memory.c 107.4 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>
L
Linus Torvalds 已提交
60

A
Alexey Dobriyan 已提交
61
#include <asm/io.h>
L
Linus Torvalds 已提交
62 63 64 65 66 67
#include <asm/pgalloc.h>
#include <asm/uaccess.h>
#include <asm/tlb.h>
#include <asm/tlbflush.h>
#include <asm/pgtable.h>

68 69
#include "internal.h"

A
Andy Whitcroft 已提交
70
#ifndef CONFIG_NEED_MULTIPLE_NODES
L
Linus Torvalds 已提交
71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91
/* 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);

92 93 94 95 96 97 98 99 100 101 102 103
/*
 * 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
104 105 106 107

static int __init disable_randmaps(char *s)
{
	randomize_va_space = 0;
108
	return 1;
109 110 111
}
__setup("norandmaps", disable_randmaps);

H
Hugh Dickins 已提交
112
unsigned long zero_pfn __read_mostly;
H
Hugh Dickins 已提交
113
unsigned long highest_memmap_pfn __read_mostly;
H
Hugh Dickins 已提交
114 115 116 117 118 119 120 121 122 123

/*
 * 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);
124

K
KAMEZAWA Hiroyuki 已提交
125

126 127
#if defined(SPLIT_RSS_COUNTING)

128
static void __sync_task_rss_stat(struct mm_struct *mm)
129 130 131 132
{
	int i;

	for (i = 0; i < NR_MM_COUNTERS; i++) {
133 134 135
		if (current->rss_stat.count[i]) {
			add_mm_counter(mm, i, current->rss_stat.count[i]);
			current->rss_stat.count[i] = 0;
136 137
		}
	}
138
	current->rss_stat.events = 0;
139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159
}

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))
160
		__sync_task_rss_stat(task->mm);
161 162
}

163
void sync_mm_rss(struct mm_struct *mm)
164
{
165
	__sync_task_rss_stat(mm);
166
}
167
#else /* SPLIT_RSS_COUNTING */
168 169 170 171 172 173 174 175

#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)
{
}

176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235
#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;
	}

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

	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;
	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;

#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);
236 237
#endif

238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277
	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;

	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
278
	VM_BUG_ON(!tlb->need_flush);
279 280 281 282 283 284 285 286 287 288 289

	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;
290
		batch = tlb->active;
291 292 293 294 295 296 297 298
	}
	VM_BUG_ON(batch->nr > batch->max);

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

#endif /* HAVE_GENERIC_MMU_GATHER */

299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373
#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);
}

374
#endif /* CONFIG_HAVE_RCU_TABLE_FREE */
375

L
Linus Torvalds 已提交
376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403
/*
 * 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.
 */
404 405
static void free_pte_range(struct mmu_gather *tlb, pmd_t *pmd,
			   unsigned long addr)
L
Linus Torvalds 已提交
406
{
407
	pgtable_t token = pmd_pgtable(*pmd);
408
	pmd_clear(pmd);
409
	pte_free_tlb(tlb, token, addr);
410
	tlb->mm->nr_ptes--;
L
Linus Torvalds 已提交
411 412
}

413 414 415
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
416 417 418
{
	pmd_t *pmd;
	unsigned long next;
419
	unsigned long start;
L
Linus Torvalds 已提交
420

421
	start = addr;
L
Linus Torvalds 已提交
422 423 424 425 426
	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
427
		free_pte_range(tlb, pmd, addr);
L
Linus Torvalds 已提交
428 429
	} while (pmd++, addr = next, addr != end);

430 431 432 433 434 435 436
	start &= PUD_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PUD_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
437
	}
438 439 440 441 442
	if (end - 1 > ceiling - 1)
		return;

	pmd = pmd_offset(pud, start);
	pud_clear(pud);
443
	pmd_free_tlb(tlb, pmd, start);
L
Linus Torvalds 已提交
444 445
}

446 447 448
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
449 450 451
{
	pud_t *pud;
	unsigned long next;
452
	unsigned long start;
L
Linus Torvalds 已提交
453

454
	start = addr;
L
Linus Torvalds 已提交
455 456 457 458 459
	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
460
		free_pmd_range(tlb, pud, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
461 462
	} while (pud++, addr = next, addr != end);

463 464 465 466 467 468 469
	start &= PGDIR_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PGDIR_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
470
	}
471 472 473 474 475
	if (end - 1 > ceiling - 1)
		return;

	pud = pud_offset(pgd, start);
	pgd_clear(pgd);
476
	pud_free_tlb(tlb, pud, start);
L
Linus Torvalds 已提交
477 478 479
}

/*
480 481
 * This function frees user-level page tables of a process.
 *
L
Linus Torvalds 已提交
482 483
 * Must be called with pagetable lock held.
 */
484
void free_pgd_range(struct mmu_gather *tlb,
485 486
			unsigned long addr, unsigned long end,
			unsigned long floor, unsigned long ceiling)
L
Linus Torvalds 已提交
487 488 489
{
	pgd_t *pgd;
	unsigned long next;
490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515

	/*
	 * 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
516

517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532
	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;

533
	pgd = pgd_offset(tlb->mm, addr);
L
Linus Torvalds 已提交
534 535 536 537
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
538
		free_pud_range(tlb, pgd, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
539
	} while (pgd++, addr = next, addr != end);
540 541
}

542
void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *vma,
543
		unsigned long floor, unsigned long ceiling)
544 545 546 547 548
{
	while (vma) {
		struct vm_area_struct *next = vma->vm_next;
		unsigned long addr = vma->vm_start;

549
		/*
N
npiggin@suse.de 已提交
550 551
		 * Hide vma from rmap and truncate_pagecache before freeing
		 * pgtables
552
		 */
553
		unlink_anon_vmas(vma);
554 555
		unlink_file_vma(vma);

556
		if (is_vm_hugetlb_page(vma)) {
557
			hugetlb_free_pgd_range(tlb, addr, vma->vm_end,
558
				floor, next? next->vm_start: ceiling);
559 560 561 562 563
		} else {
			/*
			 * Optimization: gather nearby vmas into one call down
			 */
			while (next && next->vm_start <= vma->vm_end + PMD_SIZE
564
			       && !is_vm_hugetlb_page(next)) {
565 566
				vma = next;
				next = vma->vm_next;
567
				unlink_anon_vmas(vma);
568
				unlink_file_vma(vma);
569 570 571 572
			}
			free_pgd_range(tlb, addr, vma->vm_end,
				floor, next? next->vm_start: ceiling);
		}
573 574
		vma = next;
	}
L
Linus Torvalds 已提交
575 576
}

577 578
int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma,
		pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
579
{
580
	pgtable_t new = pte_alloc_one(mm, address);
581
	int wait_split_huge_page;
582 583 584
	if (!new)
		return -ENOMEM;

585 586 587 588 589 590 591 592 593 594 595 596 597 598 599
	/*
	 * 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
600
	spin_lock(&mm->page_table_lock);
601 602
	wait_split_huge_page = 0;
	if (likely(pmd_none(*pmd))) {	/* Has another populated it ? */
L
Linus Torvalds 已提交
603 604
		mm->nr_ptes++;
		pmd_populate(mm, pmd, new);
605
		new = NULL;
606 607
	} else if (unlikely(pmd_trans_splitting(*pmd)))
		wait_split_huge_page = 1;
H
Hugh Dickins 已提交
608
	spin_unlock(&mm->page_table_lock);
609 610
	if (new)
		pte_free(mm, new);
611 612
	if (wait_split_huge_page)
		wait_split_huge_page(vma->anon_vma, pmd);
613
	return 0;
L
Linus Torvalds 已提交
614 615
}

616
int __pte_alloc_kernel(pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
617
{
618 619 620 621
	pte_t *new = pte_alloc_one_kernel(&init_mm, address);
	if (!new)
		return -ENOMEM;

622 623
	smp_wmb(); /* See comment in __pte_alloc */

624
	spin_lock(&init_mm.page_table_lock);
625
	if (likely(pmd_none(*pmd))) {	/* Has another populated it ? */
626
		pmd_populate_kernel(&init_mm, pmd, new);
627
		new = NULL;
628 629
	} else
		VM_BUG_ON(pmd_trans_splitting(*pmd));
630
	spin_unlock(&init_mm.page_table_lock);
631 632
	if (new)
		pte_free_kernel(&init_mm, new);
633
	return 0;
L
Linus Torvalds 已提交
634 635
}

K
KAMEZAWA Hiroyuki 已提交
636 637 638 639 640 641
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)
642
{
K
KAMEZAWA Hiroyuki 已提交
643 644
	int i;

645
	if (current->mm == mm)
646
		sync_mm_rss(mm);
K
KAMEZAWA Hiroyuki 已提交
647 648 649
	for (i = 0; i < NR_MM_COUNTERS; i++)
		if (rss[i])
			add_mm_counter(mm, i, rss[i]);
650 651
}

N
Nick Piggin 已提交
652
/*
653 654 655
 * 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
656 657 658
 *
 * The calling function must still handle the error.
 */
659 660
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
661
{
662 663 664 665 666
	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;
667 668 669 670 671 672 673 674 675 676 677 678 679 680
	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) {
681 682
			printk(KERN_ALERT
				"BUG: Bad page map: %lu messages suppressed\n",
683 684 685 686 687 688 689
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;
690 691 692 693

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

694 695
	printk(KERN_ALERT
		"BUG: Bad page map in process %s  pte:%08llx pmd:%08llx\n",
696 697
		current->comm,
		(long long)pte_val(pte), (long long)pmd_val(*pmd));
698 699
	if (page)
		dump_page(page);
700
	printk(KERN_ALERT
701 702 703 704 705 706
		"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)
707
		print_symbol(KERN_ALERT "vma->vm_ops->fault: %s\n",
708 709
				(unsigned long)vma->vm_ops->fault);
	if (vma->vm_file && vma->vm_file->f_op)
710
		print_symbol(KERN_ALERT "vma->vm_file->f_op->mmap: %s\n",
711
				(unsigned long)vma->vm_file->f_op->mmap);
N
Nick Piggin 已提交
712
	dump_stack();
713
	add_taint(TAINT_BAD_PAGE);
N
Nick Piggin 已提交
714 715
}

716
static inline int is_cow_mapping(vm_flags_t flags)
717 718 719 720
{
	return (flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
}

H
Hugh Dickins 已提交
721 722 723 724 725 726 727 728 729 730 731 732 733 734
#ifndef is_zero_pfn
static inline int is_zero_pfn(unsigned long pfn)
{
	return pfn == zero_pfn;
}
#endif

#ifndef my_zero_pfn
static inline unsigned long my_zero_pfn(unsigned long addr)
{
	return zero_pfn;
}
#endif

H
Hugh Dickins 已提交
735
/*
N
Nick Piggin 已提交
736
 * vm_normal_page -- This function gets the "struct page" associated with a pte.
737
 *
N
Nick Piggin 已提交
738 739 740
 * "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
741
 *
N
Nick Piggin 已提交
742 743 744 745 746 747 748 749
 * 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.
750
 *
J
Jared Hulbert 已提交
751 752
 * 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
753 754
 * set, and the vm_pgoff will point to the first PFN mapped: thus every special
 * mapping will always honor the rule
755 756 757
 *
 *	pfn_of_page == vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT)
 *
N
Nick Piggin 已提交
758 759 760 761 762 763
 * 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
764 765
 *
 *
N
Nick Piggin 已提交
766
 * In order to support COW of arbitrary special mappings, we have VM_MIXEDMAP.
J
Jared Hulbert 已提交
767 768 769 770 771 772 773 774 775
 *
 * 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
776
 */
N
Nick Piggin 已提交
777 778 779 780 781 782 783
#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
784
{
785
	unsigned long pfn = pte_pfn(pte);
N
Nick Piggin 已提交
786 787

	if (HAVE_PTE_SPECIAL) {
788 789
		if (likely(!pte_special(pte)))
			goto check_pfn;
H
Hugh Dickins 已提交
790 791
		if (vma->vm_flags & (VM_PFNMAP | VM_MIXEDMAP))
			return NULL;
H
Hugh Dickins 已提交
792
		if (!is_zero_pfn(pfn))
793
			print_bad_pte(vma, addr, pte, NULL);
N
Nick Piggin 已提交
794 795 796 797 798
		return NULL;
	}

	/* !HAVE_PTE_SPECIAL case follows: */

J
Jared Hulbert 已提交
799 800 801 802 803 804
	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
805 806
			unsigned long off;
			off = (addr - vma->vm_start) >> PAGE_SHIFT;
J
Jared Hulbert 已提交
807 808 809 810 811
			if (pfn == vma->vm_pgoff + off)
				return NULL;
			if (!is_cow_mapping(vma->vm_flags))
				return NULL;
		}
812 813
	}

H
Hugh Dickins 已提交
814 815
	if (is_zero_pfn(pfn))
		return NULL;
816 817 818 819 820
check_pfn:
	if (unlikely(pfn > highest_memmap_pfn)) {
		print_bad_pte(vma, addr, pte, NULL);
		return NULL;
	}
821 822

	/*
N
Nick Piggin 已提交
823 824
	 * NOTE! We still have PageReserved() pages in the page tables.
	 * eg. VDSO mappings can cause them to exist.
825
	 */
J
Jared Hulbert 已提交
826
out:
827
	return pfn_to_page(pfn);
H
Hugh Dickins 已提交
828 829
}

L
Linus Torvalds 已提交
830 831 832 833 834 835
/*
 * 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
836
static inline unsigned long
L
Linus Torvalds 已提交
837
copy_one_pte(struct mm_struct *dst_mm, struct mm_struct *src_mm,
N
Nick Piggin 已提交
838
		pte_t *dst_pte, pte_t *src_pte, struct vm_area_struct *vma,
H
Hugh Dickins 已提交
839
		unsigned long addr, int *rss)
L
Linus Torvalds 已提交
840
{
N
Nick Piggin 已提交
841
	unsigned long vm_flags = vma->vm_flags;
L
Linus Torvalds 已提交
842 843 844 845 846 847
	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)) {
848 849
			swp_entry_t entry = pte_to_swp_entry(pte);

H
Hugh Dickins 已提交
850 851 852
			if (swap_duplicate(entry) < 0)
				return entry.val;

L
Linus Torvalds 已提交
853 854 855
			/* make sure dst_mm is on swapoff's mmlist. */
			if (unlikely(list_empty(&dst_mm->mmlist))) {
				spin_lock(&mmlist_lock);
856 857 858
				if (list_empty(&dst_mm->mmlist))
					list_add(&dst_mm->mmlist,
						 &src_mm->mmlist);
L
Linus Torvalds 已提交
859 860
				spin_unlock(&mmlist_lock);
			}
K
KAMEZAWA Hiroyuki 已提交
861 862
			if (likely(!non_swap_entry(entry)))
				rss[MM_SWAPENTS]++;
863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880
			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);
				}
881
			}
L
Linus Torvalds 已提交
882
		}
883
		goto out_set_pte;
L
Linus Torvalds 已提交
884 885 886 887 888 889
	}

	/*
	 * If it's a COW mapping, write protect it both
	 * in the parent and the child
	 */
890
	if (is_cow_mapping(vm_flags)) {
L
Linus Torvalds 已提交
891
		ptep_set_wrprotect(src_mm, addr, src_pte);
892
		pte = pte_wrprotect(pte);
L
Linus Torvalds 已提交
893 894 895 896 897 898 899 900 901
	}

	/*
	 * 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);
902 903 904 905

	page = vm_normal_page(vma, addr, pte);
	if (page) {
		get_page(page);
H
Hugh Dickins 已提交
906
		page_dup_rmap(page);
K
KAMEZAWA Hiroyuki 已提交
907 908 909 910
		if (PageAnon(page))
			rss[MM_ANONPAGES]++;
		else
			rss[MM_FILEPAGES]++;
911
	}
912 913 914

out_set_pte:
	set_pte_at(dst_mm, addr, dst_pte, pte);
H
Hugh Dickins 已提交
915
	return 0;
L
Linus Torvalds 已提交
916 917
}

918 919 920
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
921
{
922
	pte_t *orig_src_pte, *orig_dst_pte;
L
Linus Torvalds 已提交
923
	pte_t *src_pte, *dst_pte;
H
Hugh Dickins 已提交
924
	spinlock_t *src_ptl, *dst_ptl;
925
	int progress = 0;
K
KAMEZAWA Hiroyuki 已提交
926
	int rss[NR_MM_COUNTERS];
H
Hugh Dickins 已提交
927
	swp_entry_t entry = (swp_entry_t){0};
L
Linus Torvalds 已提交
928 929

again:
K
KAMEZAWA Hiroyuki 已提交
930 931
	init_rss_vec(rss);

H
Hugh Dickins 已提交
932
	dst_pte = pte_alloc_map_lock(dst_mm, dst_pmd, addr, &dst_ptl);
L
Linus Torvalds 已提交
933 934
	if (!dst_pte)
		return -ENOMEM;
P
Peter Zijlstra 已提交
935
	src_pte = pte_offset_map(src_pmd, addr);
H
Hugh Dickins 已提交
936
	src_ptl = pte_lockptr(src_mm, src_pmd);
I
Ingo Molnar 已提交
937
	spin_lock_nested(src_ptl, SINGLE_DEPTH_NESTING);
938 939
	orig_src_pte = src_pte;
	orig_dst_pte = dst_pte;
940
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
941 942 943 944 945 946

	do {
		/*
		 * We are holding two locks at this point - either of them
		 * could generate latencies in another task on another CPU.
		 */
947 948 949
		if (progress >= 32) {
			progress = 0;
			if (need_resched() ||
N
Nick Piggin 已提交
950
			    spin_needbreak(src_ptl) || spin_needbreak(dst_ptl))
951 952
				break;
		}
L
Linus Torvalds 已提交
953 954 955 956
		if (pte_none(*src_pte)) {
			progress++;
			continue;
		}
H
Hugh Dickins 已提交
957 958 959 960
		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
961 962 963
		progress += 8;
	} while (dst_pte++, src_pte++, addr += PAGE_SIZE, addr != end);

964
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
965
	spin_unlock(src_ptl);
P
Peter Zijlstra 已提交
966
	pte_unmap(orig_src_pte);
K
KAMEZAWA Hiroyuki 已提交
967
	add_mm_rss_vec(dst_mm, rss);
968
	pte_unmap_unlock(orig_dst_pte, dst_ptl);
H
Hugh Dickins 已提交
969
	cond_resched();
H
Hugh Dickins 已提交
970 971 972 973 974 975

	if (entry.val) {
		if (add_swap_count_continuation(entry, GFP_KERNEL) < 0)
			return -ENOMEM;
		progress = 0;
	}
L
Linus Torvalds 已提交
976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993
	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);
994 995
		if (pmd_trans_huge(*src_pmd)) {
			int err;
996
			VM_BUG_ON(next-addr != HPAGE_PMD_SIZE);
997 998 999 1000 1001 1002 1003 1004
			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
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 1035 1036 1037 1038 1039 1040 1041 1042
		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;
A
Andrea Arcangeli 已提交
1043
	int ret;
L
Linus Torvalds 已提交
1044

1045 1046 1047 1048 1049 1050
	/*
	 * 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.
	 */
1051
	if (!(vma->vm_flags & (VM_HUGETLB|VM_NONLINEAR|VM_PFNMAP|VM_INSERTPAGE))) {
1052 1053 1054 1055
		if (!vma->anon_vma)
			return 0;
	}

L
Linus Torvalds 已提交
1056 1057 1058
	if (is_vm_hugetlb_page(vma))
		return copy_hugetlb_page_range(dst_mm, src_mm, vma);

1059
	if (unlikely(is_pfn_mapping(vma))) {
1060 1061 1062 1063 1064 1065 1066 1067 1068
		/*
		 * We do not free on error cases below as remove_vma
		 * gets called on error from higher level routine
		 */
		ret = track_pfn_vma_copy(vma);
		if (ret)
			return ret;
	}

A
Andrea Arcangeli 已提交
1069 1070 1071 1072 1073 1074 1075 1076 1077 1078
	/*
	 * 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.
	 */
	if (is_cow_mapping(vma->vm_flags))
		mmu_notifier_invalidate_range_start(src_mm, addr, end);

	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 1094 1095

	if (is_cow_mapping(vma->vm_flags))
		mmu_notifier_invalidate_range_end(src_mm,
						  vma->vm_start, end);
	return ret;
L
Linus Torvalds 已提交
1096 1097
}

1098
static unsigned long zap_pte_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
1099
				struct vm_area_struct *vma, pmd_t *pmd,
L
Linus Torvalds 已提交
1100
				unsigned long addr, unsigned long end,
1101
				struct zap_details *details)
L
Linus Torvalds 已提交
1102
{
N
Nick Piggin 已提交
1103
	struct mm_struct *mm = tlb->mm;
P
Peter Zijlstra 已提交
1104
	int force_flush = 0;
K
KAMEZAWA Hiroyuki 已提交
1105
	int rss[NR_MM_COUNTERS];
1106
	spinlock_t *ptl;
1107
	pte_t *start_pte;
1108
	pte_t *pte;
K
KAMEZAWA Hiroyuki 已提交
1109

P
Peter Zijlstra 已提交
1110
again:
1111
	init_rss_vec(rss);
1112 1113
	start_pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
	pte = start_pte;
1114
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
1115 1116
	do {
		pte_t ptent = *pte;
1117
		if (pte_none(ptent)) {
L
Linus Torvalds 已提交
1118
			continue;
1119
		}
1120

L
Linus Torvalds 已提交
1121
		if (pte_present(ptent)) {
H
Hugh Dickins 已提交
1122
			struct page *page;
1123

1124
			page = vm_normal_page(vma, addr, ptent);
L
Linus Torvalds 已提交
1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142
			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
1143
			ptent = ptep_get_and_clear_full(mm, addr, pte,
1144
							tlb->fullmm);
L
Linus Torvalds 已提交
1145 1146 1147 1148 1149 1150
			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
1151
				set_pte_at(mm, addr, pte,
L
Linus Torvalds 已提交
1152 1153
					   pgoff_to_pte(page->index));
			if (PageAnon(page))
K
KAMEZAWA Hiroyuki 已提交
1154
				rss[MM_ANONPAGES]--;
1155 1156 1157
			else {
				if (pte_dirty(ptent))
					set_page_dirty(page);
1158 1159
				if (pte_young(ptent) &&
				    likely(!VM_SequentialReadHint(vma)))
1160
					mark_page_accessed(page);
K
KAMEZAWA Hiroyuki 已提交
1161
				rss[MM_FILEPAGES]--;
1162
			}
1163
			page_remove_rmap(page);
1164 1165
			if (unlikely(page_mapcount(page) < 0))
				print_bad_pte(vma, addr, ptent, page);
P
Peter Zijlstra 已提交
1166 1167 1168
			force_flush = !__tlb_remove_page(tlb, page);
			if (force_flush)
				break;
L
Linus Torvalds 已提交
1169 1170 1171 1172 1173 1174 1175 1176
			continue;
		}
		/*
		 * If details->check_mapping, we leave swap entries;
		 * if details->nonlinear_vma, we leave file entries.
		 */
		if (unlikely(details))
			continue;
1177 1178 1179
		if (pte_file(ptent)) {
			if (unlikely(!(vma->vm_flags & VM_NONLINEAR)))
				print_bad_pte(vma, addr, ptent, NULL);
K
KAMEZAWA Hiroyuki 已提交
1180 1181 1182 1183 1184
		} else {
			swp_entry_t entry = pte_to_swp_entry(ptent);

			if (!non_swap_entry(entry))
				rss[MM_SWAPENTS]--;
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194
			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
1195 1196 1197
			if (unlikely(!free_swap_and_cache(entry)))
				print_bad_pte(vma, addr, ptent, NULL);
		}
1198
		pte_clear_not_present_full(mm, addr, pte, tlb->fullmm);
1199
	} while (pte++, addr += PAGE_SIZE, addr != end);
1200

K
KAMEZAWA Hiroyuki 已提交
1201
	add_mm_rss_vec(mm, rss);
1202
	arch_leave_lazy_mmu_mode();
1203
	pte_unmap_unlock(start_pte, ptl);
1204

P
Peter Zijlstra 已提交
1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216
	/*
	 * mmu_gather ran out of room to batch pages, we break out of
	 * the PTE lock to avoid doing the potential expensive TLB invalidate
	 * and page-free while holding it.
	 */
	if (force_flush) {
		force_flush = 0;
		tlb_flush_mmu(tlb);
		if (addr != end)
			goto again;
	}

1217
	return addr;
L
Linus Torvalds 已提交
1218 1219
}

1220
static inline unsigned long zap_pmd_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
1221
				struct vm_area_struct *vma, pud_t *pud,
L
Linus Torvalds 已提交
1222
				unsigned long addr, unsigned long end,
1223
				struct zap_details *details)
L
Linus Torvalds 已提交
1224 1225 1226 1227 1228 1229 1230
{
	pmd_t *pmd;
	unsigned long next;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
1231
		if (pmd_trans_huge(*pmd)) {
1232
			if (next - addr != HPAGE_PMD_SIZE) {
1233
				VM_BUG_ON(!rwsem_is_locked(&tlb->mm->mmap_sem));
1234
				split_huge_page_pmd(vma->vm_mm, pmd);
S
Shaohua Li 已提交
1235
			} else if (zap_huge_pmd(tlb, vma, pmd, addr))
1236
				goto next;
1237 1238
			/* fall through */
		}
1239 1240 1241 1242 1243 1244 1245 1246 1247
		/*
		 * 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;
1248
		next = zap_pte_range(tlb, vma, pmd, addr, next, details);
1249
next:
1250 1251
		cond_resched();
	} while (pmd++, addr = next, addr != end);
1252 1253

	return addr;
L
Linus Torvalds 已提交
1254 1255
}

1256
static inline unsigned long zap_pud_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
1257
				struct vm_area_struct *vma, pgd_t *pgd,
L
Linus Torvalds 已提交
1258
				unsigned long addr, unsigned long end,
1259
				struct zap_details *details)
L
Linus Torvalds 已提交
1260 1261 1262 1263 1264 1265 1266
{
	pud_t *pud;
	unsigned long next;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
1267
		if (pud_none_or_clear_bad(pud))
L
Linus Torvalds 已提交
1268
			continue;
1269 1270
		next = zap_pmd_range(tlb, vma, pud, addr, next, details);
	} while (pud++, addr = next, addr != end);
1271 1272

	return addr;
L
Linus Torvalds 已提交
1273 1274
}

1275 1276
static unsigned long unmap_page_range(struct mmu_gather *tlb,
				struct vm_area_struct *vma,
L
Linus Torvalds 已提交
1277
				unsigned long addr, unsigned long end,
1278
				struct zap_details *details)
L
Linus Torvalds 已提交
1279 1280 1281 1282 1283 1284 1285 1286
{
	pgd_t *pgd;
	unsigned long next;

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

	BUG_ON(addr >= end);
1287
	mem_cgroup_uncharge_start();
L
Linus Torvalds 已提交
1288 1289 1290 1291
	tlb_start_vma(tlb, vma);
	pgd = pgd_offset(vma->vm_mm, addr);
	do {
		next = pgd_addr_end(addr, end);
1292
		if (pgd_none_or_clear_bad(pgd))
L
Linus Torvalds 已提交
1293
			continue;
1294 1295
		next = zap_pud_range(tlb, vma, pgd, addr, next, details);
	} while (pgd++, addr = next, addr != end);
L
Linus Torvalds 已提交
1296
	tlb_end_vma(tlb, vma);
1297
	mem_cgroup_uncharge_end();
1298 1299

	return addr;
L
Linus Torvalds 已提交
1300 1301 1302 1303
}

/**
 * unmap_vmas - unmap a range of memory covered by a list of vma's
1304
 * @tlb: address of the caller's struct mmu_gather
L
Linus Torvalds 已提交
1305 1306 1307 1308 1309 1310
 * @vma: the starting vma
 * @start_addr: virtual address at which to start unmapping
 * @end_addr: virtual address at which to end unmapping
 * @nr_accounted: Place number of unmapped pages in vm-accountable vma's here
 * @details: details of nonlinear truncation or shared cache invalidation
 *
1311
 * Returns the end address of the unmapping (restart addr if interrupted).
L
Linus Torvalds 已提交
1312
 *
1313
 * Unmap all pages in the vma list.
L
Linus Torvalds 已提交
1314 1315 1316 1317 1318 1319 1320 1321 1322 1323
 *
 * 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.
 */
P
Peter Zijlstra 已提交
1324
unsigned long unmap_vmas(struct mmu_gather *tlb,
L
Linus Torvalds 已提交
1325 1326 1327 1328
		struct vm_area_struct *vma, unsigned long start_addr,
		unsigned long end_addr, unsigned long *nr_accounted,
		struct zap_details *details)
{
1329
	unsigned long start = start_addr;
A
Andrea Arcangeli 已提交
1330
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
1331

A
Andrea Arcangeli 已提交
1332
	mmu_notifier_invalidate_range_start(mm, start_addr, end_addr);
L
Linus Torvalds 已提交
1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345
	for ( ; vma && vma->vm_start < end_addr; vma = vma->vm_next) {
		unsigned long end;

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

		if (vma->vm_flags & VM_ACCOUNT)
			*nr_accounted += (end - start) >> PAGE_SHIFT;

1346
		if (unlikely(is_pfn_mapping(vma)))
1347 1348
			untrack_pfn_vma(vma, 0, 0);

L
Linus Torvalds 已提交
1349
		while (start != end) {
1350
			if (unlikely(is_vm_hugetlb_page(vma))) {
1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361
				/*
				 * 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.
				 */
1362
				if (vma->vm_file)
1363 1364
					unmap_hugepage_range(vma, start, end, NULL);

1365 1366
				start = end;
			} else
1367
				start = unmap_page_range(tlb, vma, start, end, details);
L
Linus Torvalds 已提交
1368 1369
		}
	}
1370

A
Andrea Arcangeli 已提交
1371
	mmu_notifier_invalidate_range_end(mm, start_addr, end_addr);
1372
	return start;	/* which is now the end (or restart) address */
L
Linus Torvalds 已提交
1373 1374 1375 1376 1377 1378 1379 1380 1381
}

/**
 * zap_page_range - 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
 */
1382
unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
L
Linus Torvalds 已提交
1383 1384 1385
		unsigned long size, struct zap_details *details)
{
	struct mm_struct *mm = vma->vm_mm;
P
Peter Zijlstra 已提交
1386
	struct mmu_gather tlb;
L
Linus Torvalds 已提交
1387 1388 1389 1390
	unsigned long end = address + size;
	unsigned long nr_accounted = 0;

	lru_add_drain();
P
Peter Zijlstra 已提交
1391
	tlb_gather_mmu(&tlb, mm, 0);
1392
	update_hiwater_rss(mm);
1393
	end = unmap_vmas(&tlb, vma, address, end, &nr_accounted, details);
P
Peter Zijlstra 已提交
1394
	tlb_finish_mmu(&tlb, address, end);
1395
	return end;
L
Linus Torvalds 已提交
1396 1397
}

1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420
/**
 * 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;
	zap_page_range(vma, address, size, NULL);
	return 0;
}
EXPORT_SYMBOL_GPL(zap_vma_ptes);

J
Johannes Weiner 已提交
1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431
/**
 * 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
1432
 */
1433
struct page *follow_page(struct vm_area_struct *vma, unsigned long address,
1434
			unsigned int flags)
L
Linus Torvalds 已提交
1435 1436 1437 1438 1439
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *ptep, pte;
1440
	spinlock_t *ptl;
L
Linus Torvalds 已提交
1441
	struct page *page;
1442
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
1443

1444 1445 1446 1447 1448
	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
1449

1450
	page = NULL;
L
Linus Torvalds 已提交
1451 1452
	pgd = pgd_offset(mm, address);
	if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
1453
		goto no_page_table;
L
Linus Torvalds 已提交
1454 1455

	pud = pud_offset(pgd, address);
A
Andi Kleen 已提交
1456
	if (pud_none(*pud))
1457
		goto no_page_table;
1458
	if (pud_huge(*pud) && vma->vm_flags & VM_HUGETLB) {
A
Andi Kleen 已提交
1459 1460 1461 1462 1463 1464 1465
		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
1466
	pmd = pmd_offset(pud, address);
1467
	if (pmd_none(*pmd))
1468
		goto no_page_table;
1469
	if (pmd_huge(*pmd) && vma->vm_flags & VM_HUGETLB) {
1470 1471
		BUG_ON(flags & FOLL_GET);
		page = follow_huge_pmd(mm, address, pmd, flags & FOLL_WRITE);
L
Linus Torvalds 已提交
1472
		goto out;
1473
	}
1474
	if (pmd_trans_huge(*pmd)) {
1475 1476 1477 1478
		if (flags & FOLL_SPLIT) {
			split_huge_page_pmd(mm, pmd);
			goto split_fallthrough;
		}
1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493
		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 {
				page = follow_trans_huge_pmd(mm, address,
							     pmd, flags);
				spin_unlock(&mm->page_table_lock);
				goto out;
			}
		} else
			spin_unlock(&mm->page_table_lock);
		/* fall through */
	}
1494
split_fallthrough:
1495 1496 1497
	if (unlikely(pmd_bad(*pmd)))
		goto no_page_table;

1498
	ptep = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
1499 1500

	pte = *ptep;
1501
	if (!pte_present(pte))
1502
		goto no_page;
1503 1504
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;
H
Hugh Dickins 已提交
1505

1506
	page = vm_normal_page(vma, address, pte);
H
Hugh Dickins 已提交
1507 1508
	if (unlikely(!page)) {
		if ((flags & FOLL_DUMP) ||
H
Hugh Dickins 已提交
1509
		    !is_zero_pfn(pte_pfn(pte)))
H
Hugh Dickins 已提交
1510 1511 1512
			goto bad_page;
		page = pte_page(pte);
	}
L
Linus Torvalds 已提交
1513

1514
	if (flags & FOLL_GET)
1515
		get_page_foll(page);
1516 1517 1518 1519
	if (flags & FOLL_TOUCH) {
		if ((flags & FOLL_WRITE) &&
		    !pte_dirty(pte) && !PageDirty(page))
			set_page_dirty(page);
1520 1521 1522 1523 1524
		/*
		 * 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().
		 */
1525 1526
		mark_page_accessed(page);
	}
1527
	if ((flags & FOLL_MLOCK) && (vma->vm_flags & VM_LOCKED)) {
1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548
		/*
		 * 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 */
			/*
			 * Because we lock page here and migration is
			 * blocked by the pte's page reference, we need
			 * only check for file-cache page truncation.
			 */
			if (page->mapping)
				mlock_vma_page(page);
			unlock_page(page);
		}
	}
1549 1550
unlock:
	pte_unmap_unlock(ptep, ptl);
L
Linus Torvalds 已提交
1551
out:
1552
	return page;
L
Linus Torvalds 已提交
1553

1554 1555 1556 1557 1558 1559 1560 1561
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
1562

1563 1564 1565
no_page_table:
	/*
	 * When core dumping an enormous anonymous area that nobody
H
Hugh Dickins 已提交
1566 1567 1568 1569 1570
	 * 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.
1571
	 */
H
Hugh Dickins 已提交
1572 1573 1574
	if ((flags & FOLL_DUMP) &&
	    (!vma->vm_ops || !vma->vm_ops->fault))
		return ERR_PTR(-EFAULT);
1575
	return page;
L
Linus Torvalds 已提交
1576 1577
}

1578 1579
static inline int stack_guard_page(struct vm_area_struct *vma, unsigned long addr)
{
1580 1581
	return stack_guard_page_start(vma, addr) ||
	       stack_guard_page_end(vma, addr+PAGE_SIZE);
1582 1583
}

H
Huang Ying 已提交
1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632
/**
 * __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
1633
int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
H
Hugh Dickins 已提交
1634
		     unsigned long start, int nr_pages, unsigned int gup_flags,
1635 1636
		     struct page **pages, struct vm_area_struct **vmas,
		     int *nonblocking)
L
Linus Torvalds 已提交
1637 1638
{
	int i;
H
Hugh Dickins 已提交
1639
	unsigned long vm_flags;
L
Linus Torvalds 已提交
1640

1641
	if (nr_pages <= 0)
1642
		return 0;
H
Hugh Dickins 已提交
1643 1644 1645

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

L
Linus Torvalds 已提交
1646 1647
	/* 
	 * Require read or write permissions.
H
Hugh Dickins 已提交
1648
	 * If FOLL_FORCE is set, we only require the "MAY" flags.
L
Linus Torvalds 已提交
1649
	 */
H
Hugh Dickins 已提交
1650 1651 1652 1653
	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);
L
Linus Torvalds 已提交
1654 1655 1656
	i = 0;

	do {
1657
		struct vm_area_struct *vma;
L
Linus Torvalds 已提交
1658 1659

		vma = find_extend_vma(mm, start);
1660
		if (!vma && in_gate_area(mm, start)) {
L
Linus Torvalds 已提交
1661 1662 1663 1664 1665
			unsigned long pg = start & PAGE_MASK;
			pgd_t *pgd;
			pud_t *pud;
			pmd_t *pmd;
			pte_t *pte;
N
Nick Piggin 已提交
1666 1667

			/* user gate pages are read-only */
H
Hugh Dickins 已提交
1668
			if (gup_flags & FOLL_WRITE)
L
Linus Torvalds 已提交
1669 1670 1671 1672 1673 1674 1675 1676 1677
				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);
1678 1679
			if (pmd_none(*pmd))
				return i ? : -EFAULT;
1680
			VM_BUG_ON(pmd_trans_huge(*pmd));
L
Linus Torvalds 已提交
1681
			pte = pte_offset_map(pmd, pg);
1682 1683 1684 1685
			if (pte_none(*pte)) {
				pte_unmap(pte);
				return i ? : -EFAULT;
			}
1686
			vma = get_gate_vma(mm);
L
Linus Torvalds 已提交
1687
			if (pages) {
1688 1689
				struct page *page;

1690
				page = vm_normal_page(vma, start, *pte);
1691 1692 1693 1694 1695 1696 1697 1698 1699
				if (!page) {
					if (!(gup_flags & FOLL_DUMP) &&
					     is_zero_pfn(pte_pfn(*pte)))
						page = pte_page(*pte);
					else {
						pte_unmap(pte);
						return i ? : -EFAULT;
					}
				}
1700
				pages[i] = page;
1701
				get_page(page);
L
Linus Torvalds 已提交
1702 1703
			}
			pte_unmap(pte);
1704
			goto next_page;
L
Linus Torvalds 已提交
1705 1706
		}

N
Nick Piggin 已提交
1707 1708
		if (!vma ||
		    (vma->vm_flags & (VM_IO | VM_PFNMAP)) ||
H
Hugh Dickins 已提交
1709
		    !(vm_flags & vma->vm_flags))
L
Linus Torvalds 已提交
1710 1711
			return i ? : -EFAULT;

H
Hugh Dickins 已提交
1712 1713
		if (is_vm_hugetlb_page(vma)) {
			i = follow_hugetlb_page(mm, vma, pages, vmas,
H
Hugh Dickins 已提交
1714
					&start, &nr_pages, i, gup_flags);
H
Hugh Dickins 已提交
1715 1716
			continue;
		}
1717

L
Linus Torvalds 已提交
1718
		do {
1719
			struct page *page;
H
Hugh Dickins 已提交
1720
			unsigned int foll_flags = gup_flags;
L
Linus Torvalds 已提交
1721

1722
			/*
1723
			 * If we have a pending SIGKILL, don't keep faulting
H
Hugh Dickins 已提交
1724
			 * pages and potentially allocating memory.
1725
			 */
H
Hugh Dickins 已提交
1726
			if (unlikely(fatal_signal_pending(current)))
1727
				return i ? i : -ERESTARTSYS;
1728

1729
			cond_resched();
1730
			while (!(page = follow_page(vma, start, foll_flags))) {
1731
				int ret;
1732 1733
				unsigned int fault_flags = 0;

1734 1735 1736 1737 1738
				/* For mlock, just skip the stack guard page. */
				if (foll_flags & FOLL_MLOCK) {
					if (stack_guard_page(vma, start))
						goto next_page;
				}
1739 1740 1741 1742
				if (foll_flags & FOLL_WRITE)
					fault_flags |= FAULT_FLAG_WRITE;
				if (nonblocking)
					fault_flags |= FAULT_FLAG_ALLOW_RETRY;
1743 1744
				if (foll_flags & FOLL_NOWAIT)
					fault_flags |= (FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_RETRY_NOWAIT);
1745

1746
				ret = handle_mm_fault(mm, vma, start,
1747
							fault_flags);
1748

N
Nick Piggin 已提交
1749 1750 1751
				if (ret & VM_FAULT_ERROR) {
					if (ret & VM_FAULT_OOM)
						return i ? i : -ENOMEM;
1752 1753 1754 1755 1756 1757 1758 1759 1760 1761
					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
1762 1763 1764
						return i ? i : -EFAULT;
					BUG();
				}
1765 1766 1767 1768 1769 1770 1771

				if (tsk) {
					if (ret & VM_FAULT_MAJOR)
						tsk->maj_flt++;
					else
						tsk->min_flt++;
				}
N
Nick Piggin 已提交
1772

1773
				if (ret & VM_FAULT_RETRY) {
1774 1775
					if (nonblocking)
						*nonblocking = 0;
1776 1777 1778
					return i;
				}

1779
				/*
N
Nick Piggin 已提交
1780 1781 1782 1783
				 * 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
1784 1785 1786 1787 1788 1789
				 * 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).
1790
				 */
1791 1792
				if ((ret & VM_FAULT_WRITE) &&
				    !(vma->vm_flags & VM_WRITE))
1793
					foll_flags &= ~FOLL_WRITE;
N
Nick Piggin 已提交
1794

1795
				cond_resched();
L
Linus Torvalds 已提交
1796
			}
1797 1798
			if (IS_ERR(page))
				return i ? i : PTR_ERR(page);
L
Linus Torvalds 已提交
1799
			if (pages) {
1800
				pages[i] = page;
1801

1802
				flush_anon_page(vma, page, start);
1803
				flush_dcache_page(page);
L
Linus Torvalds 已提交
1804
			}
1805
next_page:
L
Linus Torvalds 已提交
1806 1807 1808 1809
			if (vmas)
				vmas[i] = vma;
			i++;
			start += PAGE_SIZE;
1810 1811 1812
			nr_pages--;
		} while (nr_pages && start < vma->vm_end);
	} while (nr_pages);
L
Linus Torvalds 已提交
1813 1814
	return i;
}
H
Huang Ying 已提交
1815
EXPORT_SYMBOL(__get_user_pages);
N
Nick Piggin 已提交
1816

1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873
/*
 * 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;
}

/*
1874
 * get_user_pages() - pin user pages in memory
1875 1876
 * @tsk:	the task_struct to use for page fault accounting, or
 *		NULL if faults are not to be recorded.
1877 1878
 * @mm:		mm_struct of target mm
 * @start:	starting user address
1879
 * @nr_pages:	number of pages from start to pin
1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890
 * @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
1891
 * requested. If nr_pages is 0 or negative, returns 0. If no pages
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
 * 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
1924
int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
1925
		unsigned long start, int nr_pages, int write, int force,
N
Nick Piggin 已提交
1926 1927
		struct page **pages, struct vm_area_struct **vmas)
{
H
Hugh Dickins 已提交
1928
	int flags = FOLL_TOUCH;
N
Nick Piggin 已提交
1929

H
Hugh Dickins 已提交
1930 1931
	if (pages)
		flags |= FOLL_GET;
N
Nick Piggin 已提交
1932
	if (write)
H
Hugh Dickins 已提交
1933
		flags |= FOLL_WRITE;
N
Nick Piggin 已提交
1934
	if (force)
H
Hugh Dickins 已提交
1935
		flags |= FOLL_FORCE;
N
Nick Piggin 已提交
1936

1937 1938
	return __get_user_pages(tsk, mm, start, nr_pages, flags, pages, vmas,
				NULL);
N
Nick Piggin 已提交
1939
}
L
Linus Torvalds 已提交
1940 1941
EXPORT_SYMBOL(get_user_pages);

H
Hugh Dickins 已提交
1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962
/**
 * 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,
1963 1964
			     FOLL_FORCE | FOLL_DUMP | FOLL_GET, &page, &vma,
			     NULL) < 1)
H
Hugh Dickins 已提交
1965 1966 1967 1968 1969 1970
		return NULL;
	flush_cache_page(vma, addr, page_to_pfn(page));
	return page;
}
#endif /* CONFIG_ELF_CORE */

1971
pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
H
Harvey Harrison 已提交
1972
			spinlock_t **ptl)
1973 1974 1975 1976
{
	pgd_t * pgd = pgd_offset(mm, addr);
	pud_t * pud = pud_alloc(mm, pgd, addr);
	if (pud) {
1977
		pmd_t * pmd = pmd_alloc(mm, pud, addr);
1978 1979
		if (pmd) {
			VM_BUG_ON(pmd_trans_huge(*pmd));
1980
			return pte_alloc_map_lock(mm, pmd, addr, ptl);
1981
		}
1982 1983 1984 1985
	}
	return NULL;
}

1986 1987 1988 1989 1990 1991 1992
/*
 * 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
1993 1994
static int insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page, pgprot_t prot)
1995
{
N
Nick Piggin 已提交
1996
	struct mm_struct *mm = vma->vm_mm;
1997
	int retval;
1998
	pte_t *pte;
1999 2000
	spinlock_t *ptl;

2001
	retval = -EINVAL;
2002
	if (PageAnon(page))
2003
		goto out;
2004 2005
	retval = -ENOMEM;
	flush_dcache_page(page);
2006
	pte = get_locked_pte(mm, addr, &ptl);
2007
	if (!pte)
2008
		goto out;
2009 2010 2011 2012 2013 2014
	retval = -EBUSY;
	if (!pte_none(*pte))
		goto out_unlock;

	/* Ok, finally just insert the thing.. */
	get_page(page);
2015
	inc_mm_counter_fast(mm, MM_FILEPAGES);
2016 2017 2018 2019
	page_add_file_rmap(page);
	set_pte_at(mm, addr, pte, mk_pte(page, prot));

	retval = 0;
2020 2021
	pte_unmap_unlock(pte, ptl);
	return retval;
2022 2023 2024 2025 2026 2027
out_unlock:
	pte_unmap_unlock(pte, ptl);
out:
	return retval;
}

2028 2029 2030 2031 2032 2033
/**
 * 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
 *
2034 2035 2036 2037 2038 2039
 * 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
2040
 * (see split_page()).
2041 2042 2043 2044 2045 2046 2047 2048 2049
 *
 * 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.
 */
N
Nick Piggin 已提交
2050 2051
int vm_insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page)
2052 2053 2054 2055 2056
{
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
	if (!page_count(page))
		return -EINVAL;
2057
	vma->vm_flags |= VM_INSERTPAGE;
N
Nick Piggin 已提交
2058
	return insert_page(vma, addr, page, vma->vm_page_prot);
2059
}
2060
EXPORT_SYMBOL(vm_insert_page);
2061

N
Nick Piggin 已提交
2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080
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);
2081
	update_mmu_cache(vma, addr, pte); /* XXX: why not for insert_page? */
N
Nick Piggin 已提交
2082 2083 2084 2085 2086 2087 2088 2089

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

N
Nick Piggin 已提交
2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100
/**
 * 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
 *
 * Similar to vm_inert_page, this allows drivers to insert individual pages
 * 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
2101 2102 2103 2104 2105
 *
 * 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
2106 2107
 */
int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
N
Nick Piggin 已提交
2108
			unsigned long pfn)
N
Nick Piggin 已提交
2109
{
2110
	int ret;
2111
	pgprot_t pgprot = vma->vm_page_prot;
N
Nick Piggin 已提交
2112 2113 2114 2115 2116 2117
	/*
	 * 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
2118 2119 2120 2121 2122
	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
2123

N
Nick Piggin 已提交
2124 2125
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
2126
	if (track_pfn_vma_new(vma, &pgprot, pfn, PAGE_SIZE))
2127 2128
		return -EINVAL;

2129
	ret = insert_pfn(vma, addr, pfn, pgprot);
2130 2131 2132 2133 2134

	if (ret)
		untrack_pfn_vma(vma, pfn, PAGE_SIZE);

	return ret;
N
Nick Piggin 已提交
2135 2136
}
EXPORT_SYMBOL(vm_insert_pfn);
N
Nick Piggin 已提交
2137

N
Nick Piggin 已提交
2138 2139 2140 2141
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
2142

N
Nick Piggin 已提交
2143 2144
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
N
Nick Piggin 已提交
2145

N
Nick Piggin 已提交
2146 2147 2148 2149
	/*
	 * 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
2150 2151
	 * 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
2152 2153 2154 2155 2156 2157 2158 2159
	 */
	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
2160
}
N
Nick Piggin 已提交
2161
EXPORT_SYMBOL(vm_insert_mixed);
N
Nick Piggin 已提交
2162

L
Linus Torvalds 已提交
2163 2164 2165 2166 2167 2168 2169 2170 2171 2172
/*
 * 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
2173
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2174

H
Hugh Dickins 已提交
2175
	pte = pte_alloc_map_lock(mm, pmd, addr, &ptl);
L
Linus Torvalds 已提交
2176 2177
	if (!pte)
		return -ENOMEM;
2178
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
2179 2180
	do {
		BUG_ON(!pte_none(*pte));
N
Nick Piggin 已提交
2181
		set_pte_at(mm, addr, pte, pte_mkspecial(pfn_pte(pfn, prot)));
L
Linus Torvalds 已提交
2182 2183
		pfn++;
	} while (pte++, addr += PAGE_SIZE, addr != end);
2184
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
2185
	pte_unmap_unlock(pte - 1, ptl);
L
Linus Torvalds 已提交
2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199
	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;
2200
	VM_BUG_ON(pmd_trans_huge(*pmd));
L
Linus Torvalds 已提交
2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229
	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;
}

2230 2231 2232 2233 2234 2235 2236 2237 2238 2239
/**
 * 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
2240 2241 2242 2243 2244
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;
2245
	unsigned long end = addr + PAGE_ALIGN(size);
L
Linus Torvalds 已提交
2246 2247 2248 2249 2250 2251 2252 2253
	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).
H
Hugh Dickins 已提交
2254 2255 2256 2257 2258
	 *   VM_RESERVED is specified all over the place, because
	 *	in 2.4 it kept swapout's vma scan off this vma; but
	 *	in 2.6 the LRU scan won't even find its pages, so this
	 *	flag means no more than count its pages in reserved_vm,
	 * 	and omit it from core dump, even when VM_IO turned off.
2259 2260 2261
	 *   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.
L
Linus Torvalds 已提交
2262 2263 2264 2265
	 *
	 * 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".
L
Linus Torvalds 已提交
2266
	 */
2267
	if (addr == vma->vm_start && end == vma->vm_end) {
L
Linus Torvalds 已提交
2268
		vma->vm_pgoff = pfn;
2269
		vma->vm_flags |= VM_PFN_AT_MMAP;
2270
	} else if (is_cow_mapping(vma->vm_flags))
2271
		return -EINVAL;
L
Linus Torvalds 已提交
2272

2273
	vma->vm_flags |= VM_IO | VM_RESERVED | VM_PFNMAP;
L
Linus Torvalds 已提交
2274

2275
	err = track_pfn_vma_new(vma, &prot, pfn, PAGE_ALIGN(size));
2276 2277 2278 2279 2280 2281
	if (err) {
		/*
		 * To indicate that track_pfn related cleanup is not
		 * needed from higher level routine calling unmap_vmas
		 */
		vma->vm_flags &= ~(VM_IO | VM_RESERVED | VM_PFNMAP);
2282
		vma->vm_flags &= ~VM_PFN_AT_MMAP;
2283
		return -EINVAL;
2284
	}
2285

L
Linus Torvalds 已提交
2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296
	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);
2297 2298 2299 2300

	if (err)
		untrack_pfn_vma(vma, pfn, PAGE_ALIGN(size));

L
Linus Torvalds 已提交
2301 2302 2303 2304
	return err;
}
EXPORT_SYMBOL(remap_pfn_range);

2305 2306 2307 2308 2309 2310
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;
2311
	pgtable_t token;
2312
	spinlock_t *uninitialized_var(ptl);
2313 2314 2315 2316 2317 2318 2319 2320 2321

	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));

2322 2323
	arch_enter_lazy_mmu_mode();

2324
	token = pmd_pgtable(*pmd);
2325 2326

	do {
2327
		err = fn(pte++, token, addr, data);
2328 2329
		if (err)
			break;
2330
	} while (addr += PAGE_SIZE, addr != end);
2331

2332 2333
	arch_leave_lazy_mmu_mode();

2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346
	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
2347 2348
	BUG_ON(pud_huge(*pud));

2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389
	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;
2390
	unsigned long end = addr + size;
2391 2392 2393 2394 2395 2396 2397 2398 2399 2400
	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);
2401

2402 2403 2404 2405
	return err;
}
EXPORT_SYMBOL_GPL(apply_to_page_range);

2406 2407 2408 2409
/*
 * 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
2410
 * might give a mix of unmatched parts, do_swap_page and do_nonlinear_fault
2411 2412
 * must check under lock before unmapping the pte and proceeding
 * (but do_wp_page is only called after already making such a check;
2413
 * and do_anonymous_page can safely check later on).
2414
 */
H
Hugh Dickins 已提交
2415
static inline int pte_unmap_same(struct mm_struct *mm, pmd_t *pmd,
2416 2417 2418 2419 2420
				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
2421 2422
		spinlock_t *ptl = pte_lockptr(mm, pmd);
		spin_lock(ptl);
2423
		same = pte_same(*page_table, orig_pte);
H
Hugh Dickins 已提交
2424
		spin_unlock(ptl);
2425 2426 2427 2428 2429 2430
	}
#endif
	pte_unmap(page_table);
	return same;
}

2431
static inline void cow_user_page(struct page *dst, struct page *src, unsigned long va, struct vm_area_struct *vma)
2432 2433 2434 2435 2436 2437 2438 2439
{
	/*
	 * 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)) {
2440
		void *kaddr = kmap_atomic(dst);
L
Linus Torvalds 已提交
2441 2442 2443 2444 2445 2446 2447 2448 2449
		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))
2450
			clear_page(kaddr);
2451
		kunmap_atomic(kaddr);
2452
		flush_dcache_page(dst);
N
Nick Piggin 已提交
2453 2454
	} else
		copy_user_highpage(dst, src, va, vma);
2455 2456
}

L
Linus Torvalds 已提交
2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470
/*
 * 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.
 *
2471 2472 2473
 * 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
2474
 */
2475 2476
static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2477
		spinlock_t *ptl, pte_t orig_pte)
2478
	__releases(ptl)
L
Linus Torvalds 已提交
2479
{
2480
	struct page *old_page, *new_page;
L
Linus Torvalds 已提交
2481
	pte_t entry;
2482
	int ret = 0;
2483
	int page_mkwrite = 0;
2484
	struct page *dirty_page = NULL;
L
Linus Torvalds 已提交
2485

2486
	old_page = vm_normal_page(vma, address, orig_pte);
2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497
	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;
2498
		goto gotten;
2499
	}
L
Linus Torvalds 已提交
2500

2501
	/*
P
Peter Zijlstra 已提交
2502 2503
	 * Take out anonymous pages first, anonymous shared vmas are
	 * not dirty accountable.
2504
	 */
H
Hugh Dickins 已提交
2505
	if (PageAnon(old_page) && !PageKsm(old_page)) {
2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516
		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
2517
		}
2518
		if (reuse_swap_page(old_page)) {
2519 2520 2521 2522 2523 2524
			/*
			 * 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);
2525 2526 2527
			unlock_page(old_page);
			goto reuse;
		}
2528
		unlock_page(old_page);
P
Peter Zijlstra 已提交
2529
	} else if (unlikely((vma->vm_flags & (VM_WRITE|VM_SHARED)) ==
2530
					(VM_WRITE|VM_SHARED))) {
P
Peter Zijlstra 已提交
2531 2532 2533 2534 2535
		/*
		 * Only catch write-faults on shared writable pages,
		 * read-only shared pages can get COWed by
		 * get_user_pages(.write=1, .force=1).
		 */
2536
		if (vma->vm_ops && vma->vm_ops->page_mkwrite) {
2537 2538 2539 2540 2541 2542 2543 2544 2545
			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;

2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556
			/*
			 * 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);

2557 2558 2559 2560
			tmp = vma->vm_ops->page_mkwrite(vma, &vmf);
			if (unlikely(tmp &
					(VM_FAULT_ERROR | VM_FAULT_NOPAGE))) {
				ret = tmp;
2561
				goto unwritable_page;
2562
			}
N
Nick Piggin 已提交
2563 2564 2565 2566 2567 2568 2569 2570 2571
			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));
2572 2573 2574 2575 2576 2577 2578 2579 2580

			/*
			 * 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
2581 2582
			if (!pte_same(*page_table, orig_pte)) {
				unlock_page(old_page);
2583
				goto unlock;
N
Nick Piggin 已提交
2584
			}
2585 2586

			page_mkwrite = 1;
L
Linus Torvalds 已提交
2587
		}
2588 2589
		dirty_page = old_page;
		get_page(dirty_page);
2590

2591
reuse:
2592 2593 2594
		flush_cache_page(vma, address, pte_pfn(orig_pte));
		entry = pte_mkyoung(orig_pte);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
2595
		if (ptep_set_access_flags(vma, address, page_table, entry,1))
2596
			update_mmu_cache(vma, address, page_table);
2597
		pte_unmap_unlock(page_table, ptl);
2598
		ret |= VM_FAULT_WRITE;
2599 2600 2601 2602 2603 2604 2605 2606 2607 2608

		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.
		 *
2609
		 * __do_fault is protected similarly.
2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635
		 */
		if (!page_mkwrite) {
			wait_on_page_locked(dirty_page);
			set_page_dirty_balance(dirty_page, page_mkwrite);
		}
		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);
			}
		}

		/* file_update_time outside page_lock */
		if (vma->vm_file)
			file_update_time(vma->vm_file);

		return ret;
L
Linus Torvalds 已提交
2636 2637 2638 2639 2640
	}

	/*
	 * Ok, we need to copy. Oh, well..
	 */
N
Nick Piggin 已提交
2641
	page_cache_get(old_page);
H
Hugh Dickins 已提交
2642
gotten:
2643
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
2644 2645

	if (unlikely(anon_vma_prepare(vma)))
2646
		goto oom;
H
Hugh Dickins 已提交
2647

H
Hugh Dickins 已提交
2648
	if (is_zero_pfn(pte_pfn(orig_pte))) {
H
Hugh Dickins 已提交
2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659
		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
2660
	if (mem_cgroup_newpage_charge(new_page, mm, GFP_KERNEL))
2661 2662
		goto oom_free_new;

L
Linus Torvalds 已提交
2663 2664 2665
	/*
	 * Re-check the pte - we dropped the lock
	 */
2666
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
2667
	if (likely(pte_same(*page_table, orig_pte))) {
H
Hugh Dickins 已提交
2668 2669
		if (old_page) {
			if (!PageAnon(old_page)) {
2670 2671
				dec_mm_counter_fast(mm, MM_FILEPAGES);
				inc_mm_counter_fast(mm, MM_ANONPAGES);
H
Hugh Dickins 已提交
2672 2673
			}
		} else
2674
			inc_mm_counter_fast(mm, MM_ANONPAGES);
2675
		flush_cache_page(vma, address, pte_pfn(orig_pte));
2676 2677
		entry = mk_pte(new_page, vma->vm_page_prot);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
2678 2679 2680 2681 2682 2683
		/*
		 * 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.
		 */
2684
		ptep_clear_flush(vma, address, page_table);
N
Nick Piggin 已提交
2685
		page_add_new_anon_rmap(new_page, vma, address);
2686 2687 2688 2689 2690 2691
		/*
		 * 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);
2692
		update_mmu_cache(vma, address, page_table);
N
Nick Piggin 已提交
2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715
		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.
			 */
2716
			page_remove_rmap(old_page);
N
Nick Piggin 已提交
2717 2718
		}

L
Linus Torvalds 已提交
2719 2720
		/* Free the old page.. */
		new_page = old_page;
N
Nick Piggin 已提交
2721
		ret |= VM_FAULT_WRITE;
2722 2723 2724
	} else
		mem_cgroup_uncharge_page(new_page);

H
Hugh Dickins 已提交
2725 2726
	if (new_page)
		page_cache_release(new_page);
2727
unlock:
2728
	pte_unmap_unlock(page_table, ptl);
2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740
	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
2741
	return ret;
2742
oom_free_new:
2743
	page_cache_release(new_page);
2744
oom:
N
Nick Piggin 已提交
2745 2746 2747 2748 2749
	if (old_page) {
		if (page_mkwrite) {
			unlock_page(old_page);
			page_cache_release(old_page);
		}
H
Hugh Dickins 已提交
2750
		page_cache_release(old_page);
N
Nick Piggin 已提交
2751
	}
L
Linus Torvalds 已提交
2752
	return VM_FAULT_OOM;
2753 2754 2755

unwritable_page:
	page_cache_release(old_page);
2756
	return ret;
L
Linus Torvalds 已提交
2757 2758
}

2759
static void unmap_mapping_range_vma(struct vm_area_struct *vma,
L
Linus Torvalds 已提交
2760 2761 2762
		unsigned long start_addr, unsigned long end_addr,
		struct zap_details *details)
{
2763
	zap_page_range(vma, start_addr, end_addr - start_addr, details);
L
Linus Torvalds 已提交
2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785
}

static inline void unmap_mapping_range_tree(struct prio_tree_root *root,
					    struct zap_details *details)
{
	struct vm_area_struct *vma;
	struct prio_tree_iter iter;
	pgoff_t vba, vea, zba, zea;

	vma_prio_tree_foreach(vma, &iter, root,
			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;

2786
		unmap_mapping_range_vma(vma,
L
Linus Torvalds 已提交
2787 2788
			((zba - vba) << PAGE_SHIFT) + vma->vm_start,
			((zea - vba + 1) << PAGE_SHIFT) + vma->vm_start,
2789
				details);
L
Linus Torvalds 已提交
2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805
	}
}

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.
	 */
	list_for_each_entry(vma, head, shared.vm_set.list) {
		details->nonlinear_vma = vma;
2806
		unmap_mapping_range_vma(vma, vma->vm_start, vma->vm_end, details);
L
Linus Torvalds 已提交
2807 2808 2809 2810
	}
}

/**
2811
 * 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
2812
 * @mapping: the address space containing mmaps to be unmapped.
L
Linus Torvalds 已提交
2813 2814
 * @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
2815
 * boundary.  Note that this is different from truncate_pagecache(), which
L
Linus Torvalds 已提交
2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846
 * 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;


2847
	mutex_lock(&mapping->i_mmap_mutex);
L
Linus Torvalds 已提交
2848 2849 2850 2851
	if (unlikely(!prio_tree_empty(&mapping->i_mmap)))
		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);
2852
	mutex_unlock(&mapping->i_mmap_mutex);
L
Linus Torvalds 已提交
2853 2854 2855 2856
}
EXPORT_SYMBOL(unmap_mapping_range);

/*
2857 2858 2859
 * 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
2860
 */
2861 2862
static int do_swap_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2863
		unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
2864
{
2865
	spinlock_t *ptl;
A
Andrea Arcangeli 已提交
2866
	struct page *page, *swapcache = NULL;
2867
	swp_entry_t entry;
L
Linus Torvalds 已提交
2868
	pte_t pte;
2869
	int locked;
2870
	struct mem_cgroup *ptr;
2871
	int exclusive = 0;
N
Nick Piggin 已提交
2872
	int ret = 0;
L
Linus Torvalds 已提交
2873

H
Hugh Dickins 已提交
2874
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
2875
		goto out;
2876 2877

	entry = pte_to_swp_entry(orig_pte);
2878 2879 2880 2881 2882 2883 2884
	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
2885
			ret = VM_FAULT_SIGBUS;
2886
		}
2887 2888
		goto out;
	}
2889
	delayacct_set_flag(DELAYACCT_PF_SWAPIN);
L
Linus Torvalds 已提交
2890 2891
	page = lookup_swap_cache(entry);
	if (!page) {
H
Hugh Dickins 已提交
2892
		grab_swap_token(mm); /* Contend for token _before_ read-in */
2893 2894
		page = swapin_readahead(entry,
					GFP_HIGHUSER_MOVABLE, vma, address);
L
Linus Torvalds 已提交
2895 2896
		if (!page) {
			/*
2897 2898
			 * Back out if somebody else faulted in this pte
			 * while we released the pte lock.
L
Linus Torvalds 已提交
2899
			 */
2900
			page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
2901 2902
			if (likely(pte_same(*page_table, orig_pte)))
				ret = VM_FAULT_OOM;
2903
			delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2904
			goto unlock;
L
Linus Torvalds 已提交
2905 2906 2907 2908
		}

		/* Had to read the page from swap area: Major fault */
		ret = VM_FAULT_MAJOR;
2909
		count_vm_event(PGMAJFAULT);
2910
		mem_cgroup_count_vm_event(mm, PGMAJFAULT);
2911
	} else if (PageHWPoison(page)) {
2912 2913 2914 2915
		/*
		 * hwpoisoned dirty swapcache pages are kept for killing
		 * owner processes (which may be unknown at hwpoison time)
		 */
2916 2917
		ret = VM_FAULT_HWPOISON;
		delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2918
		goto out_release;
L
Linus Torvalds 已提交
2919 2920
	}

2921
	locked = lock_page_or_retry(page, mm, flags);
2922
	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2923 2924 2925 2926
	if (!locked) {
		ret |= VM_FAULT_RETRY;
		goto out_release;
	}
2927

A
Andrea Arcangeli 已提交
2928
	/*
2929 2930 2931 2932
	 * 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
2933
	 */
2934
	if (unlikely(!PageSwapCache(page) || page_private(page) != entry.val))
A
Andrea Arcangeli 已提交
2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946
		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
2947 2948
	}

K
KAMEZAWA Hiroyuki 已提交
2949
	if (mem_cgroup_try_charge_swapin(mm, page, GFP_KERNEL, &ptr)) {
2950
		ret = VM_FAULT_OOM;
2951
		goto out_page;
2952 2953
	}

L
Linus Torvalds 已提交
2954
	/*
2955
	 * Back out if somebody else already faulted in this pte.
L
Linus Torvalds 已提交
2956
	 */
2957
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
H
Hugh Dickins 已提交
2958
	if (unlikely(!pte_same(*page_table, orig_pte)))
2959 2960 2961 2962 2963
		goto out_nomap;

	if (unlikely(!PageUptodate(page))) {
		ret = VM_FAULT_SIGBUS;
		goto out_nomap;
L
Linus Torvalds 已提交
2964 2965
	}

2966 2967 2968 2969 2970 2971 2972 2973
	/*
	 * 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.
2974 2975 2976 2977
	 * 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().
2978
	 */
L
Linus Torvalds 已提交
2979

2980
	inc_mm_counter_fast(mm, MM_ANONPAGES);
K
KAMEZAWA Hiroyuki 已提交
2981
	dec_mm_counter_fast(mm, MM_SWAPENTS);
L
Linus Torvalds 已提交
2982
	pte = mk_pte(page, vma->vm_page_prot);
2983
	if ((flags & FAULT_FLAG_WRITE) && reuse_swap_page(page)) {
L
Linus Torvalds 已提交
2984
		pte = maybe_mkwrite(pte_mkdirty(pte), vma);
2985
		flags &= ~FAULT_FLAG_WRITE;
2986
		ret |= VM_FAULT_WRITE;
2987
		exclusive = 1;
L
Linus Torvalds 已提交
2988 2989 2990
	}
	flush_icache_page(vma, page);
	set_pte_at(mm, address, page_table, pte);
2991
	do_page_add_anon_rmap(page, vma, address, exclusive);
2992 2993
	/* It's better to call commit-charge after rmap is established */
	mem_cgroup_commit_charge_swapin(page, ptr);
L
Linus Torvalds 已提交
2994

2995
	swap_free(entry);
N
Nick Piggin 已提交
2996
	if (vm_swap_full() || (vma->vm_flags & VM_LOCKED) || PageMlocked(page))
2997
		try_to_free_swap(page);
2998
	unlock_page(page);
A
Andrea Arcangeli 已提交
2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010
	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);
	}
3011

3012
	if (flags & FAULT_FLAG_WRITE) {
3013 3014 3015
		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
3016 3017 3018 3019
		goto out;
	}

	/* No need to invalidate - it was non-present before */
3020
	update_mmu_cache(vma, address, page_table);
3021
unlock:
3022
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
3023 3024
out:
	return ret;
3025
out_nomap:
3026
	mem_cgroup_cancel_charge_swapin(ptr);
3027
	pte_unmap_unlock(page_table, ptl);
3028
out_page:
3029
	unlock_page(page);
3030
out_release:
3031
	page_cache_release(page);
A
Andrea Arcangeli 已提交
3032 3033 3034 3035
	if (swapcache) {
		unlock_page(swapcache);
		page_cache_release(swapcache);
	}
3036
	return ret;
L
Linus Torvalds 已提交
3037 3038
}

3039
/*
3040 3041
 * This is like a special single-page "expand_{down|up}wards()",
 * except we must first make sure that 'address{-|+}PAGE_SIZE'
3042 3043 3044 3045 3046 3047
 * 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) {
3048 3049 3050 3051 3052 3053 3054 3055 3056 3057
		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;
3058

3059
		expand_downwards(vma, address - PAGE_SIZE);
3060
	}
3061 3062 3063 3064 3065 3066 3067 3068 3069
	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);
	}
3070 3071 3072
	return 0;
}

L
Linus Torvalds 已提交
3073
/*
3074 3075 3076
 * 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
3077
 */
3078 3079
static int do_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
3080
		unsigned int flags)
L
Linus Torvalds 已提交
3081
{
3082 3083
	struct page *page;
	spinlock_t *ptl;
L
Linus Torvalds 已提交
3084 3085
	pte_t entry;

3086 3087 3088 3089
	pte_unmap(page_table);

	/* Check if we need to add a guard page to the stack */
	if (check_stack_guard_page(vma, address) < 0)
3090 3091
		return VM_FAULT_SIGBUS;

3092
	/* Use the zero-page for reads */
H
Hugh Dickins 已提交
3093 3094 3095
	if (!(flags & FAULT_FLAG_WRITE)) {
		entry = pte_mkspecial(pfn_pte(my_zero_pfn(address),
						vma->vm_page_prot));
3096
		page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
H
Hugh Dickins 已提交
3097 3098 3099 3100 3101
		if (!pte_none(*page_table))
			goto unlock;
		goto setpte;
	}

N
Nick Piggin 已提交
3102 3103 3104 3105 3106 3107
	/* 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
3108
	__SetPageUptodate(page);
3109

K
KAMEZAWA Hiroyuki 已提交
3110
	if (mem_cgroup_newpage_charge(page, mm, GFP_KERNEL))
3111 3112
		goto oom_free_page;

N
Nick Piggin 已提交
3113
	entry = mk_pte(page, vma->vm_page_prot);
H
Hugh Dickins 已提交
3114 3115
	if (vma->vm_flags & VM_WRITE)
		entry = pte_mkwrite(pte_mkdirty(entry));
L
Linus Torvalds 已提交
3116

N
Nick Piggin 已提交
3117
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
3118
	if (!pte_none(*page_table))
N
Nick Piggin 已提交
3119
		goto release;
H
Hugh Dickins 已提交
3120

3121
	inc_mm_counter_fast(mm, MM_ANONPAGES);
N
Nick Piggin 已提交
3122
	page_add_new_anon_rmap(page, vma, address);
H
Hugh Dickins 已提交
3123
setpte:
3124
	set_pte_at(mm, address, page_table, entry);
L
Linus Torvalds 已提交
3125 3126

	/* No need to invalidate - it was non-present before */
3127
	update_mmu_cache(vma, address, page_table);
3128
unlock:
3129
	pte_unmap_unlock(page_table, ptl);
N
Nick Piggin 已提交
3130
	return 0;
3131
release:
3132
	mem_cgroup_uncharge_page(page);
3133 3134
	page_cache_release(page);
	goto unlock;
3135
oom_free_page:
3136
	page_cache_release(page);
3137
oom:
L
Linus Torvalds 已提交
3138 3139 3140 3141
	return VM_FAULT_OOM;
}

/*
3142
 * __do_fault() tries to create a new page mapping. It aggressively
L
Linus Torvalds 已提交
3143
 * tries to share with existing pages, but makes a separate copy if
3144 3145
 * the FAULT_FLAG_WRITE is set in the flags parameter in order to avoid
 * the next page fault.
L
Linus Torvalds 已提交
3146 3147 3148 3149
 *
 * As this is called only for pages that do not currently exist, we
 * do not need to flush old virtual caches or the TLB.
 *
3150
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
3151
 * but allow concurrent faults), and pte neither mapped nor locked.
3152
 * We return with mmap_sem still held, but pte unmapped and unlocked.
L
Linus Torvalds 已提交
3153
 */
3154
static int __do_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3155
		unsigned long address, pmd_t *pmd,
3156
		pgoff_t pgoff, unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
3157
{
3158
	pte_t *page_table;
3159
	spinlock_t *ptl;
N
Nick Piggin 已提交
3160
	struct page *page;
3161
	struct page *cow_page;
L
Linus Torvalds 已提交
3162 3163
	pte_t entry;
	int anon = 0;
3164
	struct page *dirty_page = NULL;
N
Nick Piggin 已提交
3165 3166
	struct vm_fault vmf;
	int ret;
3167
	int page_mkwrite = 0;
3168

3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188
	/*
	 * 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
3189 3190 3191 3192
	vmf.virtual_address = (void __user *)(address & PAGE_MASK);
	vmf.pgoff = pgoff;
	vmf.flags = flags;
	vmf.page = NULL;
L
Linus Torvalds 已提交
3193

N
Nick Piggin 已提交
3194
	ret = vma->vm_ops->fault(vma, &vmf);
3195 3196
	if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE |
			    VM_FAULT_RETRY)))
3197
		goto uncharge_out;
L
Linus Torvalds 已提交
3198

3199 3200 3201
	if (unlikely(PageHWPoison(vmf.page))) {
		if (ret & VM_FAULT_LOCKED)
			unlock_page(vmf.page);
3202 3203
		ret = VM_FAULT_HWPOISON;
		goto uncharge_out;
3204 3205
	}

3206
	/*
N
Nick Piggin 已提交
3207
	 * For consistency in subsequent calls, make the faulted page always
3208 3209
	 * locked.
	 */
N
Nick Piggin 已提交
3210
	if (unlikely(!(ret & VM_FAULT_LOCKED)))
N
Nick Piggin 已提交
3211
		lock_page(vmf.page);
3212
	else
N
Nick Piggin 已提交
3213
		VM_BUG_ON(!PageLocked(vmf.page));
3214

L
Linus Torvalds 已提交
3215 3216 3217
	/*
	 * Should we do an early C-O-W break?
	 */
N
Nick Piggin 已提交
3218
	page = vmf.page;
3219
	if (flags & FAULT_FLAG_WRITE) {
3220
		if (!(vma->vm_flags & VM_SHARED)) {
3221
			page = cow_page;
3222
			anon = 1;
N
Nick Piggin 已提交
3223
			copy_user_highpage(page, vmf.page, address, vma);
N
Nick Piggin 已提交
3224
			__SetPageUptodate(page);
3225
		} else {
3226 3227
			/*
			 * If the page will be shareable, see if the backing
3228
			 * address space wants to know that the page is about
3229 3230
			 * to become writable
			 */
3231
			if (vma->vm_ops->page_mkwrite) {
3232 3233
				int tmp;

3234
				unlock_page(page);
N
Nick Piggin 已提交
3235
				vmf.flags = FAULT_FLAG_WRITE|FAULT_FLAG_MKWRITE;
3236 3237 3238 3239
				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
3240
					goto unwritable_page;
N
Nick Piggin 已提交
3241
				}
N
Nick Piggin 已提交
3242 3243 3244 3245 3246 3247 3248 3249 3250
				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));
3251
				page_mkwrite = 1;
3252 3253
			}
		}
3254

L
Linus Torvalds 已提交
3255 3256
	}

3257
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
3258 3259 3260 3261 3262 3263

	/*
	 * This silly early PAGE_DIRTY setting removes a race
	 * due to the bad i386 page protection. But it's valid
	 * for other architectures too.
	 *
3264
	 * Note that if FAULT_FLAG_WRITE is set, we either now have
L
Linus Torvalds 已提交
3265 3266 3267 3268 3269
	 * 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... */
3270
	if (likely(pte_same(*page_table, orig_pte))) {
3271 3272
		flush_icache_page(vma, page);
		entry = mk_pte(page, vma->vm_page_prot);
3273
		if (flags & FAULT_FLAG_WRITE)
L
Linus Torvalds 已提交
3274 3275
			entry = maybe_mkwrite(pte_mkdirty(entry), vma);
		if (anon) {
3276
			inc_mm_counter_fast(mm, MM_ANONPAGES);
3277
			page_add_new_anon_rmap(page, vma, address);
3278
		} else {
3279
			inc_mm_counter_fast(mm, MM_FILEPAGES);
3280
			page_add_file_rmap(page);
3281
			if (flags & FAULT_FLAG_WRITE) {
3282
				dirty_page = page;
3283 3284
				get_page(dirty_page);
			}
3285
		}
3286
		set_pte_at(mm, address, page_table, entry);
3287 3288

		/* no need to invalidate: a not-present page won't be cached */
3289
		update_mmu_cache(vma, address, page_table);
L
Linus Torvalds 已提交
3290
	} else {
3291 3292
		if (cow_page)
			mem_cgroup_uncharge_page(cow_page);
3293 3294 3295
		if (anon)
			page_cache_release(page);
		else
3296
			anon = 1; /* no anon but release faulted_page */
L
Linus Torvalds 已提交
3297 3298
	}

3299
	pte_unmap_unlock(page_table, ptl);
3300

N
Nick Piggin 已提交
3301 3302
	if (dirty_page) {
		struct address_space *mapping = page->mapping;
3303

N
Nick Piggin 已提交
3304 3305 3306
		if (set_page_dirty(dirty_page))
			page_mkwrite = 1;
		unlock_page(dirty_page);
3307
		put_page(dirty_page);
N
Nick Piggin 已提交
3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322
		if (page_mkwrite && mapping) {
			/*
			 * Some device drivers do not set page.mapping but still
			 * dirty their pages
			 */
			balance_dirty_pages_ratelimited(mapping);
		}

		/* file_update_time outside page_lock */
		if (vma->vm_file)
			file_update_time(vma->vm_file);
	} else {
		unlock_page(vmf.page);
		if (anon)
			page_cache_release(vmf.page);
3323
	}
3324

N
Nick Piggin 已提交
3325
	return ret;
N
Nick Piggin 已提交
3326 3327 3328 3329

unwritable_page:
	page_cache_release(page);
	return ret;
3330 3331 3332 3333 3334 3335 3336
uncharge_out:
	/* fs's fault handler get error */
	if (cow_page) {
		mem_cgroup_uncharge_page(cow_page);
		page_cache_release(cow_page);
	}
	return ret;
3337
}
3338

3339 3340
static int do_linear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
3341
		unsigned int flags, pte_t orig_pte)
3342 3343
{
	pgoff_t pgoff = (((address & PAGE_MASK)
3344
			- vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
3345

3346 3347
	pte_unmap(page_table);
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
3348 3349
}

L
Linus Torvalds 已提交
3350 3351 3352 3353
/*
 * Fault of a previously existing named mapping. Repopulate the pte
 * from the encoded file_pte if possible. This enables swappable
 * nonlinear vmas.
3354 3355 3356 3357
 *
 * 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
3358
 */
N
Nick Piggin 已提交
3359
static int do_nonlinear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3360
		unsigned long address, pte_t *page_table, pmd_t *pmd,
3361
		unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
3362
{
3363
	pgoff_t pgoff;
L
Linus Torvalds 已提交
3364

3365 3366
	flags |= FAULT_FLAG_NONLINEAR;

H
Hugh Dickins 已提交
3367
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
N
Nick Piggin 已提交
3368
		return 0;
L
Linus Torvalds 已提交
3369

3370
	if (unlikely(!(vma->vm_flags & VM_NONLINEAR))) {
3371 3372 3373
		/*
		 * Page table corrupted: show pte and kill process.
		 */
3374
		print_bad_pte(vma, address, orig_pte, NULL);
H
Hugh Dickins 已提交
3375
		return VM_FAULT_SIGBUS;
3376 3377 3378
	}

	pgoff = pte_to_pgoff(orig_pte);
3379
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
L
Linus Torvalds 已提交
3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390
}

/*
 * 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
3391 3392 3393
 * 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
3394
 */
3395 3396 3397
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
3398 3399
{
	pte_t entry;
3400
	spinlock_t *ptl;
L
Linus Torvalds 已提交
3401

3402
	entry = *pte;
L
Linus Torvalds 已提交
3403
	if (!pte_present(entry)) {
3404
		if (pte_none(entry)) {
J
Jes Sorensen 已提交
3405
			if (vma->vm_ops) {
N
Nick Piggin 已提交
3406
				if (likely(vma->vm_ops->fault))
3407
					return do_linear_fault(mm, vma, address,
3408
						pte, pmd, flags, entry);
J
Jes Sorensen 已提交
3409 3410
			}
			return do_anonymous_page(mm, vma, address,
3411
						 pte, pmd, flags);
3412
		}
L
Linus Torvalds 已提交
3413
		if (pte_file(entry))
N
Nick Piggin 已提交
3414
			return do_nonlinear_fault(mm, vma, address,
3415
					pte, pmd, flags, entry);
3416
		return do_swap_page(mm, vma, address,
3417
					pte, pmd, flags, entry);
L
Linus Torvalds 已提交
3418 3419
	}

H
Hugh Dickins 已提交
3420
	ptl = pte_lockptr(mm, pmd);
3421 3422 3423
	spin_lock(ptl);
	if (unlikely(!pte_same(*pte, entry)))
		goto unlock;
3424
	if (flags & FAULT_FLAG_WRITE) {
L
Linus Torvalds 已提交
3425
		if (!pte_write(entry))
3426 3427
			return do_wp_page(mm, vma, address,
					pte, pmd, ptl, entry);
L
Linus Torvalds 已提交
3428 3429 3430
		entry = pte_mkdirty(entry);
	}
	entry = pte_mkyoung(entry);
3431
	if (ptep_set_access_flags(vma, address, pte, entry, flags & FAULT_FLAG_WRITE)) {
3432
		update_mmu_cache(vma, address, pte);
3433 3434 3435 3436 3437 3438 3439
	} 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.
		 */
3440
		if (flags & FAULT_FLAG_WRITE)
3441
			flush_tlb_fix_spurious_fault(vma, address);
3442
	}
3443 3444
unlock:
	pte_unmap_unlock(pte, ptl);
N
Nick Piggin 已提交
3445
	return 0;
L
Linus Torvalds 已提交
3446 3447 3448 3449 3450
}

/*
 * By the time we get here, we already hold the mm semaphore
 */
N
Nick Piggin 已提交
3451
int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3452
		unsigned long address, unsigned int flags)
L
Linus Torvalds 已提交
3453 3454 3455 3456 3457 3458 3459 3460
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

	__set_current_state(TASK_RUNNING);

3461
	count_vm_event(PGFAULT);
3462
	mem_cgroup_count_vm_event(mm, PGFAULT);
L
Linus Torvalds 已提交
3463

3464 3465 3466
	/* do counter updates before entering really critical section. */
	check_sync_rss_stat(current);

3467
	if (unlikely(is_vm_hugetlb_page(vma)))
3468
		return hugetlb_fault(mm, vma, address, flags);
L
Linus Torvalds 已提交
3469 3470 3471 3472

	pgd = pgd_offset(mm, address);
	pud = pud_alloc(mm, pgd, address);
	if (!pud)
H
Hugh Dickins 已提交
3473
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
3474 3475
	pmd = pmd_alloc(mm, pud, address);
	if (!pmd)
H
Hugh Dickins 已提交
3476
		return VM_FAULT_OOM;
3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498
	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;
		barrier();
		if (pmd_trans_huge(orig_pmd)) {
			if (flags & FAULT_FLAG_WRITE &&
			    !pmd_write(orig_pmd) &&
			    !pmd_trans_splitting(orig_pmd))
				return do_huge_pmd_wp_page(mm, vma, address,
							   pmd, orig_pmd);
			return 0;
		}
	}

	/*
	 * 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.
	 */
3499
	if (unlikely(pmd_none(*pmd)) && __pte_alloc(mm, vma, pmd, address))
H
Hugh Dickins 已提交
3500
		return VM_FAULT_OOM;
3501 3502 3503 3504 3505 3506 3507 3508 3509 3510
	/* 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
3511

3512
	return handle_pte_fault(mm, vma, address, pte, pmd, flags);
L
Linus Torvalds 已提交
3513 3514 3515 3516 3517
}

#ifndef __PAGETABLE_PUD_FOLDED
/*
 * Allocate page upper directory.
H
Hugh Dickins 已提交
3518
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
3519
 */
3520
int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
L
Linus Torvalds 已提交
3521
{
H
Hugh Dickins 已提交
3522 3523
	pud_t *new = pud_alloc_one(mm, address);
	if (!new)
3524
		return -ENOMEM;
L
Linus Torvalds 已提交
3525

3526 3527
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
3528
	spin_lock(&mm->page_table_lock);
3529
	if (pgd_present(*pgd))		/* Another has populated it */
3530
		pud_free(mm, new);
3531 3532
	else
		pgd_populate(mm, pgd, new);
H
Hugh Dickins 已提交
3533
	spin_unlock(&mm->page_table_lock);
3534
	return 0;
L
Linus Torvalds 已提交
3535 3536 3537 3538 3539 3540
}
#endif /* __PAGETABLE_PUD_FOLDED */

#ifndef __PAGETABLE_PMD_FOLDED
/*
 * Allocate page middle directory.
H
Hugh Dickins 已提交
3541
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
3542
 */
3543
int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
L
Linus Torvalds 已提交
3544
{
H
Hugh Dickins 已提交
3545 3546
	pmd_t *new = pmd_alloc_one(mm, address);
	if (!new)
3547
		return -ENOMEM;
L
Linus Torvalds 已提交
3548

3549 3550
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
3551
	spin_lock(&mm->page_table_lock);
L
Linus Torvalds 已提交
3552
#ifndef __ARCH_HAS_4LEVEL_HACK
3553
	if (pud_present(*pud))		/* Another has populated it */
3554
		pmd_free(mm, new);
3555 3556
	else
		pud_populate(mm, pud, new);
L
Linus Torvalds 已提交
3557
#else
3558
	if (pgd_present(*pud))		/* Another has populated it */
3559
		pmd_free(mm, new);
3560 3561
	else
		pgd_populate(mm, pud, new);
L
Linus Torvalds 已提交
3562
#endif /* __ARCH_HAS_4LEVEL_HACK */
H
Hugh Dickins 已提交
3563
	spin_unlock(&mm->page_table_lock);
3564
	return 0;
3565
}
L
Linus Torvalds 已提交
3566 3567 3568 3569 3570 3571 3572 3573 3574
#endif /* __PAGETABLE_PMD_FOLDED */

int make_pages_present(unsigned long addr, unsigned long end)
{
	int ret, len, write;
	struct vm_area_struct * vma;

	vma = find_vma(current->mm, addr);
	if (!vma)
K
KOSAKI Motohiro 已提交
3575
		return -ENOMEM;
3576 3577 3578 3579 3580 3581
	/*
	 * We want to touch writable mappings with a write fault in order
	 * to break COW, except for shared mappings because these don't COW
	 * and we would not want to dirty them for nothing.
	 */
	write = (vma->vm_flags & (VM_WRITE | VM_SHARED)) == VM_WRITE;
3582 3583
	BUG_ON(addr >= end);
	BUG_ON(end > vma->vm_end);
3584
	len = DIV_ROUND_UP(end, PAGE_SIZE) - addr/PAGE_SIZE;
L
Linus Torvalds 已提交
3585 3586
	ret = get_user_pages(current, current->mm, addr,
			len, write, 0, NULL, NULL);
3587
	if (ret < 0)
L
Linus Torvalds 已提交
3588
		return ret;
3589
	return ret == len ? 0 : -EFAULT;
L
Linus Torvalds 已提交
3590 3591 3592 3593 3594
}

#if !defined(__HAVE_ARCH_GATE_AREA)

#if defined(AT_SYSINFO_EHDR)
3595
static struct vm_area_struct gate_vma;
L
Linus Torvalds 已提交
3596 3597 3598 3599 3600 3601

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
3602 3603
	gate_vma.vm_flags = VM_READ | VM_MAYREAD | VM_EXEC | VM_MAYEXEC;
	gate_vma.vm_page_prot = __P101;
3604 3605 3606 3607 3608 3609 3610
	/*
	 * Make sure the vDSO gets into every core dump.
	 * Dumping its contents makes post-mortem fully interpretable later
	 * without matching up the same kernel and hardware config to see
	 * what PC values meant.
	 */
	gate_vma.vm_flags |= VM_ALWAYSDUMP;
L
Linus Torvalds 已提交
3611 3612 3613 3614 3615
	return 0;
}
__initcall(gate_vma_init);
#endif

3616
struct vm_area_struct *get_gate_vma(struct mm_struct *mm)
L
Linus Torvalds 已提交
3617 3618 3619 3620 3621 3622 3623 3624
{
#ifdef AT_SYSINFO_EHDR
	return &gate_vma;
#else
	return NULL;
#endif
}

3625
int in_gate_area_no_mm(unsigned long addr)
L
Linus Torvalds 已提交
3626 3627 3628 3629 3630 3631 3632 3633 3634
{
#ifdef AT_SYSINFO_EHDR
	if ((addr >= FIXADDR_USER_START) && (addr < FIXADDR_USER_END))
		return 1;
#endif
	return 0;
}

#endif	/* __HAVE_ARCH_GATE_AREA */
3635

3636
static int __follow_pte(struct mm_struct *mm, unsigned long address,
J
Johannes Weiner 已提交
3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652
		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);
3653
	VM_BUG_ON(pmd_trans_huge(*pmd));
J
Johannes Weiner 已提交
3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673
	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;
}

3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684
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
3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713
/**
 * 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);

3714
#ifdef CONFIG_HAVE_IOREMAP_PROT
3715 3716 3717
int follow_phys(struct vm_area_struct *vma,
		unsigned long address, unsigned int flags,
		unsigned long *prot, resource_size_t *phys)
3718
{
3719
	int ret = -EINVAL;
3720 3721 3722
	pte_t *ptep, pte;
	spinlock_t *ptl;

3723 3724
	if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
		goto out;
3725

3726
	if (follow_pte(vma->vm_mm, address, &ptep, &ptl))
3727
		goto out;
3728
	pte = *ptep;
3729

3730 3731 3732 3733
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;

	*prot = pgprot_val(pte_pgprot(pte));
3734
	*phys = (resource_size_t)pte_pfn(pte) << PAGE_SHIFT;
3735

3736
	ret = 0;
3737 3738 3739
unlock:
	pte_unmap_unlock(ptep, ptl);
out:
3740
	return ret;
3741 3742 3743 3744 3745 3746 3747
}

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
3748
	void __iomem *maddr;
3749 3750
	int offset = addr & (PAGE_SIZE-1);

3751
	if (follow_phys(vma, addr, write, &prot, &phys_addr))
3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764
		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

3765
/*
3766 3767
 * Access another process' address space as given in mm.  If non-NULL, use the
 * given task for page fault accounting.
3768
 */
3769 3770
static int __access_remote_vm(struct task_struct *tsk, struct mm_struct *mm,
		unsigned long addr, void *buf, int len, int write)
3771 3772 3773 3774 3775
{
	struct vm_area_struct *vma;
	void *old_buf = buf;

	down_read(&mm->mmap_sem);
S
Simon Arlott 已提交
3776
	/* ignore errors, just check how much was successfully transferred */
3777 3778 3779
	while (len) {
		int bytes, ret, offset;
		void *maddr;
3780
		struct page *page = NULL;
3781 3782 3783

		ret = get_user_pages(tsk, mm, addr, 1,
				write, 1, &page, &vma);
3784 3785 3786 3787 3788 3789 3790
		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);
3791
			if (!vma || vma->vm_start > addr)
3792 3793 3794 3795 3796 3797 3798 3799
				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;
3800
		} else {
3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816
			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);
3817 3818 3819 3820 3821 3822 3823 3824 3825
		}
		len -= bytes;
		buf += bytes;
		addr += bytes;
	}
	up_read(&mm->mmap_sem);

	return buf - old_buf;
}
3826

S
Stephen Wilson 已提交
3827
/**
3828
 * access_remote_vm - access another process' address space
S
Stephen Wilson 已提交
3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842
 * @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);
}

3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863
/*
 * 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;
}

3864 3865 3866 3867 3868 3869 3870 3871
/*
 * 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;

3872 3873 3874 3875 3876 3877 3878
	/*
	 * Do not print if we are in atomic
	 * contexts (in exception stacks, etc.):
	 */
	if (preempt_count())
		return;

3879 3880 3881 3882 3883 3884 3885 3886
	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) {
			char *p, *s;

3887
			p = d_path(&f->f_path, buf, PAGE_SIZE);
3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900
			if (IS_ERR(p))
				p = "?";
			s = strrchr(p, '/');
			if (s)
				p = s+1;
			printk("%s%s[%lx+%lx]", prefix, p,
					vma->vm_start,
					vma->vm_end - vma->vm_start);
			free_page((unsigned long)buf);
		}
	}
	up_read(&current->mm->mmap_sem);
}
3901 3902 3903 3904

#ifdef CONFIG_PROVE_LOCKING
void might_fault(void)
{
3905 3906 3907 3908 3909 3910 3911 3912 3913
	/*
	 * 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;

3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924
	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
3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995

#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 */