page(vma, addr, page_to_pfn(page));
	return page;
}
#endif /* CONFIG_ELF_CORE */

2027
pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
H
Harvey Harrison 已提交
2028
			spinlock_t **ptl)
2029 2030 2031 2032
{
	pgd_t * pgd = pgd_offset(mm, addr);
	pud_t * pud = pud_alloc(mm, pgd, addr);
	if (pud) {
2033
		pmd_t * pmd = pmd_alloc(mm, pud, addr);
2034 2035
		if (pmd) {
			VM_BUG_ON(pmd_trans_huge(*pmd));
2036
			return pte_alloc_map_lock(mm, pmd, addr, ptl);
2037
		}
2038 2039 2040 2041
	}
	return NULL;
}

2042 2043 2044 2045 2046 2047 2048
/*
 * This is the old fallback for page remapping.
 *
 * For historical reasons, it only allows reserved pages. Only
 * old drivers should use this, and they needed to mark their
 * pages reserved for the old functions anyway.
 */
N
Nick Piggin 已提交
2049 2050
static int insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page, pgprot_t prot)
2051
{
N
Nick Piggin 已提交
2052
	struct mm_struct *mm = vma->vm_mm;
2053
	int retval;
2054
	pte_t *pte;
2055 2056
	spinlock_t *ptl;

2057
	retval = -EINVAL;
2058
	if (PageAnon(page))
2059
		goto out;
2060 2061
	retval = -ENOMEM;
	flush_dcache_page(page);
2062
	pte = get_locked_pte(mm, addr, &ptl);
2063
	if (!pte)
2064
		goto out;
2065 2066 2067 2068 2069 2070
	retval = -EBUSY;
	if (!pte_none(*pte))
		goto out_unlock;

	/* Ok, finally just insert the thing.. */
	get_page(page);
2071
	inc_mm_counter_fast(mm, MM_FILEPAGES);
2072 2073 2074 2075
	page_add_file_rmap(page);
	set_pte_at(mm, addr, pte, mk_pte(page, prot));

	retval = 0;
2076 2077
	pte_unmap_unlock(pte, ptl);
	return retval;
2078 2079 2080 2081 2082 2083
out_unlock:
	pte_unmap_unlock(pte, ptl);
out:
	return retval;
}

2084 2085 2086 2087 2088 2089
/**
 * vm_insert_page - insert single page into user vma
 * @vma: user vma to map to
 * @addr: target user address of this page
 * @page: source kernel page
 *
2090 2091 2092 2093 2094 2095
 * This allows drivers to insert individual pages they've allocated
 * into a user vma.
 *
 * The page has to be a nice clean _individual_ kernel allocation.
 * If you allocate a compound page, you need to have marked it as
 * such (__GFP_COMP), or manually just split the page up yourself
N
Nick Piggin 已提交
2096
 * (see split_page()).
2097 2098 2099 2100 2101 2102 2103 2104
 *
 * NOTE! Traditionally this was done with "remap_pfn_range()" which
 * took an arbitrary page protection parameter. This doesn't allow
 * that. Your vma protection will have to be set up correctly, which
 * means that if you want a shared writable mapping, you'd better
 * ask for a shared writable mapping!
 *
 * The page does not need to be reserved.
2105 2106 2107 2108 2109
 *
 * Usually this function is called from f_op->mmap() handler
 * under mm->mmap_sem write-lock, so it can change vma->vm_flags.
 * Caller must set VM_MIXEDMAP on vma if it wants to call this
 * function from other places, for example from page-fault handler.
2110
 */
N
Nick Piggin 已提交
2111 2112
int vm_insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page)
2113 2114 2115 2116 2117
{
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
	if (!page_count(page))
		return -EINVAL;
2118 2119 2120 2121 2122
	if (!(vma->vm_flags & VM_MIXEDMAP)) {
		BUG_ON(down_read_trylock(&vma->vm_mm->mmap_sem));
		BUG_ON(vma->vm_flags & VM_PFNMAP);
		vma->vm_flags |= VM_MIXEDMAP;
	}
N
Nick Piggin 已提交
2123
	return insert_page(vma, addr, page, vma->vm_page_prot);
2124
}
2125
EXPORT_SYMBOL(vm_insert_page);
2126

N
Nick Piggin 已提交
2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145
static int insert_pfn(struct vm_area_struct *vma, unsigned long addr,
			unsigned long pfn, pgprot_t prot)
{
	struct mm_struct *mm = vma->vm_mm;
	int retval;
	pte_t *pte, entry;
	spinlock_t *ptl;

	retval = -ENOMEM;
	pte = get_locked_pte(mm, addr, &ptl);
	if (!pte)
		goto out;
	retval = -EBUSY;
	if (!pte_none(*pte))
		goto out_unlock;

	/* Ok, finally just insert the thing.. */
	entry = pte_mkspecial(pfn_pte(pfn, prot));
	set_pte_at(mm, addr, pte, entry);
2146
	update_mmu_cache(vma, addr, pte); /* XXX: why not for insert_page? */
N
Nick Piggin 已提交
2147 2148 2149 2150 2151 2152 2153 2154

	retval = 0;
out_unlock:
	pte_unmap_unlock(pte, ptl);
out:
	return retval;
}

N
Nick Piggin 已提交
2155 2156 2157 2158 2159 2160
/**
 * vm_insert_pfn - insert single pfn into user vma
 * @vma: user vma to map to
 * @addr: target user address of this page
 * @pfn: source kernel pfn
 *
2161
 * Similar to vm_insert_page, this allows drivers to insert individual pages
N
Nick Piggin 已提交
2162 2163 2164 2165
 * they've allocated into a user vma. Same comments apply.
 *
 * This function should only be called from a vm_ops->fault handler, and
 * in that case the handler should return NULL.
N
Nick Piggin 已提交
2166 2167 2168 2169 2170
 *
 * vma cannot be a COW mapping.
 *
 * As this is called only for pages that do not currently exist, we
 * do not need to flush old virtual caches or the TLB.
N
Nick Piggin 已提交
2171 2172
 */
int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
N
Nick Piggin 已提交
2173
			unsigned long pfn)
N
Nick Piggin 已提交
2174
{
2175
	int ret;
2176
	pgprot_t pgprot = vma->vm_page_prot;
N
Nick Piggin 已提交
2177 2178 2179 2180 2181 2182
	/*
	 * Technically, architectures with pte_special can avoid all these
	 * restrictions (same for remap_pfn_range).  However we would like
	 * consistency in testing and feature parity among all, so we should
	 * try to keep these invariants in place for everybody.
	 */
J
Jared Hulbert 已提交
2183 2184 2185 2186 2187
	BUG_ON(!(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)));
	BUG_ON((vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)) ==
						(VM_PFNMAP|VM_MIXEDMAP));
	BUG_ON((vma->vm_flags & VM_PFNMAP) && is_cow_mapping(vma->vm_flags));
	BUG_ON((vma->vm_flags & VM_MIXEDMAP) && pfn_valid(pfn));
N
Nick Piggin 已提交
2188

N
Nick Piggin 已提交
2189 2190
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
2191
	if (track_pfn_insert(vma, &pgprot, pfn))
2192 2193
		return -EINVAL;

2194
	ret = insert_pfn(vma, addr, pfn, pgprot);
2195 2196

	return ret;
N
Nick Piggin 已提交
2197 2198
}
EXPORT_SYMBOL(vm_insert_pfn);
N
Nick Piggin 已提交
2199

N
Nick Piggin 已提交
2200 2201 2202 2203
int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
			unsigned long pfn)
{
	BUG_ON(!(vma->vm_flags & VM_MIXEDMAP));
N
Nick Piggin 已提交
2204

N
Nick Piggin 已提交
2205 2206
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
N
Nick Piggin 已提交
2207

N
Nick Piggin 已提交
2208 2209 2210 2211
	/*
	 * If we don't have pte special, then we have to use the pfn_valid()
	 * based VM_MIXEDMAP scheme (see vm_normal_page), and thus we *must*
	 * refcount the page if pfn_valid is true (hence insert_page rather
H
Hugh Dickins 已提交
2212 2213
	 * than insert_pfn).  If a zero_pfn were inserted into a VM_MIXEDMAP
	 * without pte special, it would there be refcounted as a normal page.
N
Nick Piggin 已提交
2214 2215 2216 2217 2218 2219 2220 2221
	 */
	if (!HAVE_PTE_SPECIAL && pfn_valid(pfn)) {
		struct page *page;

		page = pfn_to_page(pfn);
		return insert_page(vma, addr, page, vma->vm_page_prot);
	}
	return insert_pfn(vma, addr, pfn, vma->vm_page_prot);
N
Nick Piggin 已提交
2222
}
N
Nick Piggin 已提交
2223
EXPORT_SYMBOL(vm_insert_mixed);
N
Nick Piggin 已提交
2224

L
Linus Torvalds 已提交
2225 2226 2227 2228 2229 2230 2231 2232 2233 2234
/*
 * maps a range of physical memory into the requested pages. the old
 * mappings are removed. any references to nonexistent pages results
 * in null mappings (currently treated as "copy-on-access")
 */
static int remap_pte_range(struct mm_struct *mm, pmd_t *pmd,
			unsigned long addr, unsigned long end,
			unsigned long pfn, pgprot_t prot)
{
	pte_t *pte;
H
Hugh Dickins 已提交
2235
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2236

H
Hugh Dickins 已提交
2237
	pte = pte_alloc_map_lock(mm, pmd, addr, &ptl);
L
Linus Torvalds 已提交
2238 2239
	if (!pte)
		return -ENOMEM;
2240
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
2241 2242
	do {
		BUG_ON(!pte_none(*pte));
N
Nick Piggin 已提交
2243
		set_pte_at(mm, addr, pte, pte_mkspecial(pfn_pte(pfn, prot)));
L
Linus Torvalds 已提交
2244 2245
		pfn++;
	} while (pte++, addr += PAGE_SIZE, addr != end);
2246
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
2247
	pte_unmap_unlock(pte - 1, ptl);
L
Linus Torvalds 已提交
2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261
	return 0;
}

static inline int remap_pmd_range(struct mm_struct *mm, pud_t *pud,
			unsigned long addr, unsigned long end,
			unsigned long pfn, pgprot_t prot)
{
	pmd_t *pmd;
	unsigned long next;

	pfn -= addr >> PAGE_SHIFT;
	pmd = pmd_alloc(mm, pud, addr);
	if (!pmd)
		return -ENOMEM;
2262
	VM_BUG_ON(pmd_trans_huge(*pmd));
L
Linus Torvalds 已提交
2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291
	do {
		next = pmd_addr_end(addr, end);
		if (remap_pte_range(mm, pmd, addr, next,
				pfn + (addr >> PAGE_SHIFT), prot))
			return -ENOMEM;
	} while (pmd++, addr = next, addr != end);
	return 0;
}

static inline int remap_pud_range(struct mm_struct *mm, pgd_t *pgd,
			unsigned long addr, unsigned long end,
			unsigned long pfn, pgprot_t prot)
{
	pud_t *pud;
	unsigned long next;

	pfn -= addr >> PAGE_SHIFT;
	pud = pud_alloc(mm, pgd, addr);
	if (!pud)
		return -ENOMEM;
	do {
		next = pud_addr_end(addr, end);
		if (remap_pmd_range(mm, pud, addr, next,
				pfn + (addr >> PAGE_SHIFT), prot))
			return -ENOMEM;
	} while (pud++, addr = next, addr != end);
	return 0;
}

2292 2293 2294 2295 2296 2297 2298 2299 2300 2301
/**
 * remap_pfn_range - remap kernel memory to userspace
 * @vma: user vma to map to
 * @addr: target user address to start at
 * @pfn: physical address of kernel memory
 * @size: size of map area
 * @prot: page protection flags for this mapping
 *
 *  Note: this is only safe if the mm semaphore is held when called.
 */
L
Linus Torvalds 已提交
2302 2303 2304 2305 2306
int remap_pfn_range(struct vm_area_struct *vma, unsigned long addr,
		    unsigned long pfn, unsigned long size, pgprot_t prot)
{
	pgd_t *pgd;
	unsigned long next;
2307
	unsigned long end = addr + PAGE_ALIGN(size);
L
Linus Torvalds 已提交
2308 2309 2310 2311 2312 2313 2314 2315
	struct mm_struct *mm = vma->vm_mm;
	int err;

	/*
	 * Physically remapped pages are special. Tell the
	 * rest of the world about it:
	 *   VM_IO tells people not to look at these pages
	 *	(accesses can have side effects).
2316 2317 2318
	 *   VM_PFNMAP tells the core MM that the base pages are just
	 *	raw PFN mappings, and do not have a "struct page" associated
	 *	with them.
2319 2320 2321 2322
	 *   VM_DONTEXPAND
	 *      Disable vma merging and expanding with mremap().
	 *   VM_DONTDUMP
	 *      Omit vma from core dump, even when VM_IO turned off.
L
Linus Torvalds 已提交
2323 2324 2325 2326
	 *
	 * There's a horrible special case to handle copy-on-write
	 * behaviour that some programs depend on. We mark the "original"
	 * un-COW'ed pages by matching them up with "vma->vm_pgoff".
2327
	 * See vm_normal_page() for details.
L
Linus Torvalds 已提交
2328
	 */
2329 2330 2331
	if (is_cow_mapping(vma->vm_flags)) {
		if (addr != vma->vm_start || end != vma->vm_end)
			return -EINVAL;
L
Linus Torvalds 已提交
2332
		vma->vm_pgoff = pfn;
2333 2334 2335 2336
	}

	err = track_pfn_remap(vma, &prot, pfn, addr, PAGE_ALIGN(size));
	if (err)
2337
		return -EINVAL;
L
Linus Torvalds 已提交
2338

2339
	vma->vm_flags |= VM_IO | VM_PFNMAP | VM_DONTEXPAND | VM_DONTDUMP;
L
Linus Torvalds 已提交
2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351

	BUG_ON(addr >= end);
	pfn -= addr >> PAGE_SHIFT;
	pgd = pgd_offset(mm, addr);
	flush_cache_range(vma, addr, end);
	do {
		next = pgd_addr_end(addr, end);
		err = remap_pud_range(mm, pgd, addr, next,
				pfn + (addr >> PAGE_SHIFT), prot);
		if (err)
			break;
	} while (pgd++, addr = next, addr != end);
2352 2353

	if (err)
2354
		untrack_pfn(vma, pfn, PAGE_ALIGN(size));
2355

L
Linus Torvalds 已提交
2356 2357 2358 2359
	return err;
}
EXPORT_SYMBOL(remap_pfn_range);

2360 2361 2362 2363 2364 2365
static int apply_to_pte_range(struct mm_struct *mm, pmd_t *pmd,
				     unsigned long addr, unsigned long end,
				     pte_fn_t fn, void *data)
{
	pte_t *pte;
	int err;
2366
	pgtable_t token;
2367
	spinlock_t *uninitialized_var(ptl);
2368 2369 2370 2371 2372 2373 2374 2375 2376

	pte = (mm == &init_mm) ?
		pte_alloc_kernel(pmd, addr) :
		pte_alloc_map_lock(mm, pmd, addr, &ptl);
	if (!pte)
		return -ENOMEM;

	BUG_ON(pmd_huge(*pmd));

2377 2378
	arch_enter_lazy_mmu_mode();

2379
	token = pmd_pgtable(*pmd);
2380 2381

	do {
2382
		err = fn(pte++, token, addr, data);
2383 2384
		if (err)
			break;
2385
	} while (addr += PAGE_SIZE, addr != end);
2386

2387 2388
	arch_leave_lazy_mmu_mode();

2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401
	if (mm != &init_mm)
		pte_unmap_unlock(pte-1, ptl);
	return err;
}

static int apply_to_pmd_range(struct mm_struct *mm, pud_t *pud,
				     unsigned long addr, unsigned long end,
				     pte_fn_t fn, void *data)
{
	pmd_t *pmd;
	unsigned long next;
	int err;

A
Andi Kleen 已提交
2402 2403
	BUG_ON(pud_huge(*pud));

2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444
	pmd = pmd_alloc(mm, pud, addr);
	if (!pmd)
		return -ENOMEM;
	do {
		next = pmd_addr_end(addr, end);
		err = apply_to_pte_range(mm, pmd, addr, next, fn, data);
		if (err)
			break;
	} while (pmd++, addr = next, addr != end);
	return err;
}

static int apply_to_pud_range(struct mm_struct *mm, pgd_t *pgd,
				     unsigned long addr, unsigned long end,
				     pte_fn_t fn, void *data)
{
	pud_t *pud;
	unsigned long next;
	int err;

	pud = pud_alloc(mm, pgd, addr);
	if (!pud)
		return -ENOMEM;
	do {
		next = pud_addr_end(addr, end);
		err = apply_to_pmd_range(mm, pud, addr, next, fn, data);
		if (err)
			break;
	} while (pud++, addr = next, addr != end);
	return err;
}

/*
 * Scan a region of virtual memory, filling in page tables as necessary
 * and calling a provided function on each leaf page table.
 */
int apply_to_page_range(struct mm_struct *mm, unsigned long addr,
			unsigned long size, pte_fn_t fn, void *data)
{
	pgd_t *pgd;
	unsigned long next;
2445
	unsigned long end = addr + size;
2446 2447 2448 2449 2450 2451 2452 2453 2454 2455
	int err;

	BUG_ON(addr >= end);
	pgd = pgd_offset(mm, addr);
	do {
		next = pgd_addr_end(addr, end);
		err = apply_to_pud_range(mm, pgd, addr, next, fn, data);
		if (err)
			break;
	} while (pgd++, addr = next, addr != end);
2456

2457 2458 2459 2460
	return err;
}
EXPORT_SYMBOL_GPL(apply_to_page_range);

2461 2462 2463 2464
/*
 * handle_pte_fault chooses page fault handler according to an entry
 * which was read non-atomically.  Before making any commitment, on
 * those architectures or configurations (e.g. i386 with PAE) which
2465
 * might give a mix of unmatched parts, do_swap_page and do_nonlinear_fault
2466 2467
 * must check under lock before unmapping the pte and proceeding
 * (but do_wp_page is only called after already making such a check;
2468
 * and do_anonymous_page can safely check later on).
2469
 */
H
Hugh Dickins 已提交
2470
static inline int pte_unmap_same(struct mm_struct *mm, pmd_t *pmd,
2471 2472 2473 2474 2475
				pte_t *page_table, pte_t orig_pte)
{
	int same = 1;
#if defined(CONFIG_SMP) || defined(CONFIG_PREEMPT)
	if (sizeof(pte_t) > sizeof(unsigned long)) {
H
Hugh Dickins 已提交
2476 2477
		spinlock_t *ptl = pte_lockptr(mm, pmd);
		spin_lock(ptl);
2478
		same = pte_same(*page_table, orig_pte);
H
Hugh Dickins 已提交
2479
		spin_unlock(ptl);
2480 2481 2482 2483 2484 2485
	}
#endif
	pte_unmap(page_table);
	return same;
}

2486
static inline void cow_user_page(struct page *dst, struct page *src, unsigned long va, struct vm_area_struct *vma)
2487 2488 2489 2490 2491 2492 2493 2494
{
	/*
	 * If the source page was a PFN mapping, we don't have
	 * a "struct page" for it. We do a best-effort copy by
	 * just copying from the original user address. If that
	 * fails, we just zero-fill it. Live with it.
	 */
	if (unlikely(!src)) {
2495
		void *kaddr = kmap_atomic(dst);
L
Linus Torvalds 已提交
2496 2497 2498 2499 2500 2501 2502 2503 2504
		void __user *uaddr = (void __user *)(va & PAGE_MASK);

		/*
		 * This really shouldn't fail, because the page is there
		 * in the page tables. But it might just be unreadable,
		 * in which case we just give up and fill the result with
		 * zeroes.
		 */
		if (__copy_from_user_inatomic(kaddr, uaddr, PAGE_SIZE))
2505
			clear_page(kaddr);
2506
		kunmap_atomic(kaddr);
2507
		flush_dcache_page(dst);
N
Nick Piggin 已提交
2508 2509
	} else
		copy_user_highpage(dst, src, va, vma);
2510 2511
}

L
Linus Torvalds 已提交
2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525
/*
 * This routine handles present pages, when users try to write
 * to a shared page. It is done by copying the page to a new address
 * and decrementing the shared-page counter for the old page.
 *
 * Note that this routine assumes that the protection checks have been
 * done by the caller (the low-level page fault routine in most cases).
 * Thus we can safely just mark it writable once we've done any necessary
 * COW.
 *
 * We also mark the page dirty at this point even though the page will
 * change only once the write actually happens. This avoids a few races,
 * and potentially makes it more efficient.
 *
2526 2527 2528
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
 * but allow concurrent faults), with pte both mapped and locked.
 * We return with mmap_sem still held, but pte unmapped and unlocked.
L
Linus Torvalds 已提交
2529
 */
2530 2531
static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2532
		spinlock_t *ptl, pte_t orig_pte)
2533
	__releases(ptl)
L
Linus Torvalds 已提交
2534
{
2535
	struct page *old_page, *new_page = NULL;
L
Linus Torvalds 已提交
2536
	pte_t entry;
2537
	int ret = 0;
2538
	int page_mkwrite = 0;
2539
	struct page *dirty_page = NULL;
2540 2541
	unsigned long mmun_start = 0;	/* For mmu_notifiers */
	unsigned long mmun_end = 0;	/* For mmu_notifiers */
L
Linus Torvalds 已提交
2542

2543
	old_page = vm_normal_page(vma, address, orig_pte);
2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554
	if (!old_page) {
		/*
		 * VM_MIXEDMAP !pfn_valid() case
		 *
		 * We should not cow pages in a shared writeable mapping.
		 * Just mark the pages writable as we can't do any dirty
		 * accounting on raw pfn maps.
		 */
		if ((vma->vm_flags & (VM_WRITE|VM_SHARED)) ==
				     (VM_WRITE|VM_SHARED))
			goto reuse;
2555
		goto gotten;
2556
	}
L
Linus Torvalds 已提交
2557

2558
	/*
P
Peter Zijlstra 已提交
2559 2560
	 * Take out anonymous pages first, anonymous shared vmas are
	 * not dirty accountable.
2561
	 */
H
Hugh Dickins 已提交
2562
	if (PageAnon(old_page) && !PageKsm(old_page)) {
2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573
		if (!trylock_page(old_page)) {
			page_cache_get(old_page);
			pte_unmap_unlock(page_table, ptl);
			lock_page(old_page);
			page_table = pte_offset_map_lock(mm, pmd, address,
							 &ptl);
			if (!pte_same(*page_table, orig_pte)) {
				unlock_page(old_page);
				goto unlock;
			}
			page_cache_release(old_page);
P
Peter Zijlstra 已提交
2574
		}
2575
		if (reuse_swap_page(old_page)) {
2576 2577 2578 2579 2580 2581
			/*
			 * The page is all ours.  Move it to our anon_vma so
			 * the rmap code will not search our parent or siblings.
			 * Protected against the rmap code by the page lock.
			 */
			page_move_anon_rmap(old_page, vma, address);
2582 2583 2584
			unlock_page(old_page);
			goto reuse;
		}
2585
		unlock_page(old_page);
P
Peter Zijlstra 已提交
2586
	} else if (unlikely((vma->vm_flags & (VM_WRITE|VM_SHARED)) ==
2587
					(VM_WRITE|VM_SHARED))) {
P
Peter Zijlstra 已提交
2588 2589 2590 2591 2592
		/*
		 * Only catch write-faults on shared writable pages,
		 * read-only shared pages can get COWed by
		 * get_user_pages(.write=1, .force=1).
		 */
2593
		if (vma->vm_ops && vma->vm_ops->page_mkwrite) {
2594 2595 2596 2597 2598 2599 2600 2601 2602
			struct vm_fault vmf;
			int tmp;

			vmf.virtual_address = (void __user *)(address &
								PAGE_MASK);
			vmf.pgoff = old_page->index;
			vmf.flags = FAULT_FLAG_WRITE|FAULT_FLAG_MKWRITE;
			vmf.page = old_page;

2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613
			/*
			 * Notify the address space that the page is about to
			 * become writable so that it can prohibit this or wait
			 * for the page to get into an appropriate state.
			 *
			 * We do this without the lock held, so that it can
			 * sleep if it needs to.
			 */
			page_cache_get(old_page);
			pte_unmap_unlock(page_table, ptl);

2614 2615 2616 2617
			tmp = vma->vm_ops->page_mkwrite(vma, &vmf);
			if (unlikely(tmp &
					(VM_FAULT_ERROR | VM_FAULT_NOPAGE))) {
				ret = tmp;
2618
				goto unwritable_page;
2619
			}
N
Nick Piggin 已提交
2620 2621 2622 2623 2624 2625 2626 2627 2628
			if (unlikely(!(tmp & VM_FAULT_LOCKED))) {
				lock_page(old_page);
				if (!old_page->mapping) {
					ret = 0; /* retry the fault */
					unlock_page(old_page);
					goto unwritable_page;
				}
			} else
				VM_BUG_ON(!PageLocked(old_page));
2629 2630 2631 2632 2633 2634 2635 2636 2637

			/*
			 * Since we dropped the lock we need to revalidate
			 * the PTE as someone else may have changed it.  If
			 * they did, we just return, as we can count on the
			 * MMU to tell us if they didn't also make it writable.
			 */
			page_table = pte_offset_map_lock(mm, pmd, address,
							 &ptl);
N
Nick Piggin 已提交
2638 2639
			if (!pte_same(*page_table, orig_pte)) {
				unlock_page(old_page);
2640
				goto unlock;
N
Nick Piggin 已提交
2641
			}
2642 2643

			page_mkwrite = 1;
L
Linus Torvalds 已提交
2644
		}
2645 2646
		dirty_page = old_page;
		get_page(dirty_page);
2647

2648
reuse:
2649 2650 2651
		flush_cache_page(vma, address, pte_pfn(orig_pte));
		entry = pte_mkyoung(orig_pte);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
2652
		if (ptep_set_access_flags(vma, address, page_table, entry,1))
2653
			update_mmu_cache(vma, address, page_table);
2654
		pte_unmap_unlock(page_table, ptl);
2655
		ret |= VM_FAULT_WRITE;
2656 2657 2658 2659 2660 2661 2662 2663 2664 2665

		if (!dirty_page)
			return ret;

		/*
		 * Yes, Virginia, this is actually required to prevent a race
		 * with clear_page_dirty_for_io() from clearing the page dirty
		 * bit after it clear all dirty ptes, but before a racing
		 * do_wp_page installs a dirty pte.
		 *
2666
		 * __do_fault is protected similarly.
2667 2668 2669 2670
		 */
		if (!page_mkwrite) {
			wait_on_page_locked(dirty_page);
			set_page_dirty_balance(dirty_page, page_mkwrite);
2671 2672 2673
			/* file_update_time outside page_lock */
			if (vma->vm_file)
				file_update_time(vma->vm_file);
2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691
		}
		put_page(dirty_page);
		if (page_mkwrite) {
			struct address_space *mapping = dirty_page->mapping;

			set_page_dirty(dirty_page);
			unlock_page(dirty_page);
			page_cache_release(dirty_page);
			if (mapping)	{
				/*
				 * Some device drivers do not set page.mapping
				 * but still dirty their pages
				 */
				balance_dirty_pages_ratelimited(mapping);
			}
		}

		return ret;
L
Linus Torvalds 已提交
2692 2693 2694 2695 2696
	}

	/*
	 * Ok, we need to copy. Oh, well..
	 */
N
Nick Piggin 已提交
2697
	page_cache_get(old_page);
H
Hugh Dickins 已提交
2698
gotten:
2699
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
2700 2701

	if (unlikely(anon_vma_prepare(vma)))
2702
		goto oom;
H
Hugh Dickins 已提交
2703

H
Hugh Dickins 已提交
2704
	if (is_zero_pfn(pte_pfn(orig_pte))) {
H
Hugh Dickins 已提交
2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715
		new_page = alloc_zeroed_user_highpage_movable(vma, address);
		if (!new_page)
			goto oom;
	} else {
		new_page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address);
		if (!new_page)
			goto oom;
		cow_user_page(new_page, old_page, address, vma);
	}
	__SetPageUptodate(new_page);

K
KAMEZAWA Hiroyuki 已提交
2716
	if (mem_cgroup_newpage_charge(new_page, mm, GFP_KERNEL))
2717 2718
		goto oom_free_new;

2719
	mmun_start  = address & PAGE_MASK;
2720
	mmun_end    = mmun_start + PAGE_SIZE;
2721 2722
	mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);

L
Linus Torvalds 已提交
2723 2724 2725
	/*
	 * Re-check the pte - we dropped the lock
	 */
2726
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
2727
	if (likely(pte_same(*page_table, orig_pte))) {
H
Hugh Dickins 已提交
2728 2729
		if (old_page) {
			if (!PageAnon(old_page)) {
2730 2731
				dec_mm_counter_fast(mm, MM_FILEPAGES);
				inc_mm_counter_fast(mm, MM_ANONPAGES);
H
Hugh Dickins 已提交
2732 2733
			}
		} else
2734
			inc_mm_counter_fast(mm, MM_ANONPAGES);
2735
		flush_cache_page(vma, address, pte_pfn(orig_pte));
2736 2737
		entry = mk_pte(new_page, vma->vm_page_prot);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
2738 2739 2740 2741 2742 2743
		/*
		 * Clear the pte entry and flush it first, before updating the
		 * pte with the new entry. This will avoid a race condition
		 * seen in the presence of one thread doing SMC and another
		 * thread doing COW.
		 */
2744
		ptep_clear_flush(vma, address, page_table);
N
Nick Piggin 已提交
2745
		page_add_new_anon_rmap(new_page, vma, address);
2746 2747 2748 2749 2750 2751
		/*
		 * We call the notify macro here because, when using secondary
		 * mmu page tables (such as kvm shadow page tables), we want the
		 * new page to be mapped directly into the secondary page table.
		 */
		set_pte_at_notify(mm, address, page_table, entry);
2752
		update_mmu_cache(vma, address, page_table);
N
Nick Piggin 已提交
2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775
		if (old_page) {
			/*
			 * Only after switching the pte to the new page may
			 * we remove the mapcount here. Otherwise another
			 * process may come and find the rmap count decremented
			 * before the pte is switched to the new page, and
			 * "reuse" the old page writing into it while our pte
			 * here still points into it and can be read by other
			 * threads.
			 *
			 * The critical issue is to order this
			 * page_remove_rmap with the ptp_clear_flush above.
			 * Those stores are ordered by (if nothing else,)
			 * the barrier present in the atomic_add_negative
			 * in page_remove_rmap.
			 *
			 * Then the TLB flush in ptep_clear_flush ensures that
			 * no process can access the old page before the
			 * decremented mapcount is visible. And the old page
			 * cannot be reused until after the decremented
			 * mapcount is visible. So transitively, TLBs to
			 * old page will be flushed before it can be reused.
			 */
2776
			page_remove_rmap(old_page);
N
Nick Piggin 已提交
2777 2778
		}

L
Linus Torvalds 已提交
2779 2780
		/* Free the old page.. */
		new_page = old_page;
N
Nick Piggin 已提交
2781
		ret |= VM_FAULT_WRITE;
2782 2783 2784
	} else
		mem_cgroup_uncharge_page(new_page);

2785 2786
	if (new_page)
		page_cache_release(new_page);
2787
unlock:
2788
	pte_unmap_unlock(page_table, ptl);
2789
	if (mmun_end > mmun_start)
2790
		mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802
	if (old_page) {
		/*
		 * Don't let another task, with possibly unlocked vma,
		 * keep the mlocked page.
		 */
		if ((ret & VM_FAULT_WRITE) && (vma->vm_flags & VM_LOCKED)) {
			lock_page(old_page);	/* LRU manipulation */
			munlock_vma_page(old_page);
			unlock_page(old_page);
		}
		page_cache_release(old_page);
	}
N
Nick Piggin 已提交
2803
	return ret;
2804
oom_free_new:
2805
	page_cache_release(new_page);
2806
oom:
2807
	if (old_page)
H
Hugh Dickins 已提交
2808
		page_cache_release(old_page);
L
Linus Torvalds 已提交
2809
	return VM_FAULT_OOM;
2810 2811 2812

unwritable_page:
	page_cache_release(old_page);
2813
	return ret;
L
Linus Torvalds 已提交
2814 2815
}

2816
static void unmap_mapping_range_vma(struct vm_area_struct *vma,
L
Linus Torvalds 已提交
2817 2818 2819
		unsigned long start_addr, unsigned long end_addr,
		struct zap_details *details)
{
2820
	zap_page_range_single(vma, start_addr, end_addr - start_addr, details);
L
Linus Torvalds 已提交
2821 2822
}

2823
static inline void unmap_mapping_range_tree(struct rb_root *root,
L
Linus Torvalds 已提交
2824 2825 2826 2827 2828
					    struct zap_details *details)
{
	struct vm_area_struct *vma;
	pgoff_t vba, vea, zba, zea;

2829
	vma_interval_tree_foreach(vma, root,
L
Linus Torvalds 已提交
2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841
			details->first_index, details->last_index) {

		vba = vma->vm_pgoff;
		vea = vba + ((vma->vm_end - vma->vm_start) >> PAGE_SHIFT) - 1;
		/* Assume for now that PAGE_CACHE_SHIFT == PAGE_SHIFT */
		zba = details->first_index;
		if (zba < vba)
			zba = vba;
		zea = details->last_index;
		if (zea > vea)
			zea = vea;

2842
		unmap_mapping_range_vma(vma,
L
Linus Torvalds 已提交
2843 2844
			((zba - vba) << PAGE_SHIFT) + vma->vm_start,
			((zea - vba + 1) << PAGE_SHIFT) + vma->vm_start,
2845
				details);
L
Linus Torvalds 已提交
2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859
	}
}

static inline void unmap_mapping_range_list(struct list_head *head,
					    struct zap_details *details)
{
	struct vm_area_struct *vma;

	/*
	 * In nonlinear VMAs there is no correspondence between virtual address
	 * offset and file offset.  So we must perform an exhaustive search
	 * across *all* the pages in each nonlinear VMA, not just the pages
	 * whose virtual address lies outside the file truncation point.
	 */
2860
	list_for_each_entry(vma, head, shared.nonlinear) {
L
Linus Torvalds 已提交
2861
		details->nonlinear_vma = vma;
2862
		unmap_mapping_range_vma(vma, vma->vm_start, vma->vm_end, details);
L
Linus Torvalds 已提交
2863 2864 2865 2866
	}
}

/**
2867
 * unmap_mapping_range - unmap the portion of all mmaps in the specified address_space corresponding to the specified page range in the underlying file.
M
Martin Waitz 已提交
2868
 * @mapping: the address space containing mmaps to be unmapped.
L
Linus Torvalds 已提交
2869 2870
 * @holebegin: byte in first page to unmap, relative to the start of
 * the underlying file.  This will be rounded down to a PAGE_SIZE
N
npiggin@suse.de 已提交
2871
 * boundary.  Note that this is different from truncate_pagecache(), which
L
Linus Torvalds 已提交
2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902
 * must keep the partial page.  In contrast, we must get rid of
 * partial pages.
 * @holelen: size of prospective hole in bytes.  This will be rounded
 * up to a PAGE_SIZE boundary.  A holelen of zero truncates to the
 * end of the file.
 * @even_cows: 1 when truncating a file, unmap even private COWed pages;
 * but 0 when invalidating pagecache, don't throw away private data.
 */
void unmap_mapping_range(struct address_space *mapping,
		loff_t const holebegin, loff_t const holelen, int even_cows)
{
	struct zap_details details;
	pgoff_t hba = holebegin >> PAGE_SHIFT;
	pgoff_t hlen = (holelen + PAGE_SIZE - 1) >> PAGE_SHIFT;

	/* Check for overflow. */
	if (sizeof(holelen) > sizeof(hlen)) {
		long long holeend =
			(holebegin + holelen + PAGE_SIZE - 1) >> PAGE_SHIFT;
		if (holeend & ~(long long)ULONG_MAX)
			hlen = ULONG_MAX - hba + 1;
	}

	details.check_mapping = even_cows? NULL: mapping;
	details.nonlinear_vma = NULL;
	details.first_index = hba;
	details.last_index = hba + hlen - 1;
	if (details.last_index < details.first_index)
		details.last_index = ULONG_MAX;


2903
	mutex_lock(&mapping->i_mmap_mutex);
2904
	if (unlikely(!RB_EMPTY_ROOT(&mapping->i_mmap)))
L
Linus Torvalds 已提交
2905 2906 2907
		unmap_mapping_range_tree(&mapping->i_mmap, &details);
	if (unlikely(!list_empty(&mapping->i_mmap_nonlinear)))
		unmap_mapping_range_list(&mapping->i_mmap_nonlinear, &details);
2908
	mutex_unlock(&mapping->i_mmap_mutex);
L
Linus Torvalds 已提交
2909 2910 2911 2912
}
EXPORT_SYMBOL(unmap_mapping_range);

/*
2913 2914 2915
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
 * but allow concurrent faults), and pte mapped but not yet locked.
 * We return with mmap_sem still held, but pte unmapped and unlocked.
L
Linus Torvalds 已提交
2916
 */
2917 2918
static int do_swap_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2919
		unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
2920
{
2921
	spinlock_t *ptl;
A
Andrea Arcangeli 已提交
2922
	struct page *page, *swapcache = NULL;
2923
	swp_entry_t entry;
L
Linus Torvalds 已提交
2924
	pte_t pte;
2925
	int locked;
2926
	struct mem_cgroup *ptr;
2927
	int exclusive = 0;
N
Nick Piggin 已提交
2928
	int ret = 0;
L
Linus Torvalds 已提交
2929

H
Hugh Dickins 已提交
2930
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
2931
		goto out;
2932 2933

	entry = pte_to_swp_entry(orig_pte);
2934 2935 2936 2937 2938 2939 2940
	if (unlikely(non_swap_entry(entry))) {
		if (is_migration_entry(entry)) {
			migration_entry_wait(mm, pmd, address);
		} else if (is_hwpoison_entry(entry)) {
			ret = VM_FAULT_HWPOISON;
		} else {
			print_bad_pte(vma, address, orig_pte, NULL);
H
Hugh Dickins 已提交
2941
			ret = VM_FAULT_SIGBUS;
2942
		}
2943 2944
		goto out;
	}
2945
	delayacct_set_flag(DELAYACCT_PF_SWAPIN);
L
Linus Torvalds 已提交
2946 2947
	page = lookup_swap_cache(entry);
	if (!page) {
2948 2949
		page = swapin_readahead(entry,
					GFP_HIGHUSER_MOVABLE, vma, address);
L
Linus Torvalds 已提交
2950 2951
		if (!page) {
			/*
2952 2953
			 * Back out if somebody else faulted in this pte
			 * while we released the pte lock.
L
Linus Torvalds 已提交
2954
			 */
2955
			page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
2956 2957
			if (likely(pte_same(*page_table, orig_pte)))
				ret = VM_FAULT_OOM;
2958
			delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2959
			goto unlock;
L
Linus Torvalds 已提交
2960 2961 2962 2963
		}

		/* Had to read the page from swap area: Major fault */
		ret = VM_FAULT_MAJOR;
2964
		count_vm_event(PGMAJFAULT);
2965
		mem_cgroup_count_vm_event(mm, PGMAJFAULT);
2966
	} else if (PageHWPoison(page)) {
2967 2968 2969 2970
		/*
		 * hwpoisoned dirty swapcache pages are kept for killing
		 * owner processes (which may be unknown at hwpoison time)
		 */
2971 2972
		ret = VM_FAULT_HWPOISON;
		delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2973
		goto out_release;
L
Linus Torvalds 已提交
2974 2975
	}

2976
	locked = lock_page_or_retry(page, mm, flags);
R
Rik van Riel 已提交
2977

2978
	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2979 2980 2981 2982
	if (!locked) {
		ret |= VM_FAULT_RETRY;
		goto out_release;
	}
2983

A
Andrea Arcangeli 已提交
2984
	/*
2985 2986 2987 2988
	 * Make sure try_to_free_swap or reuse_swap_page or swapoff did not
	 * release the swapcache from under us.  The page pin, and pte_same
	 * test below, are not enough to exclude that.  Even if it is still
	 * swapcache, we need to check that the page's swap has not changed.
A
Andrea Arcangeli 已提交
2989
	 */
2990
	if (unlikely(!PageSwapCache(page) || page_private(page) != entry.val))
A
Andrea Arcangeli 已提交
2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002
		goto out_page;

	if (ksm_might_need_to_copy(page, vma, address)) {
		swapcache = page;
		page = ksm_does_need_to_copy(page, vma, address);

		if (unlikely(!page)) {
			ret = VM_FAULT_OOM;
			page = swapcache;
			swapcache = NULL;
			goto out_page;
		}
H
Hugh Dickins 已提交
3003 3004
	}

K
KAMEZAWA Hiroyuki 已提交
3005
	if (mem_cgroup_try_charge_swapin(mm, page, GFP_KERNEL, &ptr)) {
3006
		ret = VM_FAULT_OOM;
3007
		goto out_page;
3008 3009
	}

L
Linus Torvalds 已提交
3010
	/*
3011
	 * Back out if somebody else already faulted in this pte.
L
Linus Torvalds 已提交
3012
	 */
3013
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
H
Hugh Dickins 已提交
3014
	if (unlikely(!pte_same(*page_table, orig_pte)))
3015 3016 3017 3018 3019
		goto out_nomap;

	if (unlikely(!PageUptodate(page))) {
		ret = VM_FAULT_SIGBUS;
		goto out_nomap;
L
Linus Torvalds 已提交
3020 3021
	}

3022 3023 3024 3025 3026 3027 3028 3029
	/*
	 * The page isn't present yet, go ahead with the fault.
	 *
	 * Be careful about the sequence of operations here.
	 * To get its accounting right, reuse_swap_page() must be called
	 * while the page is counted on swap but not yet in mapcount i.e.
	 * before page_add_anon_rmap() and swap_free(); try_to_free_swap()
	 * must be called after the swap_free(), or it will never succeed.
3030 3031 3032 3033
	 * Because delete_from_swap_page() may be called by reuse_swap_page(),
	 * mem_cgroup_commit_charge_swapin() may not be able to find swp_entry
	 * in page->private. In this case, a record in swap_cgroup  is silently
	 * discarded at swap_free().
3034
	 */
L
Linus Torvalds 已提交
3035

3036
	inc_mm_counter_fast(mm, MM_ANONPAGES);
K
KAMEZAWA Hiroyuki 已提交
3037
	dec_mm_counter_fast(mm, MM_SWAPENTS);
L
Linus Torvalds 已提交
3038
	pte = mk_pte(page, vma->vm_page_prot);
3039
	if ((flags & FAULT_FLAG_WRITE) && reuse_swap_page(page)) {
L
Linus Torvalds 已提交
3040
		pte = maybe_mkwrite(pte_mkdirty(pte), vma);
3041
		flags &= ~FAULT_FLAG_WRITE;
3042
		ret |= VM_FAULT_WRITE;
3043
		exclusive = 1;
L
Linus Torvalds 已提交
3044 3045 3046
	}
	flush_icache_page(vma, page);
	set_pte_at(mm, address, page_table, pte);
3047 3048 3049 3050
	if (swapcache) /* ksm created a completely new copy */
		page_add_new_anon_rmap(page, vma, address);
	else
		do_page_add_anon_rmap(page, vma, address, exclusive);
3051 3052
	/* It's better to call commit-charge after rmap is established */
	mem_cgroup_commit_charge_swapin(page, ptr);
L
Linus Torvalds 已提交
3053

3054
	swap_free(entry);
N
Nick Piggin 已提交
3055
	if (vm_swap_full() || (vma->vm_flags & VM_LOCKED) || PageMlocked(page))
3056
		try_to_free_swap(page);
3057
	unlock_page(page);
A
Andrea Arcangeli 已提交
3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069
	if (swapcache) {
		/*
		 * Hold the lock to avoid the swap entry to be reused
		 * until we take the PT lock for the pte_same() check
		 * (to avoid false positives from pte_same). For
		 * further safety release the lock after the swap_free
		 * so that the swap count won't change under a
		 * parallel locked swapcache.
		 */
		unlock_page(swapcache);
		page_cache_release(swapcache);
	}
3070

3071
	if (flags & FAULT_FLAG_WRITE) {
3072 3073 3074
		ret |= do_wp_page(mm, vma, address, page_table, pmd, ptl, pte);
		if (ret & VM_FAULT_ERROR)
			ret &= VM_FAULT_ERROR;
L
Linus Torvalds 已提交
3075 3076 3077 3078
		goto out;
	}

	/* No need to invalidate - it was non-present before */
3079
	update_mmu_cache(vma, address, page_table);
3080
unlock:
3081
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
3082 3083
out:
	return ret;
3084
out_nomap:
3085
	mem_cgroup_cancel_charge_swapin(ptr);
3086
	pte_unmap_unlock(page_table, ptl);
3087
out_page:
3088
	unlock_page(page);
3089
out_release:
3090
	page_cache_release(page);
A
Andrea Arcangeli 已提交
3091 3092 3093 3094
	if (swapcache) {
		unlock_page(swapcache);
		page_cache_release(swapcache);
	}
3095
	return ret;
L
Linus Torvalds 已提交
3096 3097
}

3098
/*
3099 3100
 * This is like a special single-page "expand_{down|up}wards()",
 * except we must first make sure that 'address{-|+}PAGE_SIZE'
3101 3102 3103 3104 3105 3106
 * doesn't hit another vma.
 */
static inline int check_stack_guard_page(struct vm_area_struct *vma, unsigned long address)
{
	address &= PAGE_MASK;
	if ((vma->vm_flags & VM_GROWSDOWN) && address == vma->vm_start) {
3107 3108 3109 3110 3111 3112 3113 3114 3115 3116
		struct vm_area_struct *prev = vma->vm_prev;

		/*
		 * Is there a mapping abutting this one below?
		 *
		 * That's only ok if it's the same stack mapping
		 * that has gotten split..
		 */
		if (prev && prev->vm_end == address)
			return prev->vm_flags & VM_GROWSDOWN ? 0 : -ENOMEM;
3117

3118
		expand_downwards(vma, address - PAGE_SIZE);
3119
	}
3120 3121 3122 3123 3124 3125 3126 3127 3128
	if ((vma->vm_flags & VM_GROWSUP) && address + PAGE_SIZE == vma->vm_end) {
		struct vm_area_struct *next = vma->vm_next;

		/* As VM_GROWSDOWN but s/below/above/ */
		if (next && next->vm_start == address + PAGE_SIZE)
			return next->vm_flags & VM_GROWSUP ? 0 : -ENOMEM;

		expand_upwards(vma, address + PAGE_SIZE);
	}
3129 3130 3131
	return 0;
}

L
Linus Torvalds 已提交
3132
/*
3133 3134 3135
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
 * but allow concurrent faults), and pte mapped but not yet locked.
 * We return with mmap_sem still held, but pte unmapped and unlocked.
L
Linus Torvalds 已提交
3136
 */
3137 3138
static int do_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
3139
		unsigned int flags)
L
Linus Torvalds 已提交
3140
{
3141 3142
	struct page *page;
	spinlock_t *ptl;
L
Linus Torvalds 已提交
3143 3144
	pte_t entry;

3145 3146 3147 3148
	pte_unmap(page_table);

	/* Check if we need to add a guard page to the stack */
	if (check_stack_guard_page(vma, address) < 0)
3149 3150
		return VM_FAULT_SIGBUS;

3151
	/* Use the zero-page for reads */
H
Hugh Dickins 已提交
3152 3153 3154
	if (!(flags & FAULT_FLAG_WRITE)) {
		entry = pte_mkspecial(pfn_pte(my_zero_pfn(address),
						vma->vm_page_prot));
3155
		page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
H
Hugh Dickins 已提交
3156 3157 3158 3159 3160
		if (!pte_none(*page_table))
			goto unlock;
		goto setpte;
	}

N
Nick Piggin 已提交
3161 3162 3163 3164 3165 3166
	/* Allocate our own private page. */
	if (unlikely(anon_vma_prepare(vma)))
		goto oom;
	page = alloc_zeroed_user_highpage_movable(vma, address);
	if (!page)
		goto oom;
N
Nick Piggin 已提交
3167
	__SetPageUptodate(page);
3168

K
KAMEZAWA Hiroyuki 已提交
3169
	if (mem_cgroup_newpage_charge(page, mm, GFP_KERNEL))
3170 3171
		goto oom_free_page;

N
Nick Piggin 已提交
3172
	entry = mk_pte(page, vma->vm_page_prot);
H
Hugh Dickins 已提交
3173 3174
	if (vma->vm_flags & VM_WRITE)
		entry = pte_mkwrite(pte_mkdirty(entry));
L
Linus Torvalds 已提交
3175

N
Nick Piggin 已提交
3176
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
3177
	if (!pte_none(*page_table))
N
Nick Piggin 已提交
3178
		goto release;
H
Hugh Dickins 已提交
3179

3180
	inc_mm_counter_fast(mm, MM_ANONPAGES);
N
Nick Piggin 已提交
3181
	page_add_new_anon_rmap(page, vma, address);
H
Hugh Dickins 已提交
3182
setpte:
3183
	set_pte_at(mm, address, page_table, entry);
L
Linus Torvalds 已提交
3184 3185

	/* No need to invalidate - it was non-present before */
3186
	update_mmu_cache(vma, address, page_table);
3187
unlock:
3188
	pte_unmap_unlock(page_table, ptl);
N
Nick Piggin 已提交
3189
	return 0;
3190
release:
3191
	mem_cgroup_uncharge_page(page);
3192 3193
	page_cache_release(page);
	goto unlock;
3194
oom_free_page:
3195
	page_cache_release(page);
3196
oom:
L
Linus Torvalds 已提交
3197 3198 3199 3200
	return VM_FAULT_OOM;
}

/*
3201
 * __do_fault() tries to create a new page mapping. It aggressively
L
Linus Torvalds 已提交
3202
 * tries to share with existing pages, but makes a separate copy if
3203 3204
 * the FAULT_FLAG_WRITE is set in the flags parameter in order to avoid
 * the next page fault.
L
Linus Torvalds 已提交
3205 3206 3207 3208
 *
 * As this is called only for pages that do not currently exist, we
 * do not need to flush old virtual caches or the TLB.
 *
3209
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
3210
 * but allow concurrent faults), and pte neither mapped nor locked.
3211
 * We return with mmap_sem still held, but pte unmapped and unlocked.
L
Linus Torvalds 已提交
3212
 */
3213
static int __do_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3214
		unsigned long address, pmd_t *pmd,
3215
		pgoff_t pgoff, unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
3216
{
3217
	pte_t *page_table;
3218
	spinlock_t *ptl;
N
Nick Piggin 已提交
3219
	struct page *page;
3220
	struct page *cow_page;
L
Linus Torvalds 已提交
3221 3222
	pte_t entry;
	int anon = 0;
3223
	struct page *dirty_page = NULL;
N
Nick Piggin 已提交
3224 3225
	struct vm_fault vmf;
	int ret;
3226
	int page_mkwrite = 0;
3227

3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247
	/*
	 * If we do COW later, allocate page befor taking lock_page()
	 * on the file cache page. This will reduce lock holding time.
	 */
	if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) {

		if (unlikely(anon_vma_prepare(vma)))
			return VM_FAULT_OOM;

		cow_page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address);
		if (!cow_page)
			return VM_FAULT_OOM;

		if (mem_cgroup_newpage_charge(cow_page, mm, GFP_KERNEL)) {
			page_cache_release(cow_page);
			return VM_FAULT_OOM;
		}
	} else
		cow_page = NULL;

N
Nick Piggin 已提交
3248 3249 3250 3251
	vmf.virtual_address = (void __user *)(address & PAGE_MASK);
	vmf.pgoff = pgoff;
	vmf.flags = flags;
	vmf.page = NULL;
L
Linus Torvalds 已提交
3252

N
Nick Piggin 已提交
3253
	ret = vma->vm_ops->fault(vma, &vmf);
3254 3255
	if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE |
			    VM_FAULT_RETRY)))
3256
		goto uncharge_out;
L
Linus Torvalds 已提交
3257

3258 3259 3260
	if (unlikely(PageHWPoison(vmf.page))) {
		if (ret & VM_FAULT_LOCKED)
			unlock_page(vmf.page);
3261 3262
		ret = VM_FAULT_HWPOISON;
		goto uncharge_out;
3263 3264
	}

3265
	/*
N
Nick Piggin 已提交
3266
	 * For consistency in subsequent calls, make the faulted page always
3267 3268
	 * locked.
	 */
N
Nick Piggin 已提交
3269
	if (unlikely(!(ret & VM_FAULT_LOCKED)))
N
Nick Piggin 已提交
3270
		lock_page(vmf.page);
3271
	else
N
Nick Piggin 已提交
3272
		VM_BUG_ON(!PageLocked(vmf.page));
3273

L
Linus Torvalds 已提交
3274 3275 3276
	/*
	 * Should we do an early C-O-W break?
	 */
N
Nick Piggin 已提交
3277
	page = vmf.page;
3278
	if (flags & FAULT_FLAG_WRITE) {
3279
		if (!(vma->vm_flags & VM_SHARED)) {
3280
			page = cow_page;
3281
			anon = 1;
N
Nick Piggin 已提交
3282
			copy_user_highpage(page, vmf.page, address, vma);
N
Nick Piggin 已提交
3283
			__SetPageUptodate(page);
3284
		} else {
3285 3286
			/*
			 * If the page will be shareable, see if the backing
3287
			 * address space wants to know that the page is about
3288 3289
			 * to become writable
			 */
3290
			if (vma->vm_ops->page_mkwrite) {
3291 3292
				int tmp;

3293
				unlock_page(page);
N
Nick Piggin 已提交
3294
				vmf.flags = FAULT_FLAG_WRITE|FAULT_FLAG_MKWRITE;
3295 3296 3297 3298
				tmp = vma->vm_ops->page_mkwrite(vma, &vmf);
				if (unlikely(tmp &
					  (VM_FAULT_ERROR | VM_FAULT_NOPAGE))) {
					ret = tmp;
N
Nick Piggin 已提交
3299
					goto unwritable_page;
N
Nick Piggin 已提交
3300
				}
N
Nick Piggin 已提交
3301 3302 3303 3304 3305 3306 3307 3308 3309
				if (unlikely(!(tmp & VM_FAULT_LOCKED))) {
					lock_page(page);
					if (!page->mapping) {
						ret = 0; /* retry the fault */
						unlock_page(page);
						goto unwritable_page;
					}
				} else
					VM_BUG_ON(!PageLocked(page));
3310
				page_mkwrite = 1;
3311 3312
			}
		}
3313

L
Linus Torvalds 已提交
3314 3315
	}

3316
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
3317 3318 3319 3320 3321 3322

	/*
	 * This silly early PAGE_DIRTY setting removes a race
	 * due to the bad i386 page protection. But it's valid
	 * for other architectures too.
	 *
3323
	 * Note that if FAULT_FLAG_WRITE is set, we either now have
L
Linus Torvalds 已提交
3324 3325 3326 3327 3328
	 * an exclusive copy of the page, or this is a shared mapping,
	 * so we can make it writable and dirty to avoid having to
	 * handle that later.
	 */
	/* Only go through if we didn't race with anybody else... */
3329
	if (likely(pte_same(*page_table, orig_pte))) {
3330 3331
		flush_icache_page(vma, page);
		entry = mk_pte(page, vma->vm_page_prot);
3332
		if (flags & FAULT_FLAG_WRITE)
L
Linus Torvalds 已提交
3333 3334
			entry = maybe_mkwrite(pte_mkdirty(entry), vma);
		if (anon) {
3335
			inc_mm_counter_fast(mm, MM_ANONPAGES);
3336
			page_add_new_anon_rmap(page, vma, address);
3337
		} else {
3338
			inc_mm_counter_fast(mm, MM_FILEPAGES);
3339
			page_add_file_rmap(page);
3340
			if (flags & FAULT_FLAG_WRITE) {
3341
				dirty_page = page;
3342 3343
				get_page(dirty_page);
			}
3344
		}
3345
		set_pte_at(mm, address, page_table, entry);
3346 3347

		/* no need to invalidate: a not-present page won't be cached */
3348
		update_mmu_cache(vma, address, page_table);
L
Linus Torvalds 已提交
3349
	} else {
3350 3351
		if (cow_page)
			mem_cgroup_uncharge_page(cow_page);
3352 3353 3354
		if (anon)
			page_cache_release(page);
		else
3355
			anon = 1; /* no anon but release faulted_page */
L
Linus Torvalds 已提交
3356 3357
	}

3358
	pte_unmap_unlock(page_table, ptl);
3359

N
Nick Piggin 已提交
3360 3361
	if (dirty_page) {
		struct address_space *mapping = page->mapping;
3362
		int dirtied = 0;
3363

N
Nick Piggin 已提交
3364
		if (set_page_dirty(dirty_page))
3365
			dirtied = 1;
N
Nick Piggin 已提交
3366
		unlock_page(dirty_page);
3367
		put_page(dirty_page);
3368
		if ((dirtied || page_mkwrite) && mapping) {
N
Nick Piggin 已提交
3369 3370 3371 3372 3373 3374 3375 3376
			/*
			 * Some device drivers do not set page.mapping but still
			 * dirty their pages
			 */
			balance_dirty_pages_ratelimited(mapping);
		}

		/* file_update_time outside page_lock */
3377
		if (vma->vm_file && !page_mkwrite)
N
Nick Piggin 已提交
3378 3379 3380 3381 3382
			file_update_time(vma->vm_file);
	} else {
		unlock_page(vmf.page);
		if (anon)
			page_cache_release(vmf.page);
3383
	}
3384

N
Nick Piggin 已提交
3385
	return ret;
N
Nick Piggin 已提交
3386 3387 3388 3389

unwritable_page:
	page_cache_release(page);
	return ret;
3390 3391 3392 3393 3394 3395 3396
uncharge_out:
	/* fs's fault handler get error */
	if (cow_page) {
		mem_cgroup_uncharge_page(cow_page);
		page_cache_release(cow_page);
	}
	return ret;
3397
}
3398

3399 3400
static int do_linear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
3401
		unsigned int flags, pte_t orig_pte)
3402 3403
{
	pgoff_t pgoff = (((address & PAGE_MASK)
3404
			- vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
3405

3406 3407
	pte_unmap(page_table);
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
3408 3409
}

L
Linus Torvalds 已提交
3410 3411 3412 3413
/*
 * Fault of a previously existing named mapping. Repopulate the pte
 * from the encoded file_pte if possible. This enables swappable
 * nonlinear vmas.
3414 3415 3416 3417
 *
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
 * but allow concurrent faults), and pte mapped but not yet locked.
 * We return with mmap_sem still held, but pte unmapped and unlocked.
L
Linus Torvalds 已提交
3418
 */
N
Nick Piggin 已提交
3419
static int do_nonlinear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3420
		unsigned long address, pte_t *page_table, pmd_t *pmd,
3421
		unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
3422
{
3423
	pgoff_t pgoff;
L
Linus Torvalds 已提交
3424

3425 3426
	flags |= FAULT_FLAG_NONLINEAR;

H
Hugh Dickins 已提交
3427
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
N
Nick Piggin 已提交
3428
		return 0;
L
Linus Torvalds 已提交
3429

3430
	if (unlikely(!(vma->vm_flags & VM_NONLINEAR))) {
3431 3432 3433
		/*
		 * Page table corrupted: show pte and kill process.
		 */
3434
		print_bad_pte(vma, address, orig_pte, NULL);
H
Hugh Dickins 已提交
3435
		return VM_FAULT_SIGBUS;
3436 3437 3438
	}

	pgoff = pte_to_pgoff(orig_pte);
3439
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
L
Linus Torvalds 已提交
3440 3441
}

3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453
int numa_migrate_prep(struct page *page, struct vm_area_struct *vma,
				unsigned long addr, int current_nid)
{
	get_page(page);

	count_vm_numa_event(NUMA_HINT_FAULTS);
	if (current_nid == numa_node_id())
		count_vm_numa_event(NUMA_HINT_FAULTS_LOCAL);

	return mpol_misplaced(page, vma, addr);
}

3454 3455 3456
int do_numa_page(struct mm_struct *mm, struct vm_area_struct *vma,
		   unsigned long addr, pte_t pte, pte_t *ptep, pmd_t *pmd)
{
3457
	struct page *page = NULL;
3458
	spinlock_t *ptl;
3459 3460
	int current_nid = -1;
	int target_nid;
3461
	bool migrated = false;
3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473

	/*
	* The "pte" at this point cannot be used safely without
	* validation through pte_unmap_same(). It's of NUMA type but
	* the pfn may be screwed if the read is non atomic.
	*
	* ptep_modify_prot_start is not called as this is clearing
	* the _PAGE_NUMA bit and it is not really expected that there
	* would be concurrent hardware modifications to the PTE.
	*/
	ptl = pte_lockptr(mm, pmd);
	spin_lock(ptl);
3474 3475 3476 3477 3478
	if (unlikely(!pte_same(*ptep, pte))) {
		pte_unmap_unlock(ptep, ptl);
		goto out;
	}

3479 3480 3481 3482 3483 3484 3485 3486 3487 3488
	pte = pte_mknonnuma(pte);
	set_pte_at(mm, addr, ptep, pte);
	update_mmu_cache(vma, addr, ptep);

	page = vm_normal_page(vma, addr, pte);
	if (!page) {
		pte_unmap_unlock(ptep, ptl);
		return 0;
	}

3489
	current_nid = page_to_nid(page);
3490
	target_nid = numa_migrate_prep(page, vma, addr, current_nid);
3491
	pte_unmap_unlock(ptep, ptl);
3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502
	if (target_nid == -1) {
		/*
		 * Account for the fault against the current node if it not
		 * being replaced regardless of where the page is located.
		 */
		current_nid = numa_node_id();
		put_page(page);
		goto out;
	}

	/* Migrate to the requested node */
3503 3504
	migrated = migrate_misplaced_page(page, target_nid);
	if (migrated)
3505 3506 3507
		current_nid = target_nid;

out:
3508
	if (current_nid != -1)
3509
		task_numa_fault(current_nid, 1, migrated);
3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523
	return 0;
}

/* NUMA hinting page fault entry point for regular pmds */
#ifdef CONFIG_NUMA_BALANCING
static int do_pmd_numa_page(struct mm_struct *mm, struct vm_area_struct *vma,
		     unsigned long addr, pmd_t *pmdp)
{
	pmd_t pmd;
	pte_t *pte, *orig_pte;
	unsigned long _addr = addr & PMD_MASK;
	unsigned long offset;
	spinlock_t *ptl;
	bool numa = false;
3524
	int local_nid = numa_node_id();
3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546

	spin_lock(&mm->page_table_lock);
	pmd = *pmdp;
	if (pmd_numa(pmd)) {
		set_pmd_at(mm, _addr, pmdp, pmd_mknonnuma(pmd));
		numa = true;
	}
	spin_unlock(&mm->page_table_lock);

	if (!numa)
		return 0;

	/* we're in a page fault so some vma must be in the range */
	BUG_ON(!vma);
	BUG_ON(vma->vm_start >= _addr + PMD_SIZE);
	offset = max(_addr, vma->vm_start) & ~PMD_MASK;
	VM_BUG_ON(offset >= PMD_SIZE);
	orig_pte = pte = pte_offset_map_lock(mm, pmdp, _addr, &ptl);
	pte += offset >> PAGE_SHIFT;
	for (addr = _addr + offset; addr < _addr + PMD_SIZE; pte++, addr += PAGE_SIZE) {
		pte_t pteval = *pte;
		struct page *page;
3547 3548
		int curr_nid = local_nid;
		int target_nid;
3549
		bool migrated;
3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566
		if (!pte_present(pteval))
			continue;
		if (!pte_numa(pteval))
			continue;
		if (addr >= vma->vm_end) {
			vma = find_vma(mm, addr);
			/* there's a pte present so there must be a vma */
			BUG_ON(!vma);
			BUG_ON(addr < vma->vm_start);
		}
		if (pte_numa(pteval)) {
			pteval = pte_mknonnuma(pteval);
			set_pte_at(mm, addr, pte, pteval);
		}
		page = vm_normal_page(vma, addr, pteval);
		if (unlikely(!page))
			continue;
3567 3568 3569 3570
		/* only check non-shared pages */
		if (unlikely(page_mapcount(page) != 1))
			continue;

3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582
		/*
		 * Note that the NUMA fault is later accounted to either
		 * the node that is currently running or where the page is
		 * migrated to.
		 */
		curr_nid = local_nid;
		target_nid = numa_migrate_prep(page, vma, addr,
					       page_to_nid(page));
		if (target_nid == -1) {
			put_page(page);
			continue;
		}
3583

3584 3585
		/* Migrate to the requested node */
		pte_unmap_unlock(pte, ptl);
3586 3587
		migrated = migrate_misplaced_page(page, target_nid);
		if (migrated)
3588
			curr_nid = target_nid;
3589
		task_numa_fault(curr_nid, 1, migrated);
3590

3591
		pte = pte_offset_map_lock(mm, pmdp, addr, &ptl);
3592 3593 3594 3595 3596 3597 3598 3599 3600 3601
	}
	pte_unmap_unlock(orig_pte, ptl);

	return 0;
}
#else
static int do_pmd_numa_page(struct mm_struct *mm, struct vm_area_struct *vma,
		     unsigned long addr, pmd_t *pmdp)
{
	BUG();
A
Andrew Morton 已提交
3602
	return 0;
3603 3604 3605
}
#endif /* CONFIG_NUMA_BALANCING */

L
Linus Torvalds 已提交
3606 3607 3608 3609 3610 3611 3612 3613 3614
/*
 * These routines also need to handle stuff like marking pages dirty
 * and/or accessed for architectures that don't do it in hardware (most
 * RISC architectures).  The early dirtying is also good on the i386.
 *
 * There is also a hook called "update_mmu_cache()" that architectures
 * with external mmu caches can use to update those (ie the Sparc or
 * PowerPC hashed page tables that act as extended TLBs).
 *
H
Hugh Dickins 已提交
3615 3616 3617
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
 * but allow concurrent faults), and pte mapped but not yet locked.
 * We return with mmap_sem still held, but pte unmapped and unlocked.
L
Linus Torvalds 已提交
3618
 */
3619 3620 3621
int handle_pte_fault(struct mm_struct *mm,
		     struct vm_area_struct *vma, unsigned long address,
		     pte_t *pte, pmd_t *pmd, unsigned int flags)
L
Linus Torvalds 已提交
3622 3623
{
	pte_t entry;
3624
	spinlock_t *ptl;
L
Linus Torvalds 已提交
3625

3626
	entry = *pte;
L
Linus Torvalds 已提交
3627
	if (!pte_present(entry)) {
3628
		if (pte_none(entry)) {
J
Jes Sorensen 已提交
3629
			if (vma->vm_ops) {
N
Nick Piggin 已提交
3630
				if (likely(vma->vm_ops->fault))
3631
					return do_linear_fault(mm, vma, address,
3632
						pte, pmd, flags, entry);
J
Jes Sorensen 已提交
3633 3634
			}
			return do_anonymous_page(mm, vma, address,
3635
						 pte, pmd, flags);
3636
		}
L
Linus Torvalds 已提交
3637
		if (pte_file(entry))
N
Nick Piggin 已提交
3638
			return do_nonlinear_fault(mm, vma, address,
3639
					pte, pmd, flags, entry);
3640
		return do_swap_page(mm, vma, address,
3641
					pte, pmd, flags, entry);
L
Linus Torvalds 已提交
3642 3643
	}

3644 3645 3646
	if (pte_numa(entry))
		return do_numa_page(mm, vma, address, entry, pte, pmd);

H
Hugh Dickins 已提交
3647
	ptl = pte_lockptr(mm, pmd);
3648 3649 3650
	spin_lock(ptl);
	if (unlikely(!pte_same(*pte, entry)))
		goto unlock;
3651
	if (flags & FAULT_FLAG_WRITE) {
L
Linus Torvalds 已提交
3652
		if (!pte_write(entry))
3653 3654
			return do_wp_page(mm, vma, address,
					pte, pmd, ptl, entry);
L
Linus Torvalds 已提交
3655 3656 3657
		entry = pte_mkdirty(entry);
	}
	entry = pte_mkyoung(entry);
3658
	if (ptep_set_access_flags(vma, address, pte, entry, flags & FAULT_FLAG_WRITE)) {
3659
		update_mmu_cache(vma, address, pte);
3660 3661 3662 3663 3664 3665 3666
	} else {
		/*
		 * This is needed only for protection faults but the arch code
		 * is not yet telling us if this is a protection fault or not.
		 * This still avoids useless tlb flushes for .text page faults
		 * with threads.
		 */
3667
		if (flags & FAULT_FLAG_WRITE)
3668
			flush_tlb_fix_spurious_fault(vma, address);
3669
	}
3670 3671
unlock:
	pte_unmap_unlock(pte, ptl);
N
Nick Piggin 已提交
3672
	return 0;
L
Linus Torvalds 已提交
3673 3674 3675 3676 3677
}

/*
 * By the time we get here, we already hold the mm semaphore
 */
N
Nick Piggin 已提交
3678
int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3679
		unsigned long address, unsigned int flags)
L
Linus Torvalds 已提交
3680 3681 3682 3683 3684 3685 3686 3687
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

	__set_current_state(TASK_RUNNING);

3688
	count_vm_event(PGFAULT);
3689
	mem_cgroup_count_vm_event(mm, PGFAULT);
L
Linus Torvalds 已提交
3690

3691 3692 3693
	/* do counter updates before entering really critical section. */
	check_sync_rss_stat(current);

3694
	if (unlikely(is_vm_hugetlb_page(vma)))
3695
		return hugetlb_fault(mm, vma, address, flags);
L
Linus Torvalds 已提交
3696

3697
retry:
L
Linus Torvalds 已提交
3698 3699 3700
	pgd = pgd_offset(mm, address);
	pud = pud_alloc(mm, pgd, address);
	if (!pud)
H
Hugh Dickins 已提交
3701
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
3702 3703
	pmd = pmd_alloc(mm, pud, address);
	if (!pmd)
H
Hugh Dickins 已提交
3704
		return VM_FAULT_OOM;
3705 3706 3707 3708 3709 3710
	if (pmd_none(*pmd) && transparent_hugepage_enabled(vma)) {
		if (!vma->vm_ops)
			return do_huge_pmd_anonymous_page(mm, vma, address,
							  pmd, flags);
	} else {
		pmd_t orig_pmd = *pmd;
3711 3712
		int ret;

3713 3714
		barrier();
		if (pmd_trans_huge(orig_pmd)) {
3715 3716
			unsigned int dirty = flags & FAULT_FLAG_WRITE;

3717 3718 3719 3720 3721 3722 3723 3724
			/*
			 * If the pmd is splitting, return and retry the
			 * the fault.  Alternative: wait until the split
			 * is done, and goto retry.
			 */
			if (pmd_trans_splitting(orig_pmd))
				return 0;

3725
			if (pmd_numa(orig_pmd))
3726
				return do_huge_pmd_numa_page(mm, vma, address,
3727 3728
							     orig_pmd, pmd);

3729
			if (dirty && !pmd_write(orig_pmd)) {
3730 3731 3732 3733 3734 3735 3736 3737 3738 3739
				ret = do_huge_pmd_wp_page(mm, vma, address, pmd,
							  orig_pmd);
				/*
				 * If COW results in an oom, the huge pmd will
				 * have been split, so retry the fault on the
				 * pte for a smaller charge.
				 */
				if (unlikely(ret & VM_FAULT_OOM))
					goto retry;
				return ret;
3740 3741 3742
			} else {
				huge_pmd_set_accessed(mm, vma, address, pmd,
						      orig_pmd, dirty);
3743
			}
3744

3745 3746 3747 3748
			return 0;
		}
	}

3749 3750 3751
	if (pmd_numa(*pmd))
		return do_pmd_numa_page(mm, vma, address, pmd);

3752 3753 3754 3755 3756
	/*
	 * Use __pte_alloc instead of pte_alloc_map, because we can't
	 * run pte_offset_map on the pmd, if an huge pmd could
	 * materialize from under us from a different thread.
	 */
3757 3758
	if (unlikely(pmd_none(*pmd)) &&
	    unlikely(__pte_alloc(mm, vma, pmd, address)))
H
Hugh Dickins 已提交
3759
		return VM_FAULT_OOM;
3760 3761 3762 3763 3764 3765 3766 3767 3768 3769
	/* if an huge pmd materialized from under us just retry later */
	if (unlikely(pmd_trans_huge(*pmd)))
		return 0;
	/*
	 * A regular pmd is established and it can't morph into a huge pmd
	 * from under us anymore at this point because we hold the mmap_sem
	 * read mode and khugepaged takes it in write mode. So now it's
	 * safe to run pte_offset_map().
	 */
	pte = pte_offset_map(pmd, address);
L
Linus Torvalds 已提交
3770

3771
	return handle_pte_fault(mm, vma, address, pte, pmd, flags);
L
Linus Torvalds 已提交
3772 3773 3774 3775 3776
}

#ifndef __PAGETABLE_PUD_FOLDED
/*
 * Allocate page upper directory.
H
Hugh Dickins 已提交
3777
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
3778
 */
3779
int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
L
Linus Torvalds 已提交
3780
{
H
Hugh Dickins 已提交
3781 3782
	pud_t *new = pud_alloc_one(mm, address);
	if (!new)
3783
		return -ENOMEM;
L
Linus Torvalds 已提交
3784

3785 3786
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
3787
	spin_lock(&mm->page_table_lock);
3788
	if (pgd_present(*pgd))		/* Another has populated it */
3789
		pud_free(mm, new);
3790 3791
	else
		pgd_populate(mm, pgd, new);
H
Hugh Dickins 已提交
3792
	spin_unlock(&mm->page_table_lock);
3793
	return 0;
L
Linus Torvalds 已提交
3794 3795 3796 3797 3798 3799
}
#endif /* __PAGETABLE_PUD_FOLDED */

#ifndef __PAGETABLE_PMD_FOLDED
/*
 * Allocate page middle directory.
H
Hugh Dickins 已提交
3800
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
3801
 */
3802
int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
L
Linus Torvalds 已提交
3803
{
H
Hugh Dickins 已提交
3804 3805
	pmd_t *new = pmd_alloc_one(mm, address);
	if (!new)
3806
		return -ENOMEM;
L
Linus Torvalds 已提交
3807

3808 3809
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
3810
	spin_lock(&mm->page_table_lock);
L
Linus Torvalds 已提交
3811
#ifndef __ARCH_HAS_4LEVEL_HACK
3812
	if (pud_present(*pud))		/* Another has populated it */
3813
		pmd_free(mm, new);
3814 3815
	else
		pud_populate(mm, pud, new);
L
Linus Torvalds 已提交
3816
#else
3817
	if (pgd_present(*pud))		/* Another has populated it */
3818
		pmd_free(mm, new);
3819 3820
	else
		pgd_populate(mm, pud, new);
L
Linus Torvalds 已提交
3821
#endif /* __ARCH_HAS_4LEVEL_HACK */
H
Hugh Dickins 已提交
3822
	spin_unlock(&mm->page_table_lock);
3823
	return 0;
3824
}
L
Linus Torvalds 已提交
3825 3826 3827 3828 3829
#endif /* __PAGETABLE_PMD_FOLDED */

#if !defined(__HAVE_ARCH_GATE_AREA)

#if defined(AT_SYSINFO_EHDR)
3830
static struct vm_area_struct gate_vma;
L
Linus Torvalds 已提交
3831 3832 3833 3834 3835 3836

static int __init gate_vma_init(void)
{
	gate_vma.vm_mm = NULL;
	gate_vma.vm_start = FIXADDR_USER_START;
	gate_vma.vm_end = FIXADDR_USER_END;
R
Roland McGrath 已提交
3837 3838
	gate_vma.vm_flags = VM_READ | VM_MAYREAD | VM_EXEC | VM_MAYEXEC;
	gate_vma.vm_page_prot = __P101;
3839

L
Linus Torvalds 已提交
3840 3841 3842 3843 3844
	return 0;
}
__initcall(gate_vma_init);
#endif

3845
struct vm_area_struct *get_gate_vma(struct mm_struct *mm)
L
Linus Torvalds 已提交
3846 3847 3848 3849 3850 3851 3852 3853
{
#ifdef AT_SYSINFO_EHDR
	return &gate_vma;
#else
	return NULL;
#endif
}

3854
int in_gate_area_no_mm(unsigned long addr)
L
Linus Torvalds 已提交
3855 3856 3857 3858 3859 3860 3861 3862 3863
{
#ifdef AT_SYSINFO_EHDR
	if ((addr >= FIXADDR_USER_START) && (addr < FIXADDR_USER_END))
		return 1;
#endif
	return 0;
}

#endif	/* __HAVE_ARCH_GATE_AREA */
3864

3865
static int __follow_pte(struct mm_struct *mm, unsigned long address,
J
Johannes Weiner 已提交
3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881
		pte_t **ptepp, spinlock_t **ptlp)
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *ptep;

	pgd = pgd_offset(mm, address);
	if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
		goto out;

	pud = pud_offset(pgd, address);
	if (pud_none(*pud) || unlikely(pud_bad(*pud)))
		goto out;

	pmd = pmd_offset(pud, address);
3882
	VM_BUG_ON(pmd_trans_huge(*pmd));
J
Johannes Weiner 已提交
3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902
	if (pmd_none(*pmd) || unlikely(pmd_bad(*pmd)))
		goto out;

	/* We cannot handle huge page PFN maps. Luckily they don't exist. */
	if (pmd_huge(*pmd))
		goto out;

	ptep = pte_offset_map_lock(mm, pmd, address, ptlp);
	if (!ptep)
		goto out;
	if (!pte_present(*ptep))
		goto unlock;
	*ptepp = ptep;
	return 0;
unlock:
	pte_unmap_unlock(ptep, *ptlp);
out:
	return -EINVAL;
}

3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913
static inline int follow_pte(struct mm_struct *mm, unsigned long address,
			     pte_t **ptepp, spinlock_t **ptlp)
{
	int res;

	/* (void) is needed to make gcc happy */
	(void) __cond_lock(*ptlp,
			   !(res = __follow_pte(mm, address, ptepp, ptlp)));
	return res;
}

J
Johannes Weiner 已提交
3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942
/**
 * follow_pfn - look up PFN at a user virtual address
 * @vma: memory mapping
 * @address: user virtual address
 * @pfn: location to store found PFN
 *
 * Only IO mappings and raw PFN mappings are allowed.
 *
 * Returns zero and the pfn at @pfn on success, -ve otherwise.
 */
int follow_pfn(struct vm_area_struct *vma, unsigned long address,
	unsigned long *pfn)
{
	int ret = -EINVAL;
	spinlock_t *ptl;
	pte_t *ptep;

	if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
		return ret;

	ret = follow_pte(vma->vm_mm, address, &ptep, &ptl);
	if (ret)
		return ret;
	*pfn = pte_pfn(*ptep);
	pte_unmap_unlock(ptep, ptl);
	return 0;
}
EXPORT_SYMBOL(follow_pfn);

3943
#ifdef CONFIG_HAVE_IOREMAP_PROT
3944 3945 3946
int follow_phys(struct vm_area_struct *vma,
		unsigned long address, unsigned int flags,
		unsigned long *prot, resource_size_t *phys)
3947
{
3948
	int ret = -EINVAL;
3949 3950 3951
	pte_t *ptep, pte;
	spinlock_t *ptl;

3952 3953
	if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
		goto out;
3954

3955
	if (follow_pte(vma->vm_mm, address, &ptep, &ptl))
3956
		goto out;
3957
	pte = *ptep;
3958

3959 3960 3961 3962
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;

	*prot = pgprot_val(pte_pgprot(pte));
3963
	*phys = (resource_size_t)pte_pfn(pte) << PAGE_SHIFT;
3964

3965
	ret = 0;
3966 3967 3968
unlock:
	pte_unmap_unlock(ptep, ptl);
out:
3969
	return ret;
3970 3971 3972 3973 3974 3975 3976
}

int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
			void *buf, int len, int write)
{
	resource_size_t phys_addr;
	unsigned long prot = 0;
K
KOSAKI Motohiro 已提交
3977
	void __iomem *maddr;
3978 3979
	int offset = addr & (PAGE_SIZE-1);

3980
	if (follow_phys(vma, addr, write, &prot, &phys_addr))
3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993
		return -EINVAL;

	maddr = ioremap_prot(phys_addr, PAGE_SIZE, prot);
	if (write)
		memcpy_toio(maddr + offset, buf, len);
	else
		memcpy_fromio(buf, maddr + offset, len);
	iounmap(maddr);

	return len;
}
#endif

3994
/*
3995 3996
 * Access another process' address space as given in mm.  If non-NULL, use the
 * given task for page fault accounting.
3997
 */
3998 3999
static int __access_remote_vm(struct task_struct *tsk, struct mm_struct *mm,
		unsigned long addr, void *buf, int len, int write)
4000 4001 4002 4003 4004
{
	struct vm_area_struct *vma;
	void *old_buf = buf;

	down_read(&mm->mmap_sem);
S
Simon Arlott 已提交
4005
	/* ignore errors, just check how much was successfully transferred */
4006 4007 4008
	while (len) {
		int bytes, ret, offset;
		void *maddr;
4009
		struct page *page = NULL;
4010 4011 4012

		ret = get_user_pages(tsk, mm, addr, 1,
				write, 1, &page, &vma);
4013 4014 4015 4016 4017 4018 4019
		if (ret <= 0) {
			/*
			 * Check if this is a VM_IO | VM_PFNMAP VMA, which
			 * we can access using slightly different code.
			 */
#ifdef CONFIG_HAVE_IOREMAP_PROT
			vma = find_vma(mm, addr);
4020
			if (!vma || vma->vm_start > addr)
4021 4022 4023 4024 4025 4026 4027 4028
				break;
			if (vma->vm_ops && vma->vm_ops->access)
				ret = vma->vm_ops->access(vma, addr, buf,
							  len, write);
			if (ret <= 0)
#endif
				break;
			bytes = ret;
4029
		} else {
4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045
			bytes = len;
			offset = addr & (PAGE_SIZE-1);
			if (bytes > PAGE_SIZE-offset)
				bytes = PAGE_SIZE-offset;

			maddr = kmap(page);
			if (write) {
				copy_to_user_page(vma, page, addr,
						  maddr + offset, buf, bytes);
				set_page_dirty_lock(page);
			} else {
				copy_from_user_page(vma, page, addr,
						    buf, maddr + offset, bytes);
			}
			kunmap(page);
			page_cache_release(page);
4046 4047 4048 4049 4050 4051 4052 4053 4054
		}
		len -= bytes;
		buf += bytes;
		addr += bytes;
	}
	up_read(&mm->mmap_sem);

	return buf - old_buf;
}
4055

S
Stephen Wilson 已提交
4056
/**
4057
 * access_remote_vm - access another process' address space
S
Stephen Wilson 已提交
4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071
 * @mm:		the mm_struct of the target address space
 * @addr:	start address to access
 * @buf:	source or destination buffer
 * @len:	number of bytes to transfer
 * @write:	whether the access is a write
 *
 * The caller must hold a reference on @mm.
 */
int access_remote_vm(struct mm_struct *mm, unsigned long addr,
		void *buf, int len, int write)
{
	return __access_remote_vm(NULL, mm, addr, buf, len, write);
}

4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092
/*
 * Access another process' address space.
 * Source/target buffer must be kernel space,
 * Do not walk the page table directly, use get_user_pages
 */
int access_process_vm(struct task_struct *tsk, unsigned long addr,
		void *buf, int len, int write)
{
	struct mm_struct *mm;
	int ret;

	mm = get_task_mm(tsk);
	if (!mm)
		return 0;

	ret = __access_remote_vm(tsk, mm, addr, buf, len, write);
	mmput(mm);

	return ret;
}

4093 4094 4095 4096 4097 4098 4099 4100
/*
 * Print the name of a VMA.
 */
void print_vma_addr(char *prefix, unsigned long ip)
{
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma;

4101 4102 4103 4104 4105 4106 4107
	/*
	 * Do not print if we are in atomic
	 * contexts (in exception stacks, etc.):
	 */
	if (preempt_count())
		return;

4108 4109 4110 4111 4112 4113
	down_read(&mm->mmap_sem);
	vma = find_vma(mm, ip);
	if (vma && vma->vm_file) {
		struct file *f = vma->vm_file;
		char *buf = (char *)__get_free_page(GFP_KERNEL);
		if (buf) {
A
Andy Shevchenko 已提交
4114
			char *p;
4115

4116
			p = d_path(&f->f_path, buf, PAGE_SIZE);
4117 4118
			if (IS_ERR(p))
				p = "?";
A
Andy Shevchenko 已提交
4119
			printk("%s%s[%lx+%lx]", prefix, kbasename(p),
4120 4121 4122 4123 4124
					vma->vm_start,
					vma->vm_end - vma->vm_start);
			free_page((unsigned long)buf);
		}
	}
4125
	up_read(&mm->mmap_sem);
4126
}
4127 4128 4129 4130

#ifdef CONFIG_PROVE_LOCKING
void might_fault(void)
{
4131 4132 4133 4134 4135 4136 4137 4138 4139
	/*
	 * Some code (nfs/sunrpc) uses socket ops on kernel memory while
	 * holding the mmap_sem, this is safe because kernel memory doesn't
	 * get paged out, therefore we'll never actually fault, and the
	 * below annotations will generate false positives.
	 */
	if (segment_eq(get_fs(), KERNEL_DS))
		return;

4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150
	might_sleep();
	/*
	 * it would be nicer only to annotate paths which are not under
	 * pagefault_disable, however that requires a larger audit and
	 * providing helpers like get_user_atomic.
	 */
	if (!in_atomic() && current->mm)
		might_lock_read(&current->mm->mmap_sem);
}
EXPORT_SYMBOL(might_fault);
#endif
A
Andrea Arcangeli 已提交
4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221

#if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
static void clear_gigantic_page(struct page *page,
				unsigned long addr,
				unsigned int pages_per_huge_page)
{
	int i;
	struct page *p = page;

	might_sleep();
	for (i = 0; i < pages_per_huge_page;
	     i++, p = mem_map_next(p, page, i)) {
		cond_resched();
		clear_user_highpage(p, addr + i * PAGE_SIZE);
	}
}
void clear_huge_page(struct page *page,
		     unsigned long addr, unsigned int pages_per_huge_page)
{
	int i;

	if (unlikely(pages_per_huge_page > MAX_ORDER_NR_PAGES)) {
		clear_gigantic_page(page, addr, pages_per_huge_page);
		return;
	}

	might_sleep();
	for (i = 0; i < pages_per_huge_page; i++) {
		cond_resched();
		clear_user_highpage(page + i, addr + i * PAGE_SIZE);
	}
}

static void copy_user_gigantic_page(struct page *dst, struct page *src,
				    unsigned long addr,
				    struct vm_area_struct *vma,
				    unsigned int pages_per_huge_page)
{
	int i;
	struct page *dst_base = dst;
	struct page *src_base = src;

	for (i = 0; i < pages_per_huge_page; ) {
		cond_resched();
		copy_user_highpage(dst, src, addr + i*PAGE_SIZE, vma);

		i++;
		dst = mem_map_next(dst, dst_base, i);
		src = mem_map_next(src, src_base, i);
	}
}

void copy_user_huge_page(struct page *dst, struct page *src,
			 unsigned long addr, struct vm_area_struct *vma,
			 unsigned int pages_per_huge_page)
{
	int i;

	if (unlikely(pages_per_huge_page > MAX_ORDER_NR_PAGES)) {
		copy_user_gigantic_page(dst, src, addr, vma,
					pages_per_huge_page);
		return;
	}

	might_sleep();
	for (i = 0; i < pages_per_huge_page; i++) {
		cond_resched();
		copy_user_highpage(dst + i, src + i, addr + i*PAGE_SIZE, vma);
	}
}
#endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */