memory.c 106.5 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>
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>
53
#include <linux/pfn_t.h>
54
#include <linux/writeback.h>
55
#include <linux/memcontrol.h>
A
Andrea Arcangeli 已提交
56
#include <linux/mmu_notifier.h>
57 58 59
#include <linux/kallsyms.h>
#include <linux/swapops.h>
#include <linux/elf.h>
60
#include <linux/gfp.h>
61
#include <linux/migrate.h>
A
Andy Shevchenko 已提交
62
#include <linux/string.h>
63
#include <linux/dma-debug.h>
64
#include <linux/debugfs.h>
65
#include <linux/userfaultfd_k.h>
66
#include <linux/dax.h>
L
Linus Torvalds 已提交
67

68
#include <asm/io.h>
69
#include <asm/mmu_context.h>
L
Linus Torvalds 已提交
70 71 72 73 74 75
#include <asm/pgalloc.h>
#include <asm/uaccess.h>
#include <asm/tlb.h>
#include <asm/tlbflush.h>
#include <asm/pgtable.h>

76 77
#include "internal.h"

78 79
#ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
#warning Unfortunate NUMA and NUMA Balancing config, growing page-frame for last_cpupid.
80 81
#endif

82
#ifndef CONFIG_NEED_MULTIPLE_NODES
L
Linus Torvalds 已提交
83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
/* 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

/*
 * 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(high_memory);

102 103 104 105 106 107 108 109 110 111 112 113
/*
 * 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
114 115 116 117

static int __init disable_randmaps(char *s)
{
	randomize_va_space = 0;
118
	return 1;
119 120 121
}
__setup("norandmaps", disable_randmaps);

122
unsigned long zero_pfn __read_mostly;
123
unsigned long highest_memmap_pfn __read_mostly;
H
Hugh Dickins 已提交
124

125 126
EXPORT_SYMBOL(zero_pfn);

H
Hugh Dickins 已提交
127 128 129 130 131 132 133 134 135
/*
 * 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);
136

137

138 139
#if defined(SPLIT_RSS_COUNTING)

140
void sync_mm_rss(struct mm_struct *mm)
141 142 143 144
{
	int i;

	for (i = 0; i < NR_MM_COUNTERS; i++) {
145 146 147
		if (current->rss_stat.count[i]) {
			add_mm_counter(mm, i, current->rss_stat.count[i]);
			current->rss_stat.count[i] = 0;
148 149
		}
	}
150
	current->rss_stat.events = 0;
151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
}

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))
172
		sync_mm_rss(task->mm);
173
}
174
#else /* SPLIT_RSS_COUNTING */
175 176 177 178 179 180 181 182

#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)
{
}

183 184 185 186
#endif /* SPLIT_RSS_COUNTING */

#ifdef HAVE_GENERIC_MMU_GATHER

187
static bool tlb_next_batch(struct mmu_gather *tlb)
188 189 190 191 192 193
{
	struct mmu_gather_batch *batch;

	batch = tlb->active;
	if (batch->next) {
		tlb->active = batch->next;
194
		return true;
195 196
	}

197
	if (tlb->batch_count == MAX_GATHER_BATCH_COUNT)
198
		return false;
199

200 201
	batch = (void *)__get_free_pages(GFP_NOWAIT | __GFP_NOWARN, 0);
	if (!batch)
202
		return false;
203

204
	tlb->batch_count++;
205 206 207 208 209 210 211
	batch->next = NULL;
	batch->nr   = 0;
	batch->max  = MAX_GATHER_BATCH;

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

212
	return true;
213 214 215 216 217 218 219
}

/* 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).
 */
220
void tlb_gather_mmu(struct mmu_gather *tlb, struct mm_struct *mm, unsigned long start, unsigned long end)
221 222 223
{
	tlb->mm = mm;

224 225
	/* Is it from 0 to ~0? */
	tlb->fullmm     = !(start | (end+1));
226
	tlb->need_flush_all = 0;
227 228 229 230
	tlb->local.next = NULL;
	tlb->local.nr   = 0;
	tlb->local.max  = ARRAY_SIZE(tlb->__pages);
	tlb->active     = &tlb->local;
231
	tlb->batch_count = 0;
232 233 234 235

#ifdef CONFIG_HAVE_RCU_TABLE_FREE
	tlb->batch = NULL;
#endif
236 237

	__tlb_reset_range(tlb);
238 239
}

240
static void tlb_flush_mmu_tlbonly(struct mmu_gather *tlb)
241
{
242 243 244
	if (!tlb->end)
		return;

245
	tlb_flush(tlb);
246
	mmu_notifier_invalidate_range(tlb->mm, tlb->start, tlb->end);
247 248
#ifdef CONFIG_HAVE_RCU_TABLE_FREE
	tlb_table_flush(tlb);
249
#endif
250
	__tlb_reset_range(tlb);
251 252 253 254 255
}

static void tlb_flush_mmu_free(struct mmu_gather *tlb)
{
	struct mmu_gather_batch *batch;
256

257
	for (batch = &tlb->local; batch && batch->nr; batch = batch->next) {
258 259 260 261 262 263
		free_pages_and_swap_cache(batch->pages, batch->nr);
		batch->nr = 0;
	}
	tlb->active = &tlb->local;
}

264 265 266 267 268 269
void tlb_flush_mmu(struct mmu_gather *tlb)
{
	tlb_flush_mmu_tlbonly(tlb);
	tlb_flush_mmu_free(tlb);
}

270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299
/* 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;

300
	VM_BUG_ON(!tlb->end);
301 302 303 304 305 306

	batch = tlb->active;
	batch->pages[batch->nr++] = page;
	if (batch->nr == batch->max) {
		if (!tlb_next_batch(tlb))
			return 0;
307
		batch = tlb->active;
308
	}
309
	VM_BUG_ON_PAGE(batch->nr > batch->max, page);
310 311 312 313 314 315

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

#endif /* HAVE_GENERIC_MMU_GATHER */

316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388
#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;

	/*
	 * 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);
}

389
#endif /* CONFIG_HAVE_RCU_TABLE_FREE */
390

L
Linus Torvalds 已提交
391 392 393 394
/*
 * Note: this doesn't free the actual pages themselves. That
 * has been handled earlier when unmapping all the memory regions.
 */
395 396
static void free_pte_range(struct mmu_gather *tlb, pmd_t *pmd,
			   unsigned long addr)
L
Linus Torvalds 已提交
397
{
398
	pgtable_t token = pmd_pgtable(*pmd);
399
	pmd_clear(pmd);
400
	pte_free_tlb(tlb, token, addr);
401
	atomic_long_dec(&tlb->mm->nr_ptes);
L
Linus Torvalds 已提交
402 403
}

404 405 406
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
407 408 409
{
	pmd_t *pmd;
	unsigned long next;
410
	unsigned long start;
L
Linus Torvalds 已提交
411

412
	start = addr;
L
Linus Torvalds 已提交
413 414 415 416 417
	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
418
		free_pte_range(tlb, pmd, addr);
L
Linus Torvalds 已提交
419 420
	} while (pmd++, addr = next, addr != end);

421 422 423 424 425 426 427
	start &= PUD_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PUD_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
428
	}
429 430 431 432 433
	if (end - 1 > ceiling - 1)
		return;

	pmd = pmd_offset(pud, start);
	pud_clear(pud);
434
	pmd_free_tlb(tlb, pmd, start);
435
	mm_dec_nr_pmds(tlb->mm);
L
Linus Torvalds 已提交
436 437
}

438 439 440
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
441 442 443
{
	pud_t *pud;
	unsigned long next;
444
	unsigned long start;
L
Linus Torvalds 已提交
445

446
	start = addr;
L
Linus Torvalds 已提交
447 448 449 450 451
	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
452
		free_pmd_range(tlb, pud, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
453 454
	} while (pud++, addr = next, addr != end);

455 456 457 458 459 460 461
	start &= PGDIR_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PGDIR_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
462
	}
463 464 465 466 467
	if (end - 1 > ceiling - 1)
		return;

	pud = pud_offset(pgd, start);
	pgd_clear(pgd);
468
	pud_free_tlb(tlb, pud, start);
L
Linus Torvalds 已提交
469 470 471
}

/*
472
 * This function frees user-level page tables of a process.
L
Linus Torvalds 已提交
473
 */
474
void free_pgd_range(struct mmu_gather *tlb,
475 476
			unsigned long addr, unsigned long end,
			unsigned long floor, unsigned long ceiling)
L
Linus Torvalds 已提交
477 478 479
{
	pgd_t *pgd;
	unsigned long next;
480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505

	/*
	 * 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
506

507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522
	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;

523
	pgd = pgd_offset(tlb->mm, addr);
L
Linus Torvalds 已提交
524 525 526 527
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
528
		free_pud_range(tlb, pgd, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
529
	} while (pgd++, addr = next, addr != end);
530 531
}

532
void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *vma,
533
		unsigned long floor, unsigned long ceiling)
534 535 536 537 538
{
	while (vma) {
		struct vm_area_struct *next = vma->vm_next;
		unsigned long addr = vma->vm_start;

539
		/*
N
npiggin@suse.de 已提交
540 541
		 * Hide vma from rmap and truncate_pagecache before freeing
		 * pgtables
542
		 */
543
		unlink_anon_vmas(vma);
544 545
		unlink_file_vma(vma);

546
		if (is_vm_hugetlb_page(vma)) {
547
			hugetlb_free_pgd_range(tlb, addr, vma->vm_end,
548
				floor, next? next->vm_start: ceiling);
549 550 551 552 553
		} else {
			/*
			 * Optimization: gather nearby vmas into one call down
			 */
			while (next && next->vm_start <= vma->vm_end + PMD_SIZE
554
			       && !is_vm_hugetlb_page(next)) {
555 556
				vma = next;
				next = vma->vm_next;
557
				unlink_anon_vmas(vma);
558
				unlink_file_vma(vma);
559 560 561 562
			}
			free_pgd_range(tlb, addr, vma->vm_end,
				floor, next? next->vm_start: ceiling);
		}
563 564
		vma = next;
	}
L
Linus Torvalds 已提交
565 566
}

567
int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
568
{
569
	spinlock_t *ptl;
570
	pgtable_t new = pte_alloc_one(mm, address);
571 572 573
	if (!new)
		return -ENOMEM;

574 575 576 577 578 579 580 581 582 583 584 585 586 587 588
	/*
	 * 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 */

589
	ptl = pmd_lock(mm, pmd);
590
	if (likely(pmd_none(*pmd))) {	/* Has another populated it ? */
591
		atomic_long_inc(&mm->nr_ptes);
L
Linus Torvalds 已提交
592
		pmd_populate(mm, pmd, new);
593
		new = NULL;
594
	}
595
	spin_unlock(ptl);
596 597
	if (new)
		pte_free(mm, new);
598
	return 0;
L
Linus Torvalds 已提交
599 600
}

601
int __pte_alloc_kernel(pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
602
{
603 604 605 606
	pte_t *new = pte_alloc_one_kernel(&init_mm, address);
	if (!new)
		return -ENOMEM;

607 608
	smp_wmb(); /* See comment in __pte_alloc */

609
	spin_lock(&init_mm.page_table_lock);
610
	if (likely(pmd_none(*pmd))) {	/* Has another populated it ? */
611
		pmd_populate_kernel(&init_mm, pmd, new);
612
		new = NULL;
613
	}
614
	spin_unlock(&init_mm.page_table_lock);
615 616
	if (new)
		pte_free_kernel(&init_mm, new);
617
	return 0;
L
Linus Torvalds 已提交
618 619
}

620 621 622 623 624 625
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)
626
{
627 628
	int i;

629
	if (current->mm == mm)
630
		sync_mm_rss(mm);
631 632 633
	for (i = 0; i < NR_MM_COUNTERS; i++)
		if (rss[i])
			add_mm_counter(mm, i, rss[i]);
634 635
}

636
/*
637 638 639
 * 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.
640 641 642
 *
 * The calling function must still handle the error.
 */
643 644
static void print_bad_pte(struct vm_area_struct *vma, unsigned long addr,
			  pte_t pte, struct page *page)
645
{
646 647 648 649 650
	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;
651 652 653 654 655 656 657 658 659 660 661 662 663 664
	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) {
665 666
			pr_alert("BUG: Bad page map: %lu messages suppressed\n",
				 nr_unshown);
667 668 669 670 671 672
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;
673 674 675 676

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

677 678 679
	pr_alert("BUG: Bad page map in process %s  pte:%08llx pmd:%08llx\n",
		 current->comm,
		 (long long)pte_val(pte), (long long)pmd_val(*pmd));
680
	if (page)
681
		dump_page(page, "bad pte");
682 683
	pr_alert("addr:%p vm_flags:%08lx anon_vma:%p mapping:%p index:%lx\n",
		 (void *)addr, vma->vm_flags, vma->anon_vma, mapping, index);
684 685 686
	/*
	 * Choose text because data symbols depend on CONFIG_KALLSYMS_ALL=y
	 */
687 688 689 690 691
	pr_alert("file:%pD fault:%pf mmap:%pf readpage:%pf\n",
		 vma->vm_file,
		 vma->vm_ops ? vma->vm_ops->fault : NULL,
		 vma->vm_file ? vma->vm_file->f_op->mmap : NULL,
		 mapping ? mapping->a_ops->readpage : NULL);
692
	dump_stack();
693
	add_taint(TAINT_BAD_PAGE, LOCKDEP_NOW_UNRELIABLE);
694 695
}

696
/*
697
 * vm_normal_page -- This function gets the "struct page" associated with a pte.
698
 *
699 700 701
 * "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
702
 *
703 704 705 706 707 708 709 710
 * 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.
711
 *
J
Jared Hulbert 已提交
712 713
 * 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
714 715
 * set, and the vm_pgoff will point to the first PFN mapped: thus every special
 * mapping will always honor the rule
716 717 718
 *
 *	pfn_of_page == vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT)
 *
719 720 721 722 723 724
 * 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
725 726
 *
 *
727
 * In order to support COW of arbitrary special mappings, we have VM_MIXEDMAP.
J
Jared Hulbert 已提交
728 729 730 731 732 733 734 735 736
 *
 * 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.
 *
737
 */
738 739 740 741 742 743 744
#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)
745
{
746
	unsigned long pfn = pte_pfn(pte);
747 748

	if (HAVE_PTE_SPECIAL) {
749
		if (likely(!pte_special(pte)))
750
			goto check_pfn;
751 752
		if (vma->vm_ops && vma->vm_ops->find_special_page)
			return vma->vm_ops->find_special_page(vma, addr);
H
Hugh Dickins 已提交
753 754
		if (vma->vm_flags & (VM_PFNMAP | VM_MIXEDMAP))
			return NULL;
755
		if (!is_zero_pfn(pfn))
756
			print_bad_pte(vma, addr, pte, NULL);
757 758 759 760 761
		return NULL;
	}

	/* !HAVE_PTE_SPECIAL case follows: */

J
Jared Hulbert 已提交
762 763 764 765 766 767
	if (unlikely(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP))) {
		if (vma->vm_flags & VM_MIXEDMAP) {
			if (!pfn_valid(pfn))
				return NULL;
			goto out;
		} else {
768 769
			unsigned long off;
			off = (addr - vma->vm_start) >> PAGE_SHIFT;
J
Jared Hulbert 已提交
770 771 772 773 774
			if (pfn == vma->vm_pgoff + off)
				return NULL;
			if (!is_cow_mapping(vma->vm_flags))
				return NULL;
		}
775 776
	}

777 778
	if (is_zero_pfn(pfn))
		return NULL;
779 780 781 782 783
check_pfn:
	if (unlikely(pfn > highest_memmap_pfn)) {
		print_bad_pte(vma, addr, pte, NULL);
		return NULL;
	}
784 785

	/*
786 787
	 * NOTE! We still have PageReserved() pages in the page tables.
	 * eg. VDSO mappings can cause them to exist.
788
	 */
J
Jared Hulbert 已提交
789
out:
790
	return pfn_to_page(pfn);
791 792
}

L
Linus Torvalds 已提交
793 794 795 796 797 798
/*
 * 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.
 */

799
static inline unsigned long
L
Linus Torvalds 已提交
800
copy_one_pte(struct mm_struct *dst_mm, struct mm_struct *src_mm,
801
		pte_t *dst_pte, pte_t *src_pte, struct vm_area_struct *vma,
802
		unsigned long addr, int *rss)
L
Linus Torvalds 已提交
803
{
804
	unsigned long vm_flags = vma->vm_flags;
L
Linus Torvalds 已提交
805 806 807 808 809
	pte_t pte = *src_pte;
	struct page *page;

	/* pte contains position in swap or file, so copy. */
	if (unlikely(!pte_present(pte))) {
810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827
		swp_entry_t entry = pte_to_swp_entry(pte);

		if (likely(!non_swap_entry(entry))) {
			if (swap_duplicate(entry) < 0)
				return entry.val;

			/* make sure dst_mm is on swapoff's mmlist. */
			if (unlikely(list_empty(&dst_mm->mmlist))) {
				spin_lock(&mmlist_lock);
				if (list_empty(&dst_mm->mmlist))
					list_add(&dst_mm->mmlist,
							&src_mm->mmlist);
				spin_unlock(&mmlist_lock);
			}
			rss[MM_SWAPENTS]++;
		} else if (is_migration_entry(entry)) {
			page = migration_entry_to_page(entry);

828
			rss[mm_counter(page)]++;
829 830 831 832 833 834 835 836 837 838 839 840

			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);
				if (pte_swp_soft_dirty(*src_pte))
					pte = pte_swp_mksoft_dirty(pte);
				set_pte_at(src_mm, addr, src_pte, pte);
841
			}
L
Linus Torvalds 已提交
842
		}
843
		goto out_set_pte;
L
Linus Torvalds 已提交
844 845 846 847 848 849
	}

	/*
	 * If it's a COW mapping, write protect it both
	 * in the parent and the child
	 */
850
	if (is_cow_mapping(vm_flags)) {
L
Linus Torvalds 已提交
851
		ptep_set_wrprotect(src_mm, addr, src_pte);
852
		pte = pte_wrprotect(pte);
L
Linus Torvalds 已提交
853 854 855 856 857 858 859 860 861
	}

	/*
	 * 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);
862 863 864 865

	page = vm_normal_page(vma, addr, pte);
	if (page) {
		get_page(page);
866
		page_dup_rmap(page, false);
867
		rss[mm_counter(page)]++;
868
	}
869 870 871

out_set_pte:
	set_pte_at(dst_mm, addr, dst_pte, pte);
872
	return 0;
L
Linus Torvalds 已提交
873 874
}

875
static int copy_pte_range(struct mm_struct *dst_mm, struct mm_struct *src_mm,
876 877
		   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
878
{
879
	pte_t *orig_src_pte, *orig_dst_pte;
L
Linus Torvalds 已提交
880
	pte_t *src_pte, *dst_pte;
881
	spinlock_t *src_ptl, *dst_ptl;
882
	int progress = 0;
883
	int rss[NR_MM_COUNTERS];
884
	swp_entry_t entry = (swp_entry_t){0};
L
Linus Torvalds 已提交
885 886

again:
887 888
	init_rss_vec(rss);

889
	dst_pte = pte_alloc_map_lock(dst_mm, dst_pmd, addr, &dst_ptl);
L
Linus Torvalds 已提交
890 891
	if (!dst_pte)
		return -ENOMEM;
892
	src_pte = pte_offset_map(src_pmd, addr);
893
	src_ptl = pte_lockptr(src_mm, src_pmd);
894
	spin_lock_nested(src_ptl, SINGLE_DEPTH_NESTING);
895 896
	orig_src_pte = src_pte;
	orig_dst_pte = dst_pte;
897
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
898 899 900 901 902 903

	do {
		/*
		 * We are holding two locks at this point - either of them
		 * could generate latencies in another task on another CPU.
		 */
904 905 906
		if (progress >= 32) {
			progress = 0;
			if (need_resched() ||
N
Nick Piggin 已提交
907
			    spin_needbreak(src_ptl) || spin_needbreak(dst_ptl))
908 909
				break;
		}
L
Linus Torvalds 已提交
910 911 912 913
		if (pte_none(*src_pte)) {
			progress++;
			continue;
		}
914 915 916 917
		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
918 919 920
		progress += 8;
	} while (dst_pte++, src_pte++, addr += PAGE_SIZE, addr != end);

921
	arch_leave_lazy_mmu_mode();
922
	spin_unlock(src_ptl);
923
	pte_unmap(orig_src_pte);
924
	add_mm_rss_vec(dst_mm, rss);
925
	pte_unmap_unlock(orig_dst_pte, dst_ptl);
926
	cond_resched();
927 928 929 930 931 932

	if (entry.val) {
		if (add_swap_count_continuation(entry, GFP_KERNEL) < 0)
			return -ENOMEM;
		progress = 0;
	}
L
Linus Torvalds 已提交
933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950
	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);
951
		if (pmd_trans_huge(*src_pmd) || pmd_devmap(*src_pmd)) {
952
			int err;
953
			VM_BUG_ON(next-addr != HPAGE_PMD_SIZE);
954 955 956 957 958 959 960 961
			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
962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999
		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;
1000 1001 1002
	unsigned long mmun_start;	/* For mmu_notifiers */
	unsigned long mmun_end;		/* For mmu_notifiers */
	bool is_cow;
A
Andrea Arcangeli 已提交
1003
	int ret;
L
Linus Torvalds 已提交
1004

1005 1006 1007 1008 1009 1010
	/*
	 * 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.
	 */
1011 1012 1013
	if (!(vma->vm_flags & (VM_HUGETLB | VM_PFNMAP | VM_MIXEDMAP)) &&
			!vma->anon_vma)
		return 0;
1014

L
Linus Torvalds 已提交
1015 1016 1017
	if (is_vm_hugetlb_page(vma))
		return copy_hugetlb_page_range(dst_mm, src_mm, vma);

1018
	if (unlikely(vma->vm_flags & VM_PFNMAP)) {
1019 1020 1021 1022
		/*
		 * We do not free on error cases below as remove_vma
		 * gets called on error from higher level routine
		 */
1023
		ret = track_pfn_copy(vma);
1024 1025 1026 1027
		if (ret)
			return ret;
	}

A
Andrea Arcangeli 已提交
1028 1029 1030 1031 1032 1033
	/*
	 * 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.
	 */
1034 1035 1036 1037 1038 1039
	is_cow = is_cow_mapping(vma->vm_flags);
	mmun_start = addr;
	mmun_end   = end;
	if (is_cow)
		mmu_notifier_invalidate_range_start(src_mm, mmun_start,
						    mmun_end);
A
Andrea Arcangeli 已提交
1040 1041

	ret = 0;
L
Linus Torvalds 已提交
1042 1043 1044 1045 1046 1047
	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
1048 1049 1050 1051 1052
		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
1053
	} while (dst_pgd++, src_pgd++, addr = next, addr != end);
A
Andrea Arcangeli 已提交
1054

1055 1056
	if (is_cow)
		mmu_notifier_invalidate_range_end(src_mm, mmun_start, mmun_end);
A
Andrea Arcangeli 已提交
1057
	return ret;
L
Linus Torvalds 已提交
1058 1059
}

1060
static unsigned long zap_pte_range(struct mmu_gather *tlb,
1061
				struct vm_area_struct *vma, pmd_t *pmd,
L
Linus Torvalds 已提交
1062
				unsigned long addr, unsigned long end,
1063
				struct zap_details *details)
L
Linus Torvalds 已提交
1064
{
1065
	struct mm_struct *mm = tlb->mm;
P
Peter Zijlstra 已提交
1066
	int force_flush = 0;
1067
	int rss[NR_MM_COUNTERS];
1068
	spinlock_t *ptl;
1069
	pte_t *start_pte;
1070
	pte_t *pte;
1071
	swp_entry_t entry;
1072

P
Peter Zijlstra 已提交
1073
again:
1074
	init_rss_vec(rss);
1075 1076
	start_pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
	pte = start_pte;
1077
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
1078 1079
	do {
		pte_t ptent = *pte;
1080
		if (pte_none(ptent)) {
L
Linus Torvalds 已提交
1081
			continue;
1082
		}
1083

L
Linus Torvalds 已提交
1084
		if (pte_present(ptent)) {
1085
			struct page *page;
1086

1087
			page = vm_normal_page(vma, addr, ptent);
L
Linus Torvalds 已提交
1088 1089 1090 1091 1092 1093 1094 1095 1096 1097
			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;
			}
1098
			ptent = ptep_get_and_clear_full(mm, addr, pte,
1099
							tlb->fullmm);
L
Linus Torvalds 已提交
1100 1101 1102
			tlb_remove_tlb_entry(tlb, pte, addr);
			if (unlikely(!page))
				continue;
1103 1104

			if (!PageAnon(page)) {
1105
				if (pte_dirty(ptent)) {
1106 1107 1108 1109 1110 1111
					/*
					 * oom_reaper cannot tear down dirty
					 * pages
					 */
					if (unlikely(details && details->ignore_dirty))
						continue;
1112
					force_flush = 1;
1113
					set_page_dirty(page);
1114
				}
1115
				if (pte_young(ptent) &&
1116
				    likely(!(vma->vm_flags & VM_SEQ_READ)))
1117
					mark_page_accessed(page);
1118
			}
1119
			rss[mm_counter(page)]--;
1120
			page_remove_rmap(page, false);
1121 1122
			if (unlikely(page_mapcount(page) < 0))
				print_bad_pte(vma, addr, ptent, page);
1123 1124
			if (unlikely(!__tlb_remove_page(tlb, page))) {
				force_flush = 1;
1125
				addr += PAGE_SIZE;
P
Peter Zijlstra 已提交
1126
				break;
1127
			}
L
Linus Torvalds 已提交
1128 1129
			continue;
		}
1130 1131
		/* only check swap_entries if explicitly asked for in details */
		if (unlikely(details && !details->check_swap_entries))
L
Linus Torvalds 已提交
1132
			continue;
K
KAMEZAWA Hiroyuki 已提交
1133

1134 1135 1136 1137 1138
		entry = pte_to_swp_entry(ptent);
		if (!non_swap_entry(entry))
			rss[MM_SWAPENTS]--;
		else if (is_migration_entry(entry)) {
			struct page *page;
1139

1140
			page = migration_entry_to_page(entry);
1141
			rss[mm_counter(page)]--;
K
KAMEZAWA Hiroyuki 已提交
1142
		}
1143 1144
		if (unlikely(!free_swap_and_cache(entry)))
			print_bad_pte(vma, addr, ptent, NULL);
1145
		pte_clear_not_present_full(mm, addr, pte, tlb->fullmm);
1146
	} while (pte++, addr += PAGE_SIZE, addr != end);
1147

1148
	add_mm_rss_vec(mm, rss);
1149
	arch_leave_lazy_mmu_mode();
1150

1151
	/* Do the actual TLB flush before dropping ptl */
1152
	if (force_flush)
1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164
		tlb_flush_mmu_tlbonly(tlb);
	pte_unmap_unlock(start_pte, ptl);

	/*
	 * If we forced a TLB flush (either due to running out of
	 * batch buffers or because we needed to flush dirty TLB
	 * entries before releasing the ptl), free the batched
	 * memory too. Restart if we didn't do everything.
	 */
	if (force_flush) {
		force_flush = 0;
		tlb_flush_mmu_free(tlb);
1165 1166

		if (addr != end)
P
Peter Zijlstra 已提交
1167 1168 1169
			goto again;
	}

1170
	return addr;
L
Linus Torvalds 已提交
1171 1172
}

1173
static inline unsigned long zap_pmd_range(struct mmu_gather *tlb,
1174
				struct vm_area_struct *vma, pud_t *pud,
L
Linus Torvalds 已提交
1175
				unsigned long addr, unsigned long end,
1176
				struct zap_details *details)
L
Linus Torvalds 已提交
1177 1178 1179 1180 1181 1182 1183
{
	pmd_t *pmd;
	unsigned long next;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
1184
		if (pmd_trans_huge(*pmd) || pmd_devmap(*pmd)) {
1185
			if (next - addr != HPAGE_PMD_SIZE) {
1186 1187 1188 1189 1190 1191 1192 1193 1194
#ifdef CONFIG_DEBUG_VM
				if (!rwsem_is_locked(&tlb->mm->mmap_sem)) {
					pr_err("%s: mmap_sem is unlocked! addr=0x%lx end=0x%lx vma->vm_start=0x%lx vma->vm_end=0x%lx\n",
						__func__, addr, end,
						vma->vm_start,
						vma->vm_end);
					BUG();
				}
#endif
1195
				split_huge_pmd(vma, pmd, addr);
1196
			} else if (zap_huge_pmd(tlb, vma, pmd, addr))
1197
				goto next;
1198 1199
			/* fall through */
		}
1200 1201 1202 1203 1204 1205 1206 1207 1208
		/*
		 * 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;
1209
		next = zap_pte_range(tlb, vma, pmd, addr, next, details);
1210
next:
1211 1212
		cond_resched();
	} while (pmd++, addr = next, addr != end);
1213 1214

	return addr;
L
Linus Torvalds 已提交
1215 1216
}

1217
static inline unsigned long zap_pud_range(struct mmu_gather *tlb,
1218
				struct vm_area_struct *vma, pgd_t *pgd,
L
Linus Torvalds 已提交
1219
				unsigned long addr, unsigned long end,
1220
				struct zap_details *details)
L
Linus Torvalds 已提交
1221 1222 1223 1224 1225 1226 1227
{
	pud_t *pud;
	unsigned long next;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
1228
		if (pud_none_or_clear_bad(pud))
L
Linus Torvalds 已提交
1229
			continue;
1230 1231
		next = zap_pmd_range(tlb, vma, pud, addr, next, details);
	} while (pud++, addr = next, addr != end);
1232 1233

	return addr;
L
Linus Torvalds 已提交
1234 1235
}

1236
void unmap_page_range(struct mmu_gather *tlb,
1237 1238 1239
			     struct vm_area_struct *vma,
			     unsigned long addr, unsigned long end,
			     struct zap_details *details)
L
Linus Torvalds 已提交
1240 1241 1242 1243 1244 1245 1246 1247 1248
{
	pgd_t *pgd;
	unsigned long next;

	BUG_ON(addr >= end);
	tlb_start_vma(tlb, vma);
	pgd = pgd_offset(vma->vm_mm, addr);
	do {
		next = pgd_addr_end(addr, end);
1249
		if (pgd_none_or_clear_bad(pgd))
L
Linus Torvalds 已提交
1250
			continue;
1251 1252
		next = zap_pud_range(tlb, vma, pgd, addr, next, details);
	} while (pgd++, addr = next, addr != end);
L
Linus Torvalds 已提交
1253 1254
	tlb_end_vma(tlb, vma);
}
1255

1256 1257 1258

static void unmap_single_vma(struct mmu_gather *tlb,
		struct vm_area_struct *vma, unsigned long start_addr,
1259
		unsigned long end_addr,
1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
		struct zap_details *details)
{
	unsigned long start = max(vma->vm_start, start_addr);
	unsigned long end;

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

1271 1272 1273
	if (vma->vm_file)
		uprobe_munmap(vma, start, end);

1274
	if (unlikely(vma->vm_flags & VM_PFNMAP))
1275
		untrack_pfn(vma, 0, 0);
1276 1277 1278 1279 1280 1281 1282

	if (start != end) {
		if (unlikely(is_vm_hugetlb_page(vma))) {
			/*
			 * It is undesirable to test vma->vm_file as it
			 * should be non-null for valid hugetlb area.
			 * However, vm_file will be NULL in the error
1283
			 * cleanup path of mmap_region. When
1284
			 * hugetlbfs ->mmap method fails,
1285
			 * mmap_region() nullifies vma->vm_file
1286 1287 1288 1289
			 * before calling this function to clean up.
			 * Since no pte has actually been setup, it is
			 * safe to do nothing in this case.
			 */
1290
			if (vma->vm_file) {
1291
				i_mmap_lock_write(vma->vm_file->f_mapping);
1292
				__unmap_hugepage_range_final(tlb, vma, start, end, NULL);
1293
				i_mmap_unlock_write(vma->vm_file->f_mapping);
1294
			}
1295 1296 1297
		} else
			unmap_page_range(tlb, vma, start, end, details);
	}
L
Linus Torvalds 已提交
1298 1299 1300 1301
}

/**
 * unmap_vmas - unmap a range of memory covered by a list of vma's
1302
 * @tlb: address of the caller's struct mmu_gather
L
Linus Torvalds 已提交
1303 1304 1305 1306
 * @vma: the starting vma
 * @start_addr: virtual address at which to start unmapping
 * @end_addr: virtual address at which to end unmapping
 *
1307
 * Unmap all pages in the vma list.
L
Linus Torvalds 已提交
1308 1309 1310 1311 1312 1313 1314 1315 1316 1317
 *
 * 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.
 */
1318
void unmap_vmas(struct mmu_gather *tlb,
L
Linus Torvalds 已提交
1319
		struct vm_area_struct *vma, unsigned long start_addr,
1320
		unsigned long end_addr)
L
Linus Torvalds 已提交
1321
{
A
Andrea Arcangeli 已提交
1322
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
1323

A
Andrea Arcangeli 已提交
1324
	mmu_notifier_invalidate_range_start(mm, start_addr, end_addr);
1325
	for ( ; vma && vma->vm_start < end_addr; vma = vma->vm_next)
1326
		unmap_single_vma(tlb, vma, start_addr, end_addr, NULL);
A
Andrea Arcangeli 已提交
1327
	mmu_notifier_invalidate_range_end(mm, start_addr, end_addr);
L
Linus Torvalds 已提交
1328 1329 1330 1331 1332
}

/**
 * zap_page_range - remove user pages in a given range
 * @vma: vm_area_struct holding the applicable pages
1333
 * @start: starting address of pages to zap
L
Linus Torvalds 已提交
1334
 * @size: number of bytes to zap
1335
 * @details: details of shared cache invalidation
1336 1337
 *
 * Caller must protect the VMA list
L
Linus Torvalds 已提交
1338
 */
1339
void zap_page_range(struct vm_area_struct *vma, unsigned long start,
L
Linus Torvalds 已提交
1340 1341 1342
		unsigned long size, struct zap_details *details)
{
	struct mm_struct *mm = vma->vm_mm;
P
Peter Zijlstra 已提交
1343
	struct mmu_gather tlb;
1344
	unsigned long end = start + size;
L
Linus Torvalds 已提交
1345 1346

	lru_add_drain();
1347
	tlb_gather_mmu(&tlb, mm, start, end);
1348
	update_hiwater_rss(mm);
1349 1350
	mmu_notifier_invalidate_range_start(mm, start, end);
	for ( ; vma && vma->vm_start < end; vma = vma->vm_next)
1351
		unmap_single_vma(&tlb, vma, start, end, details);
1352 1353
	mmu_notifier_invalidate_range_end(mm, start, end);
	tlb_finish_mmu(&tlb, start, end);
L
Linus Torvalds 已提交
1354 1355
}

1356 1357 1358 1359 1360
/**
 * zap_page_range_single - remove user pages in a given range
 * @vma: vm_area_struct holding the applicable pages
 * @address: starting address of pages to zap
 * @size: number of bytes to zap
1361
 * @details: details of shared cache invalidation
1362 1363
 *
 * The range must fit into one VMA.
L
Linus Torvalds 已提交
1364
 */
1365
static void zap_page_range_single(struct vm_area_struct *vma, unsigned long address,
L
Linus Torvalds 已提交
1366 1367 1368
		unsigned long size, struct zap_details *details)
{
	struct mm_struct *mm = vma->vm_mm;
P
Peter Zijlstra 已提交
1369
	struct mmu_gather tlb;
L
Linus Torvalds 已提交
1370 1371 1372
	unsigned long end = address + size;

	lru_add_drain();
1373
	tlb_gather_mmu(&tlb, mm, address, end);
1374
	update_hiwater_rss(mm);
1375
	mmu_notifier_invalidate_range_start(mm, address, end);
1376
	unmap_single_vma(&tlb, vma, address, end, details);
1377
	mmu_notifier_invalidate_range_end(mm, address, end);
P
Peter Zijlstra 已提交
1378
	tlb_finish_mmu(&tlb, address, end);
L
Linus Torvalds 已提交
1379 1380
}

1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398
/**
 * 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;
1399
	zap_page_range_single(vma, address, size, NULL);
1400 1401 1402 1403
	return 0;
}
EXPORT_SYMBOL_GPL(zap_vma_ptes);

1404
pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
1405
			spinlock_t **ptl)
1406 1407 1408 1409
{
	pgd_t * pgd = pgd_offset(mm, addr);
	pud_t * pud = pud_alloc(mm, pgd, addr);
	if (pud) {
1410
		pmd_t * pmd = pmd_alloc(mm, pud, addr);
1411 1412
		if (pmd) {
			VM_BUG_ON(pmd_trans_huge(*pmd));
1413
			return pte_alloc_map_lock(mm, pmd, addr, ptl);
1414
		}
1415 1416 1417 1418
	}
	return NULL;
}

1419 1420 1421 1422 1423 1424 1425
/*
 * 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
1426 1427
static int insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page, pgprot_t prot)
1428
{
N
Nick Piggin 已提交
1429
	struct mm_struct *mm = vma->vm_mm;
1430
	int retval;
1431
	pte_t *pte;
1432 1433
	spinlock_t *ptl;

1434
	retval = -EINVAL;
1435
	if (PageAnon(page))
1436
		goto out;
1437 1438
	retval = -ENOMEM;
	flush_dcache_page(page);
1439
	pte = get_locked_pte(mm, addr, &ptl);
1440
	if (!pte)
1441
		goto out;
1442 1443 1444 1445 1446 1447
	retval = -EBUSY;
	if (!pte_none(*pte))
		goto out_unlock;

	/* Ok, finally just insert the thing.. */
	get_page(page);
1448
	inc_mm_counter_fast(mm, mm_counter_file(page));
1449 1450 1451 1452
	page_add_file_rmap(page);
	set_pte_at(mm, addr, pte, mk_pte(page, prot));

	retval = 0;
1453 1454
	pte_unmap_unlock(pte, ptl);
	return retval;
1455 1456 1457 1458 1459 1460
out_unlock:
	pte_unmap_unlock(pte, ptl);
out:
	return retval;
}

1461 1462 1463 1464 1465 1466
/**
 * 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
 *
1467 1468 1469 1470 1471 1472
 * 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
1473
 * (see split_page()).
1474 1475 1476 1477 1478 1479 1480 1481
 *
 * 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.
1482 1483 1484 1485 1486
 *
 * Usually this function is called from f_op->mmap() handler
 * under mm->mmap_sem write-lock, so it can change vma->vm_flags.
 * Caller must set VM_MIXEDMAP on vma if it wants to call this
 * function from other places, for example from page-fault handler.
1487
 */
N
Nick Piggin 已提交
1488 1489
int vm_insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page)
1490 1491 1492 1493 1494
{
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
	if (!page_count(page))
		return -EINVAL;
1495 1496 1497 1498 1499
	if (!(vma->vm_flags & VM_MIXEDMAP)) {
		BUG_ON(down_read_trylock(&vma->vm_mm->mmap_sem));
		BUG_ON(vma->vm_flags & VM_PFNMAP);
		vma->vm_flags |= VM_MIXEDMAP;
	}
N
Nick Piggin 已提交
1500
	return insert_page(vma, addr, page, vma->vm_page_prot);
1501
}
1502
EXPORT_SYMBOL(vm_insert_page);
1503

N
Nick Piggin 已提交
1504
static int insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1505
			pfn_t pfn, pgprot_t prot)
N
Nick Piggin 已提交
1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520
{
	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.. */
1521 1522 1523 1524
	if (pfn_t_devmap(pfn))
		entry = pte_mkdevmap(pfn_t_pte(pfn, prot));
	else
		entry = pte_mkspecial(pfn_t_pte(pfn, prot));
N
Nick Piggin 已提交
1525
	set_pte_at(mm, addr, pte, entry);
1526
	update_mmu_cache(vma, addr, pte); /* XXX: why not for insert_page? */
N
Nick Piggin 已提交
1527 1528 1529 1530 1531 1532 1533 1534

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

N
Nick Piggin 已提交
1535 1536 1537 1538 1539 1540
/**
 * 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
 *
1541
 * Similar to vm_insert_page, this allows drivers to insert individual pages
N
Nick Piggin 已提交
1542 1543 1544 1545
 * 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
1546 1547 1548 1549 1550
 *
 * 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
1551 1552
 */
int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
N
Nick Piggin 已提交
1553
			unsigned long pfn)
1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575
{
	return vm_insert_pfn_prot(vma, addr, pfn, vma->vm_page_prot);
}
EXPORT_SYMBOL(vm_insert_pfn);

/**
 * vm_insert_pfn_prot - insert single pfn into user vma with specified pgprot
 * @vma: user vma to map to
 * @addr: target user address of this page
 * @pfn: source kernel pfn
 * @pgprot: pgprot flags for the inserted page
 *
 * This is exactly like vm_insert_pfn, except that it allows drivers to
 * to override pgprot on a per-page basis.
 *
 * This only makes sense for IO mappings, and it makes no sense for
 * cow mappings.  In general, using multiple vmas is preferable;
 * vm_insert_pfn_prot should only be used if using multiple VMAs is
 * impractical.
 */
int vm_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr,
			unsigned long pfn, pgprot_t pgprot)
N
Nick Piggin 已提交
1576
{
1577
	int ret;
1578 1579 1580 1581 1582 1583
	/*
	 * 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
1584 1585 1586 1587 1588
	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
1589

N
Nick Piggin 已提交
1590 1591
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
1592
	if (track_pfn_insert(vma, &pgprot, __pfn_to_pfn_t(pfn, PFN_DEV)))
1593 1594
		return -EINVAL;

1595
	ret = insert_pfn(vma, addr, __pfn_to_pfn_t(pfn, PFN_DEV), pgprot);
1596 1597

	return ret;
N
Nick Piggin 已提交
1598
}
1599
EXPORT_SYMBOL(vm_insert_pfn_prot);
N
Nick Piggin 已提交
1600

N
Nick Piggin 已提交
1601
int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1602
			pfn_t pfn)
N
Nick Piggin 已提交
1603 1604
{
	BUG_ON(!(vma->vm_flags & VM_MIXEDMAP));
N
Nick Piggin 已提交
1605

N
Nick Piggin 已提交
1606 1607
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
N
Nick Piggin 已提交
1608

N
Nick Piggin 已提交
1609 1610 1611 1612
	/*
	 * 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
1613 1614
	 * 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
1615
	 */
1616
	if (!HAVE_PTE_SPECIAL && !pfn_t_devmap(pfn) && pfn_t_valid(pfn)) {
N
Nick Piggin 已提交
1617 1618
		struct page *page;

1619 1620 1621 1622 1623 1624
		/*
		 * At this point we are committed to insert_page()
		 * regardless of whether the caller specified flags that
		 * result in pfn_t_has_page() == false.
		 */
		page = pfn_to_page(pfn_t_to_pfn(pfn));
N
Nick Piggin 已提交
1625 1626 1627
		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
1628
}
N
Nick Piggin 已提交
1629
EXPORT_SYMBOL(vm_insert_mixed);
N
Nick Piggin 已提交
1630

L
Linus Torvalds 已提交
1631 1632 1633 1634 1635 1636 1637 1638 1639 1640
/*
 * 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;
1641
	spinlock_t *ptl;
L
Linus Torvalds 已提交
1642

1643
	pte = pte_alloc_map_lock(mm, pmd, addr, &ptl);
L
Linus Torvalds 已提交
1644 1645
	if (!pte)
		return -ENOMEM;
1646
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
1647 1648
	do {
		BUG_ON(!pte_none(*pte));
1649
		set_pte_at(mm, addr, pte, pte_mkspecial(pfn_pte(pfn, prot)));
L
Linus Torvalds 已提交
1650 1651
		pfn++;
	} while (pte++, addr += PAGE_SIZE, addr != end);
1652
	arch_leave_lazy_mmu_mode();
1653
	pte_unmap_unlock(pte - 1, ptl);
L
Linus Torvalds 已提交
1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667
	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;
1668
	VM_BUG_ON(pmd_trans_huge(*pmd));
L
Linus Torvalds 已提交
1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697
	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;
}

1698 1699 1700 1701 1702 1703 1704 1705 1706 1707
/**
 * 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
1708 1709 1710 1711 1712
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;
1713
	unsigned long end = addr + PAGE_ALIGN(size);
L
Linus Torvalds 已提交
1714 1715 1716 1717 1718 1719 1720 1721
	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).
1722 1723 1724
	 *   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.
1725 1726 1727 1728
	 *   VM_DONTEXPAND
	 *      Disable vma merging and expanding with mremap().
	 *   VM_DONTDUMP
	 *      Omit vma from core dump, even when VM_IO turned off.
1729 1730 1731 1732
	 *
	 * 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".
1733
	 * See vm_normal_page() for details.
L
Linus Torvalds 已提交
1734
	 */
1735 1736 1737
	if (is_cow_mapping(vma->vm_flags)) {
		if (addr != vma->vm_start || end != vma->vm_end)
			return -EINVAL;
1738
		vma->vm_pgoff = pfn;
1739 1740 1741 1742
	}

	err = track_pfn_remap(vma, &prot, pfn, addr, PAGE_ALIGN(size));
	if (err)
1743
		return -EINVAL;
1744

1745
	vma->vm_flags |= VM_IO | VM_PFNMAP | VM_DONTEXPAND | VM_DONTDUMP;
L
Linus Torvalds 已提交
1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757

	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);
1758 1759

	if (err)
1760
		untrack_pfn(vma, pfn, PAGE_ALIGN(size));
1761

L
Linus Torvalds 已提交
1762 1763 1764 1765
	return err;
}
EXPORT_SYMBOL(remap_pfn_range);

1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812
/**
 * vm_iomap_memory - remap memory to userspace
 * @vma: user vma to map to
 * @start: start of area
 * @len: size of area
 *
 * This is a simplified io_remap_pfn_range() for common driver use. The
 * driver just needs to give us the physical memory range to be mapped,
 * we'll figure out the rest from the vma information.
 *
 * NOTE! Some drivers might want to tweak vma->vm_page_prot first to get
 * whatever write-combining details or similar.
 */
int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len)
{
	unsigned long vm_len, pfn, pages;

	/* Check that the physical memory area passed in looks valid */
	if (start + len < start)
		return -EINVAL;
	/*
	 * You *really* shouldn't map things that aren't page-aligned,
	 * but we've historically allowed it because IO memory might
	 * just have smaller alignment.
	 */
	len += start & ~PAGE_MASK;
	pfn = start >> PAGE_SHIFT;
	pages = (len + ~PAGE_MASK) >> PAGE_SHIFT;
	if (pfn + pages < pfn)
		return -EINVAL;

	/* We start the mapping 'vm_pgoff' pages into the area */
	if (vma->vm_pgoff > pages)
		return -EINVAL;
	pfn += vma->vm_pgoff;
	pages -= vma->vm_pgoff;

	/* Can we fit all of the mapping? */
	vm_len = vma->vm_end - vma->vm_start;
	if (vm_len >> PAGE_SHIFT > pages)
		return -EINVAL;

	/* Ok, let it rip */
	return io_remap_pfn_range(vma, vma->vm_start, pfn, vm_len, vma->vm_page_prot);
}
EXPORT_SYMBOL(vm_iomap_memory);

1813 1814 1815 1816 1817 1818
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;
1819
	pgtable_t token;
1820
	spinlock_t *uninitialized_var(ptl);
1821 1822 1823 1824 1825 1826 1827 1828 1829

	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));

1830 1831
	arch_enter_lazy_mmu_mode();

1832
	token = pmd_pgtable(*pmd);
1833 1834

	do {
1835
		err = fn(pte++, token, addr, data);
1836 1837
		if (err)
			break;
1838
	} while (addr += PAGE_SIZE, addr != end);
1839

1840 1841
	arch_leave_lazy_mmu_mode();

1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854
	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
1855 1856
	BUG_ON(pud_huge(*pud));

1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897
	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;
1898
	unsigned long end = addr + size;
1899 1900
	int err;

1901 1902 1903
	if (WARN_ON(addr >= end))
		return -EINVAL;

1904 1905 1906 1907 1908 1909 1910
	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);
1911

1912 1913 1914 1915
	return err;
}
EXPORT_SYMBOL_GPL(apply_to_page_range);

1916
/*
1917 1918 1919 1920 1921
 * 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 might give a mix of unmatched
 * parts, do_swap_page must check under lock before unmapping the pte and
 * proceeding (but do_wp_page is only called after already making such a check;
1922
 * and do_anonymous_page can safely check later on).
1923
 */
1924
static inline int pte_unmap_same(struct mm_struct *mm, pmd_t *pmd,
1925 1926 1927 1928 1929
				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)) {
1930 1931
		spinlock_t *ptl = pte_lockptr(mm, pmd);
		spin_lock(ptl);
1932
		same = pte_same(*page_table, orig_pte);
1933
		spin_unlock(ptl);
1934 1935 1936 1937 1938 1939
	}
#endif
	pte_unmap(page_table);
	return same;
}

1940
static inline void cow_user_page(struct page *dst, struct page *src, unsigned long va, struct vm_area_struct *vma)
1941
{
1942 1943
	debug_dma_assert_idle(src);

1944 1945 1946 1947 1948 1949 1950
	/*
	 * 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)) {
1951
		void *kaddr = kmap_atomic(dst);
1952 1953 1954 1955 1956 1957 1958 1959 1960
		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))
1961
			clear_page(kaddr);
1962
		kunmap_atomic(kaddr);
1963
		flush_dcache_page(dst);
1964 1965
	} else
		copy_user_highpage(dst, src, va, vma);
1966 1967
}

1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981
static gfp_t __get_fault_gfp_mask(struct vm_area_struct *vma)
{
	struct file *vm_file = vma->vm_file;

	if (vm_file)
		return mapping_gfp_mask(vm_file->f_mapping) | __GFP_FS | __GFP_IO;

	/*
	 * Special mappings (e.g. VDSO) do not have any file so fake
	 * a default GFP_KERNEL for them.
	 */
	return GFP_KERNEL;
}

1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
/*
 * 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.
 */
static int do_page_mkwrite(struct vm_area_struct *vma, struct page *page,
	       unsigned long address)
{
	struct vm_fault vmf;
	int ret;

	vmf.virtual_address = (void __user *)(address & PAGE_MASK);
	vmf.pgoff = page->index;
	vmf.flags = FAULT_FLAG_WRITE|FAULT_FLAG_MKWRITE;
1997
	vmf.gfp_mask = __get_fault_gfp_mask(vma);
1998
	vmf.page = page;
1999
	vmf.cow_page = NULL;
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

	ret = vma->vm_ops->page_mkwrite(vma, &vmf);
	if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE)))
		return ret;
	if (unlikely(!(ret & VM_FAULT_LOCKED))) {
		lock_page(page);
		if (!page->mapping) {
			unlock_page(page);
			return 0; /* retry */
		}
		ret |= VM_FAULT_LOCKED;
	} else
		VM_BUG_ON_PAGE(!PageLocked(page), page);
	return ret;
}

2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057
/*
 * Handle write page faults for pages that can be reused in the current vma
 *
 * This can happen either due to the mapping being with the VM_SHARED flag,
 * or due to us being the last reference standing to the page. In either
 * case, all we need to do here is to mark the page as writable and update
 * any related book-keeping.
 */
static inline int wp_page_reuse(struct mm_struct *mm,
			struct vm_area_struct *vma, unsigned long address,
			pte_t *page_table, spinlock_t *ptl, pte_t orig_pte,
			struct page *page, int page_mkwrite,
			int dirty_shared)
	__releases(ptl)
{
	pte_t entry;
	/*
	 * Clear the pages cpupid information as the existing
	 * information potentially belongs to a now completely
	 * unrelated process.
	 */
	if (page)
		page_cpupid_xchg_last(page, (1 << LAST_CPUPID_SHIFT) - 1);

	flush_cache_page(vma, address, pte_pfn(orig_pte));
	entry = pte_mkyoung(orig_pte);
	entry = maybe_mkwrite(pte_mkdirty(entry), vma);
	if (ptep_set_access_flags(vma, address, page_table, entry, 1))
		update_mmu_cache(vma, address, page_table);
	pte_unmap_unlock(page_table, ptl);

	if (dirty_shared) {
		struct address_space *mapping;
		int dirtied;

		if (!page_mkwrite)
			lock_page(page);

		dirtied = set_page_dirty(page);
		VM_BUG_ON_PAGE(PageAnon(page), page);
		mapping = page->mapping;
		unlock_page(page);
2058
		put_page(page);
2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074

		if ((dirtied || page_mkwrite) && mapping) {
			/*
			 * Some device drivers do not set page.mapping
			 * but still dirty their pages
			 */
			balance_dirty_pages_ratelimited(mapping);
		}

		if (!page_mkwrite)
			file_update_time(vma->vm_file);
	}

	return VM_FAULT_WRITE;
}

2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116
/*
 * Handle the case of a page which we actually need to copy to a new page.
 *
 * Called with mmap_sem locked and the old page referenced, but
 * without the ptl held.
 *
 * High level logic flow:
 *
 * - Allocate a page, copy the content of the old page to the new one.
 * - Handle book keeping and accounting - cgroups, mmu-notifiers, etc.
 * - Take the PTL. If the pte changed, bail out and release the allocated page
 * - If the pte is still the way we remember it, update the page table and all
 *   relevant references. This includes dropping the reference the page-table
 *   held to the old page, as well as updating the rmap.
 * - In any case, unlock the PTL and drop the reference we took to the old page.
 */
static int wp_page_copy(struct mm_struct *mm, struct vm_area_struct *vma,
			unsigned long address, pte_t *page_table, pmd_t *pmd,
			pte_t orig_pte, struct page *old_page)
{
	struct page *new_page = NULL;
	spinlock_t *ptl = NULL;
	pte_t entry;
	int page_copied = 0;
	const unsigned long mmun_start = address & PAGE_MASK;	/* For mmu_notifiers */
	const unsigned long mmun_end = mmun_start + PAGE_SIZE;	/* For mmu_notifiers */
	struct mem_cgroup *memcg;

	if (unlikely(anon_vma_prepare(vma)))
		goto oom;

	if (is_zero_pfn(pte_pfn(orig_pte))) {
		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);
	}

2117
	if (mem_cgroup_try_charge(new_page, mm, GFP_KERNEL, &memcg, false))
2118 2119
		goto oom_free_new;

2120 2121
	__SetPageUptodate(new_page);

2122 2123 2124 2125 2126 2127 2128 2129 2130
	mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);

	/*
	 * Re-check the pte - we dropped the lock
	 */
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
	if (likely(pte_same(*page_table, orig_pte))) {
		if (old_page) {
			if (!PageAnon(old_page)) {
2131 2132
				dec_mm_counter_fast(mm,
						mm_counter_file(old_page));
2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147
				inc_mm_counter_fast(mm, MM_ANONPAGES);
			}
		} else {
			inc_mm_counter_fast(mm, MM_ANONPAGES);
		}
		flush_cache_page(vma, address, pte_pfn(orig_pte));
		entry = mk_pte(new_page, vma->vm_page_prot);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
		/*
		 * 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.
		 */
		ptep_clear_flush_notify(vma, address, page_table);
2148
		page_add_new_anon_rmap(new_page, vma, address, false);
2149
		mem_cgroup_commit_charge(new_page, memcg, false, false);
2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180
		lru_cache_add_active_or_unevictable(new_page, vma);
		/*
		 * 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);
		update_mmu_cache(vma, address, page_table);
		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.
			 */
2181
			page_remove_rmap(old_page, false);
2182 2183 2184 2185 2186 2187
		}

		/* Free the old page.. */
		new_page = old_page;
		page_copied = 1;
	} else {
2188
		mem_cgroup_cancel_charge(new_page, memcg, false);
2189 2190 2191
	}

	if (new_page)
2192
		put_page(new_page);
2193 2194 2195 2196 2197 2198 2199 2200 2201 2202

	pte_unmap_unlock(page_table, ptl);
	mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
	if (old_page) {
		/*
		 * Don't let another task, with possibly unlocked vma,
		 * keep the mlocked page.
		 */
		if (page_copied && (vma->vm_flags & VM_LOCKED)) {
			lock_page(old_page);	/* LRU manipulation */
2203 2204
			if (PageMlocked(old_page))
				munlock_vma_page(old_page);
2205 2206
			unlock_page(old_page);
		}
2207
		put_page(old_page);
2208 2209 2210
	}
	return page_copied ? VM_FAULT_WRITE : 0;
oom_free_new:
2211
	put_page(new_page);
2212 2213
oom:
	if (old_page)
2214
		put_page(old_page);
2215 2216 2217
	return VM_FAULT_OOM;
}

2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253
/*
 * Handle write page faults for VM_MIXEDMAP or VM_PFNMAP for a VM_SHARED
 * mapping
 */
static int wp_pfn_shared(struct mm_struct *mm,
			struct vm_area_struct *vma, unsigned long address,
			pte_t *page_table, spinlock_t *ptl, pte_t orig_pte,
			pmd_t *pmd)
{
	if (vma->vm_ops && vma->vm_ops->pfn_mkwrite) {
		struct vm_fault vmf = {
			.page = NULL,
			.pgoff = linear_page_index(vma, address),
			.virtual_address = (void __user *)(address & PAGE_MASK),
			.flags = FAULT_FLAG_WRITE | FAULT_FLAG_MKWRITE,
		};
		int ret;

		pte_unmap_unlock(page_table, ptl);
		ret = vma->vm_ops->pfn_mkwrite(vma, &vmf);
		if (ret & VM_FAULT_ERROR)
			return ret;
		page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
		/*
		 * We might have raced with another page fault while we
		 * released the pte_offset_map_lock.
		 */
		if (!pte_same(*page_table, orig_pte)) {
			pte_unmap_unlock(page_table, ptl);
			return 0;
		}
	}
	return wp_page_reuse(mm, vma, address, page_table, ptl, orig_pte,
			     NULL, 0, 0);
}

2254 2255 2256 2257 2258 2259 2260 2261
static int wp_page_shared(struct mm_struct *mm, struct vm_area_struct *vma,
			  unsigned long address, pte_t *page_table,
			  pmd_t *pmd, spinlock_t *ptl, pte_t orig_pte,
			  struct page *old_page)
	__releases(ptl)
{
	int page_mkwrite = 0;

2262
	get_page(old_page);
2263 2264 2265 2266 2267 2268 2269 2270

	if (vma->vm_ops && vma->vm_ops->page_mkwrite) {
		int tmp;

		pte_unmap_unlock(page_table, ptl);
		tmp = do_page_mkwrite(vma, old_page, address);
		if (unlikely(!tmp || (tmp &
				      (VM_FAULT_ERROR | VM_FAULT_NOPAGE)))) {
2271
			put_page(old_page);
2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284
			return tmp;
		}
		/*
		 * 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);
		if (!pte_same(*page_table, orig_pte)) {
			unlock_page(old_page);
			pte_unmap_unlock(page_table, ptl);
2285
			put_page(old_page);
2286 2287 2288 2289 2290 2291 2292 2293 2294
			return 0;
		}
		page_mkwrite = 1;
	}

	return wp_page_reuse(mm, vma, address, page_table, ptl,
			     orig_pte, old_page, page_mkwrite, 1);
}

L
Linus Torvalds 已提交
2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308
/*
 * 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.
 *
2309 2310 2311
 * 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
2312
 */
2313 2314
static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2315
		spinlock_t *ptl, pte_t orig_pte)
2316
	__releases(ptl)
L
Linus Torvalds 已提交
2317
{
2318
	struct page *old_page;
L
Linus Torvalds 已提交
2319

2320
	old_page = vm_normal_page(vma, address, orig_pte);
2321 2322
	if (!old_page) {
		/*
2323 2324
		 * VM_MIXEDMAP !pfn_valid() case, or VM_SOFTDIRTY clear on a
		 * VM_PFNMAP VMA.
2325 2326
		 *
		 * We should not cow pages in a shared writeable mapping.
2327
		 * Just mark the pages writable and/or call ops->pfn_mkwrite.
2328 2329 2330
		 */
		if ((vma->vm_flags & (VM_WRITE|VM_SHARED)) ==
				     (VM_WRITE|VM_SHARED))
2331 2332
			return wp_pfn_shared(mm, vma, address, page_table, ptl,
					     orig_pte, pmd);
2333 2334 2335 2336

		pte_unmap_unlock(page_table, ptl);
		return wp_page_copy(mm, vma, address, page_table, pmd,
				    orig_pte, old_page);
2337
	}
L
Linus Torvalds 已提交
2338

2339
	/*
2340 2341
	 * Take out anonymous pages first, anonymous shared vmas are
	 * not dirty accountable.
2342
	 */
2343
	if (PageAnon(old_page) && !PageKsm(old_page)) {
2344
		if (!trylock_page(old_page)) {
2345
			get_page(old_page);
2346 2347 2348 2349 2350 2351
			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);
2352
				pte_unmap_unlock(page_table, ptl);
2353
				put_page(old_page);
2354
				return 0;
2355
			}
2356
			put_page(old_page);
2357
		}
2358
		if (reuse_swap_page(old_page)) {
2359 2360 2361 2362 2363 2364
			/*
			 * 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);
2365
			unlock_page(old_page);
2366 2367
			return wp_page_reuse(mm, vma, address, page_table, ptl,
					     orig_pte, old_page, 0, 0);
2368
		}
2369
		unlock_page(old_page);
2370
	} else if (unlikely((vma->vm_flags & (VM_WRITE|VM_SHARED)) ==
2371
					(VM_WRITE|VM_SHARED))) {
2372 2373
		return wp_page_shared(mm, vma, address, page_table, pmd,
				      ptl, orig_pte, old_page);
L
Linus Torvalds 已提交
2374 2375 2376 2377 2378
	}

	/*
	 * Ok, we need to copy. Oh, well..
	 */
2379
	get_page(old_page);
2380

2381
	pte_unmap_unlock(page_table, ptl);
2382 2383
	return wp_page_copy(mm, vma, address, page_table, pmd,
			    orig_pte, old_page);
L
Linus Torvalds 已提交
2384 2385
}

2386
static void unmap_mapping_range_vma(struct vm_area_struct *vma,
L
Linus Torvalds 已提交
2387 2388 2389
		unsigned long start_addr, unsigned long end_addr,
		struct zap_details *details)
{
2390
	zap_page_range_single(vma, start_addr, end_addr - start_addr, details);
L
Linus Torvalds 已提交
2391 2392
}

2393
static inline void unmap_mapping_range_tree(struct rb_root *root,
L
Linus Torvalds 已提交
2394 2395 2396 2397 2398
					    struct zap_details *details)
{
	struct vm_area_struct *vma;
	pgoff_t vba, vea, zba, zea;

2399
	vma_interval_tree_foreach(vma, root,
L
Linus Torvalds 已提交
2400 2401 2402
			details->first_index, details->last_index) {

		vba = vma->vm_pgoff;
2403
		vea = vba + vma_pages(vma) - 1;
L
Linus Torvalds 已提交
2404 2405 2406 2407 2408 2409 2410
		zba = details->first_index;
		if (zba < vba)
			zba = vba;
		zea = details->last_index;
		if (zea > vea)
			zea = vea;

2411
		unmap_mapping_range_vma(vma,
L
Linus Torvalds 已提交
2412 2413
			((zba - vba) << PAGE_SHIFT) + vma->vm_start,
			((zea - vba + 1) << PAGE_SHIFT) + vma->vm_start,
2414
				details);
L
Linus Torvalds 已提交
2415 2416 2417 2418
	}
}

/**
2419 2420 2421 2422
 * unmap_mapping_range - unmap the portion of all mmaps in the specified
 * address_space corresponding to the specified page range in the underlying
 * file.
 *
2423
 * @mapping: the address space containing mmaps to be unmapped.
L
Linus Torvalds 已提交
2424 2425
 * @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
2426
 * boundary.  Note that this is different from truncate_pagecache(), which
L
Linus Torvalds 已提交
2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437
 * 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)
{
2438
	struct zap_details details = { };
L
Linus Torvalds 已提交
2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455
	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.first_index = hba;
	details.last_index = hba + hlen - 1;
	if (details.last_index < details.first_index)
		details.last_index = ULONG_MAX;

2456 2457

	/* DAX uses i_mmap_lock to serialise file truncate vs page fault */
2458
	i_mmap_lock_write(mapping);
2459
	if (unlikely(!RB_EMPTY_ROOT(&mapping->i_mmap)))
L
Linus Torvalds 已提交
2460
		unmap_mapping_range_tree(&mapping->i_mmap, &details);
2461
	i_mmap_unlock_write(mapping);
L
Linus Torvalds 已提交
2462 2463 2464 2465
}
EXPORT_SYMBOL(unmap_mapping_range);

/*
2466 2467
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
 * but allow concurrent faults), and pte mapped but not yet locked.
2468 2469 2470 2471
 * We return with pte unmapped and unlocked.
 *
 * We return with the mmap_sem locked or unlocked in the same cases
 * as does filemap_fault().
L
Linus Torvalds 已提交
2472
 */
2473 2474
static int do_swap_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2475
		unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
2476
{
2477
	spinlock_t *ptl;
2478
	struct page *page, *swapcache;
2479
	struct mem_cgroup *memcg;
2480
	swp_entry_t entry;
L
Linus Torvalds 已提交
2481
	pte_t pte;
2482
	int locked;
2483
	int exclusive = 0;
N
Nick Piggin 已提交
2484
	int ret = 0;
L
Linus Torvalds 已提交
2485

2486
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
2487
		goto out;
2488 2489

	entry = pte_to_swp_entry(orig_pte);
2490 2491 2492 2493 2494 2495 2496
	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);
2497
			ret = VM_FAULT_SIGBUS;
2498
		}
2499 2500
		goto out;
	}
2501
	delayacct_set_flag(DELAYACCT_PF_SWAPIN);
L
Linus Torvalds 已提交
2502 2503
	page = lookup_swap_cache(entry);
	if (!page) {
2504 2505
		page = swapin_readahead(entry,
					GFP_HIGHUSER_MOVABLE, vma, address);
L
Linus Torvalds 已提交
2506 2507
		if (!page) {
			/*
2508 2509
			 * Back out if somebody else faulted in this pte
			 * while we released the pte lock.
L
Linus Torvalds 已提交
2510
			 */
2511
			page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
2512 2513
			if (likely(pte_same(*page_table, orig_pte)))
				ret = VM_FAULT_OOM;
2514
			delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2515
			goto unlock;
L
Linus Torvalds 已提交
2516 2517 2518 2519
		}

		/* Had to read the page from swap area: Major fault */
		ret = VM_FAULT_MAJOR;
2520
		count_vm_event(PGMAJFAULT);
2521
		mem_cgroup_count_vm_event(mm, PGMAJFAULT);
2522
	} else if (PageHWPoison(page)) {
2523 2524 2525 2526
		/*
		 * hwpoisoned dirty swapcache pages are kept for killing
		 * owner processes (which may be unknown at hwpoison time)
		 */
2527 2528
		ret = VM_FAULT_HWPOISON;
		delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2529
		swapcache = page;
2530
		goto out_release;
L
Linus Torvalds 已提交
2531 2532
	}

2533
	swapcache = page;
2534
	locked = lock_page_or_retry(page, mm, flags);
R
Rik van Riel 已提交
2535

2536
	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2537 2538 2539 2540
	if (!locked) {
		ret |= VM_FAULT_RETRY;
		goto out_release;
	}
2541

2542
	/*
2543 2544 2545 2546
	 * 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.
2547
	 */
2548
	if (unlikely(!PageSwapCache(page) || page_private(page) != entry.val))
2549 2550
		goto out_page;

2551 2552 2553 2554 2555
	page = ksm_might_need_to_copy(page, vma, address);
	if (unlikely(!page)) {
		ret = VM_FAULT_OOM;
		page = swapcache;
		goto out_page;
2556 2557
	}

2558
	if (mem_cgroup_try_charge(page, mm, GFP_KERNEL, &memcg, false)) {
2559
		ret = VM_FAULT_OOM;
2560
		goto out_page;
2561 2562
	}

L
Linus Torvalds 已提交
2563
	/*
2564
	 * Back out if somebody else already faulted in this pte.
L
Linus Torvalds 已提交
2565
	 */
2566
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
2567
	if (unlikely(!pte_same(*page_table, orig_pte)))
2568 2569 2570 2571 2572
		goto out_nomap;

	if (unlikely(!PageUptodate(page))) {
		ret = VM_FAULT_SIGBUS;
		goto out_nomap;
L
Linus Torvalds 已提交
2573 2574
	}

2575 2576 2577 2578 2579 2580 2581 2582 2583
	/*
	 * 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.
	 */
L
Linus Torvalds 已提交
2584

2585
	inc_mm_counter_fast(mm, MM_ANONPAGES);
K
KAMEZAWA Hiroyuki 已提交
2586
	dec_mm_counter_fast(mm, MM_SWAPENTS);
L
Linus Torvalds 已提交
2587
	pte = mk_pte(page, vma->vm_page_prot);
2588
	if ((flags & FAULT_FLAG_WRITE) && reuse_swap_page(page)) {
L
Linus Torvalds 已提交
2589
		pte = maybe_mkwrite(pte_mkdirty(pte), vma);
2590
		flags &= ~FAULT_FLAG_WRITE;
2591
		ret |= VM_FAULT_WRITE;
2592
		exclusive = RMAP_EXCLUSIVE;
L
Linus Torvalds 已提交
2593 2594
	}
	flush_icache_page(vma, page);
2595 2596
	if (pte_swp_soft_dirty(orig_pte))
		pte = pte_mksoft_dirty(pte);
L
Linus Torvalds 已提交
2597
	set_pte_at(mm, address, page_table, pte);
2598
	if (page == swapcache) {
2599
		do_page_add_anon_rmap(page, vma, address, exclusive);
2600
		mem_cgroup_commit_charge(page, memcg, true, false);
2601
	} else { /* ksm created a completely new copy */
2602
		page_add_new_anon_rmap(page, vma, address, false);
2603
		mem_cgroup_commit_charge(page, memcg, false, false);
2604 2605
		lru_cache_add_active_or_unevictable(page, vma);
	}
L
Linus Torvalds 已提交
2606

2607
	swap_free(entry);
2608 2609
	if (mem_cgroup_swap_full(page) ||
	    (vma->vm_flags & VM_LOCKED) || PageMlocked(page))
2610
		try_to_free_swap(page);
2611
	unlock_page(page);
2612
	if (page != swapcache) {
2613 2614 2615 2616 2617 2618 2619 2620 2621
		/*
		 * 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);
2622
		put_page(swapcache);
2623
	}
2624

2625
	if (flags & FAULT_FLAG_WRITE) {
2626 2627 2628
		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
2629 2630 2631 2632
		goto out;
	}

	/* No need to invalidate - it was non-present before */
2633
	update_mmu_cache(vma, address, page_table);
2634
unlock:
2635
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
2636 2637
out:
	return ret;
2638
out_nomap:
2639
	mem_cgroup_cancel_charge(page, memcg, false);
2640
	pte_unmap_unlock(page_table, ptl);
2641
out_page:
2642
	unlock_page(page);
2643
out_release:
2644
	put_page(page);
2645
	if (page != swapcache) {
2646
		unlock_page(swapcache);
2647
		put_page(swapcache);
2648
	}
2649
	return ret;
L
Linus Torvalds 已提交
2650 2651
}

2652
/*
2653 2654
 * This is like a special single-page "expand_{down|up}wards()",
 * except we must first make sure that 'address{-|+}PAGE_SIZE'
2655 2656 2657 2658 2659 2660
 * 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) {
2661 2662 2663 2664 2665 2666 2667 2668 2669 2670
		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;
2671

2672
		return expand_downwards(vma, address - PAGE_SIZE);
2673
	}
2674 2675 2676 2677 2678 2679 2680
	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;

2681
		return expand_upwards(vma, address + PAGE_SIZE);
2682
	}
2683 2684 2685
	return 0;
}

L
Linus Torvalds 已提交
2686
/*
2687 2688 2689
 * 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
2690
 */
2691 2692
static int do_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2693
		unsigned int flags)
L
Linus Torvalds 已提交
2694
{
2695
	struct mem_cgroup *memcg;
2696 2697
	struct page *page;
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2698 2699
	pte_t entry;

2700 2701
	pte_unmap(page_table);

2702 2703 2704 2705
	/* File mapping without ->vm_ops ? */
	if (vma->vm_flags & VM_SHARED)
		return VM_FAULT_SIGBUS;

2706 2707
	/* Check if we need to add a guard page to the stack */
	if (check_stack_guard_page(vma, address) < 0)
2708
		return VM_FAULT_SIGSEGV;
2709

2710
	/* Use the zero-page for reads */
2711
	if (!(flags & FAULT_FLAG_WRITE) && !mm_forbids_zeropage(mm)) {
2712 2713
		entry = pte_mkspecial(pfn_pte(my_zero_pfn(address),
						vma->vm_page_prot));
2714
		page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
H
Hugh Dickins 已提交
2715 2716
		if (!pte_none(*page_table))
			goto unlock;
2717 2718 2719 2720 2721 2722
		/* Deliver the page fault to userland, check inside PT lock */
		if (userfaultfd_missing(vma)) {
			pte_unmap_unlock(page_table, ptl);
			return handle_userfault(vma, address, flags,
						VM_UFFD_MISSING);
		}
H
Hugh Dickins 已提交
2723 2724 2725
		goto setpte;
	}

N
Nick Piggin 已提交
2726 2727 2728 2729 2730 2731
	/* 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;
2732

2733
	if (mem_cgroup_try_charge(page, mm, GFP_KERNEL, &memcg, false))
2734 2735
		goto oom_free_page;

2736 2737 2738 2739 2740
	/*
	 * The memory barrier inside __SetPageUptodate makes sure that
	 * preceeding stores to the page contents become visible before
	 * the set_pte_at() write.
	 */
2741
	__SetPageUptodate(page);
2742

N
Nick Piggin 已提交
2743
	entry = mk_pte(page, vma->vm_page_prot);
H
Hugh Dickins 已提交
2744 2745
	if (vma->vm_flags & VM_WRITE)
		entry = pte_mkwrite(pte_mkdirty(entry));
L
Linus Torvalds 已提交
2746

N
Nick Piggin 已提交
2747
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
2748
	if (!pte_none(*page_table))
N
Nick Piggin 已提交
2749
		goto release;
H
Hugh Dickins 已提交
2750

2751 2752 2753
	/* Deliver the page fault to userland, check inside PT lock */
	if (userfaultfd_missing(vma)) {
		pte_unmap_unlock(page_table, ptl);
2754
		mem_cgroup_cancel_charge(page, memcg, false);
2755
		put_page(page);
2756 2757 2758 2759
		return handle_userfault(vma, address, flags,
					VM_UFFD_MISSING);
	}

2760
	inc_mm_counter_fast(mm, MM_ANONPAGES);
2761
	page_add_new_anon_rmap(page, vma, address, false);
2762
	mem_cgroup_commit_charge(page, memcg, false, false);
2763
	lru_cache_add_active_or_unevictable(page, vma);
H
Hugh Dickins 已提交
2764
setpte:
2765
	set_pte_at(mm, address, page_table, entry);
L
Linus Torvalds 已提交
2766 2767

	/* No need to invalidate - it was non-present before */
2768
	update_mmu_cache(vma, address, page_table);
2769
unlock:
2770
	pte_unmap_unlock(page_table, ptl);
N
Nick Piggin 已提交
2771
	return 0;
2772
release:
2773
	mem_cgroup_cancel_charge(page, memcg, false);
2774
	put_page(page);
2775
	goto unlock;
2776
oom_free_page:
2777
	put_page(page);
2778
oom:
L
Linus Torvalds 已提交
2779 2780 2781
	return VM_FAULT_OOM;
}

2782 2783 2784 2785 2786
/*
 * The mmap_sem must have been held on entry, and may have been
 * released depending on flags and vma->vm_ops->fault() return value.
 * See filemap_fault() and __lock_page_retry().
 */
2787
static int __do_fault(struct vm_area_struct *vma, unsigned long address,
2788
			pgoff_t pgoff, unsigned int flags,
2789 2790
			struct page *cow_page, struct page **page,
			void **entry)
2791 2792 2793 2794 2795 2796 2797 2798
{
	struct vm_fault vmf;
	int ret;

	vmf.virtual_address = (void __user *)(address & PAGE_MASK);
	vmf.pgoff = pgoff;
	vmf.flags = flags;
	vmf.page = NULL;
2799
	vmf.gfp_mask = __get_fault_gfp_mask(vma);
2800
	vmf.cow_page = cow_page;
2801 2802 2803 2804

	ret = vma->vm_ops->fault(vma, &vmf);
	if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE | VM_FAULT_RETRY)))
		return ret;
2805 2806 2807 2808
	if (ret & VM_FAULT_DAX_LOCKED) {
		*entry = vmf.entry;
		return ret;
	}
2809 2810 2811 2812

	if (unlikely(PageHWPoison(vmf.page))) {
		if (ret & VM_FAULT_LOCKED)
			unlock_page(vmf.page);
2813
		put_page(vmf.page);
2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825
		return VM_FAULT_HWPOISON;
	}

	if (unlikely(!(ret & VM_FAULT_LOCKED)))
		lock_page(vmf.page);
	else
		VM_BUG_ON_PAGE(!PageLocked(vmf.page), vmf.page);

	*page = vmf.page;
	return ret;
}

2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841
/**
 * do_set_pte - setup new PTE entry for given page and add reverse page mapping.
 *
 * @vma: virtual memory area
 * @address: user virtual address
 * @page: page to map
 * @pte: pointer to target page table entry
 * @write: true, if new entry is writable
 * @anon: true, if it's anonymous page
 *
 * Caller must hold page table lock relevant for @pte.
 *
 * Target users are page handler itself and implementations of
 * vm_ops->map_pages.
 */
void do_set_pte(struct vm_area_struct *vma, unsigned long address,
2842 2843 2844 2845 2846 2847 2848 2849 2850 2851
		struct page *page, pte_t *pte, bool write, bool anon)
{
	pte_t entry;

	flush_icache_page(vma, page);
	entry = mk_pte(page, vma->vm_page_prot);
	if (write)
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
	if (anon) {
		inc_mm_counter_fast(vma->vm_mm, MM_ANONPAGES);
2852
		page_add_new_anon_rmap(page, vma, address, false);
2853
	} else {
2854
		inc_mm_counter_fast(vma->vm_mm, mm_counter_file(page));
2855 2856 2857 2858 2859 2860 2861 2862
		page_add_file_rmap(page);
	}
	set_pte_at(vma->vm_mm, address, pte, entry);

	/* no need to invalidate: a not-present page won't be cached */
	update_mmu_cache(vma, address, pte);
}

2863 2864
static unsigned long fault_around_bytes __read_mostly =
	rounddown_pow_of_two(65536);
2865 2866 2867

#ifdef CONFIG_DEBUG_FS
static int fault_around_bytes_get(void *data, u64 *val)
2868
{
2869
	*val = fault_around_bytes;
2870 2871 2872
	return 0;
}

2873 2874 2875 2876 2877
/*
 * fault_around_pages() and fault_around_mask() expects fault_around_bytes
 * rounded down to nearest page order. It's what do_fault_around() expects to
 * see.
 */
2878
static int fault_around_bytes_set(void *data, u64 val)
2879
{
2880
	if (val / PAGE_SIZE > PTRS_PER_PTE)
2881
		return -EINVAL;
2882 2883 2884 2885
	if (val > PAGE_SIZE)
		fault_around_bytes = rounddown_pow_of_two(val);
	else
		fault_around_bytes = PAGE_SIZE; /* rounddown_pow_of_two(0) is undefined */
2886 2887
	return 0;
}
2888 2889
DEFINE_SIMPLE_ATTRIBUTE(fault_around_bytes_fops,
		fault_around_bytes_get, fault_around_bytes_set, "%llu\n");
2890 2891 2892 2893 2894

static int __init fault_around_debugfs(void)
{
	void *ret;

2895 2896
	ret = debugfs_create_file("fault_around_bytes", 0644, NULL, NULL,
			&fault_around_bytes_fops);
2897
	if (!ret)
2898
		pr_warn("Failed to create fault_around_bytes in debugfs");
2899 2900 2901 2902
	return 0;
}
late_initcall(fault_around_debugfs);
#endif
2903

2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926
/*
 * do_fault_around() tries to map few pages around the fault address. The hope
 * is that the pages will be needed soon and this will lower the number of
 * faults to handle.
 *
 * It uses vm_ops->map_pages() to map the pages, which skips the page if it's
 * not ready to be mapped: not up-to-date, locked, etc.
 *
 * This function is called with the page table lock taken. In the split ptlock
 * case the page table lock only protects only those entries which belong to
 * the page table corresponding to the fault address.
 *
 * This function doesn't cross the VMA boundaries, in order to call map_pages()
 * only once.
 *
 * fault_around_pages() defines how many pages we'll try to map.
 * do_fault_around() expects it to return a power of two less than or equal to
 * PTRS_PER_PTE.
 *
 * The virtual address of the area that we map is naturally aligned to the
 * fault_around_pages() value (and therefore to page order).  This way it's
 * easier to guarantee that we don't cross page table boundaries.
 */
2927 2928 2929
static void do_fault_around(struct vm_area_struct *vma, unsigned long address,
		pte_t *pte, pgoff_t pgoff, unsigned int flags)
{
2930
	unsigned long start_addr, nr_pages, mask;
2931 2932 2933 2934
	pgoff_t max_pgoff;
	struct vm_fault vmf;
	int off;

2935
	nr_pages = READ_ONCE(fault_around_bytes) >> PAGE_SHIFT;
2936 2937 2938
	mask = ~(nr_pages * PAGE_SIZE - 1) & PAGE_MASK;

	start_addr = max(address & mask, vma->vm_start);
2939 2940 2941 2942 2943 2944
	off = ((address - start_addr) >> PAGE_SHIFT) & (PTRS_PER_PTE - 1);
	pte -= off;
	pgoff -= off;

	/*
	 *  max_pgoff is either end of page table or end of vma
2945
	 *  or fault_around_pages() from pgoff, depending what is nearest.
2946 2947 2948 2949
	 */
	max_pgoff = pgoff - ((start_addr >> PAGE_SHIFT) & (PTRS_PER_PTE - 1)) +
		PTRS_PER_PTE - 1;
	max_pgoff = min3(max_pgoff, vma_pages(vma) + vma->vm_pgoff - 1,
2950
			pgoff + nr_pages - 1);
2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966

	/* Check if it makes any sense to call ->map_pages */
	while (!pte_none(*pte)) {
		if (++pgoff > max_pgoff)
			return;
		start_addr += PAGE_SIZE;
		if (start_addr >= vma->vm_end)
			return;
		pte++;
	}

	vmf.virtual_address = (void __user *) start_addr;
	vmf.pte = pte;
	vmf.pgoff = pgoff;
	vmf.max_pgoff = max_pgoff;
	vmf.flags = flags;
2967
	vmf.gfp_mask = __get_fault_gfp_mask(vma);
2968 2969 2970
	vma->vm_ops->map_pages(vma, &vmf);
}

2971 2972 2973 2974 2975 2976
static int do_read_fault(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pmd_t *pmd,
		pgoff_t pgoff, unsigned int flags, pte_t orig_pte)
{
	struct page *fault_page;
	spinlock_t *ptl;
2977
	pte_t *pte;
2978 2979 2980 2981 2982 2983 2984
	int ret = 0;

	/*
	 * Let's call ->map_pages() first and use ->fault() as fallback
	 * if page by the offset is not ready to be mapped (cold cache or
	 * something).
	 */
2985
	if (vma->vm_ops->map_pages && fault_around_bytes >> PAGE_SHIFT > 1) {
2986 2987 2988 2989 2990 2991
		pte = pte_offset_map_lock(mm, pmd, address, &ptl);
		do_fault_around(vma, address, pte, pgoff, flags);
		if (!pte_same(*pte, orig_pte))
			goto unlock_out;
		pte_unmap_unlock(pte, ptl);
	}
2992

2993
	ret = __do_fault(vma, address, pgoff, flags, NULL, &fault_page, NULL);
2994 2995 2996 2997 2998 2999 3000
	if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE | VM_FAULT_RETRY)))
		return ret;

	pte = pte_offset_map_lock(mm, pmd, address, &ptl);
	if (unlikely(!pte_same(*pte, orig_pte))) {
		pte_unmap_unlock(pte, ptl);
		unlock_page(fault_page);
3001
		put_page(fault_page);
3002 3003
		return ret;
	}
3004
	do_set_pte(vma, address, fault_page, pte, false, false);
3005
	unlock_page(fault_page);
3006 3007
unlock_out:
	pte_unmap_unlock(pte, ptl);
3008 3009 3010
	return ret;
}

3011 3012 3013 3014 3015
static int do_cow_fault(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pmd_t *pmd,
		pgoff_t pgoff, unsigned int flags, pte_t orig_pte)
{
	struct page *fault_page, *new_page;
3016
	void *fault_entry;
3017
	struct mem_cgroup *memcg;
3018
	spinlock_t *ptl;
3019
	pte_t *pte;
3020 3021 3022 3023 3024 3025 3026 3027 3028
	int ret;

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

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

3029
	if (mem_cgroup_try_charge(new_page, mm, GFP_KERNEL, &memcg, false)) {
3030
		put_page(new_page);
3031 3032 3033
		return VM_FAULT_OOM;
	}

3034 3035
	ret = __do_fault(vma, address, pgoff, flags, new_page, &fault_page,
			 &fault_entry);
3036 3037 3038
	if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE | VM_FAULT_RETRY)))
		goto uncharge_out;

3039
	if (!(ret & VM_FAULT_DAX_LOCKED))
3040
		copy_user_highpage(new_page, fault_page, address, vma);
3041 3042 3043 3044 3045
	__SetPageUptodate(new_page);

	pte = pte_offset_map_lock(mm, pmd, address, &ptl);
	if (unlikely(!pte_same(*pte, orig_pte))) {
		pte_unmap_unlock(pte, ptl);
3046
		if (!(ret & VM_FAULT_DAX_LOCKED)) {
3047
			unlock_page(fault_page);
3048
			put_page(fault_page);
3049
		} else {
3050 3051
			dax_unlock_mapping_entry(vma->vm_file->f_mapping,
						 pgoff);
3052
		}
3053 3054
		goto uncharge_out;
	}
3055
	do_set_pte(vma, address, new_page, pte, true, true);
3056
	mem_cgroup_commit_charge(new_page, memcg, false, false);
3057
	lru_cache_add_active_or_unevictable(new_page, vma);
3058
	pte_unmap_unlock(pte, ptl);
3059
	if (!(ret & VM_FAULT_DAX_LOCKED)) {
3060
		unlock_page(fault_page);
3061
		put_page(fault_page);
3062
	} else {
3063
		dax_unlock_mapping_entry(vma->vm_file->f_mapping, pgoff);
3064
	}
3065 3066
	return ret;
uncharge_out:
3067
	mem_cgroup_cancel_charge(new_page, memcg, false);
3068
	put_page(new_page);
3069 3070 3071
	return ret;
}

3072
static int do_shared_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3073
		unsigned long address, pmd_t *pmd,
3074
		pgoff_t pgoff, unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
3075
{
3076 3077
	struct page *fault_page;
	struct address_space *mapping;
3078
	spinlock_t *ptl;
3079
	pte_t *pte;
3080 3081
	int dirtied = 0;
	int ret, tmp;
3082

3083
	ret = __do_fault(vma, address, pgoff, flags, NULL, &fault_page, NULL);
3084
	if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE | VM_FAULT_RETRY)))
3085
		return ret;
L
Linus Torvalds 已提交
3086 3087

	/*
3088 3089
	 * Check if the backing address space wants to know that the page is
	 * about to become writable
L
Linus Torvalds 已提交
3090
	 */
3091 3092 3093 3094 3095
	if (vma->vm_ops->page_mkwrite) {
		unlock_page(fault_page);
		tmp = do_page_mkwrite(vma, fault_page, address);
		if (unlikely(!tmp ||
				(tmp & (VM_FAULT_ERROR | VM_FAULT_NOPAGE)))) {
3096
			put_page(fault_page);
3097
			return tmp;
3098
		}
3099 3100
	}

3101 3102 3103 3104
	pte = pte_offset_map_lock(mm, pmd, address, &ptl);
	if (unlikely(!pte_same(*pte, orig_pte))) {
		pte_unmap_unlock(pte, ptl);
		unlock_page(fault_page);
3105
		put_page(fault_page);
3106
		return ret;
L
Linus Torvalds 已提交
3107
	}
3108
	do_set_pte(vma, address, fault_page, pte, true, false);
3109
	pte_unmap_unlock(pte, ptl);
3110

3111 3112
	if (set_page_dirty(fault_page))
		dirtied = 1;
3113 3114 3115 3116 3117 3118
	/*
	 * Take a local copy of the address_space - page.mapping may be zeroed
	 * by truncate after unlock_page().   The address_space itself remains
	 * pinned by vma->vm_file's reference.  We rely on unlock_page()'s
	 * release semantics to prevent the compiler from undoing this copying.
	 */
3119
	mapping = page_rmapping(fault_page);
3120 3121 3122 3123 3124 3125 3126
	unlock_page(fault_page);
	if ((dirtied || vma->vm_ops->page_mkwrite) && mapping) {
		/*
		 * Some device drivers do not set page.mapping but still
		 * dirty their pages
		 */
		balance_dirty_pages_ratelimited(mapping);
3127
	}
3128

3129
	if (!vma->vm_ops->page_mkwrite)
3130
		file_update_time(vma->vm_file);
3131

3132
	return ret;
3133
}
3134

3135 3136 3137 3138 3139 3140
/*
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
 * but allow concurrent faults).
 * The mmap_sem may have been released depending on flags and our
 * return value.  See filemap_fault() and __lock_page_or_retry().
 */
3141
static int do_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3142
		unsigned long address, pte_t *page_table, pmd_t *pmd,
3143
		unsigned int flags, pte_t orig_pte)
3144
{
3145
	pgoff_t pgoff = linear_page_index(vma, address);
3146

3147
	pte_unmap(page_table);
3148 3149 3150
	/* The VMA was not fully populated on mmap() or missing VM_DONTEXPAND */
	if (!vma->vm_ops->fault)
		return VM_FAULT_SIGBUS;
3151 3152 3153
	if (!(flags & FAULT_FLAG_WRITE))
		return do_read_fault(mm, vma, address, pmd, pgoff, flags,
				orig_pte);
3154 3155 3156
	if (!(vma->vm_flags & VM_SHARED))
		return do_cow_fault(mm, vma, address, pmd, pgoff, flags,
				orig_pte);
3157
	return do_shared_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
3158 3159
}

3160
static int numa_migrate_prep(struct page *page, struct vm_area_struct *vma,
3161 3162
				unsigned long addr, int page_nid,
				int *flags)
3163 3164 3165 3166
{
	get_page(page);

	count_vm_numa_event(NUMA_HINT_FAULTS);
3167
	if (page_nid == numa_node_id()) {
3168
		count_vm_numa_event(NUMA_HINT_FAULTS_LOCAL);
3169 3170
		*flags |= TNF_FAULT_LOCAL;
	}
3171 3172 3173 3174

	return mpol_misplaced(page, vma, addr);
}

3175
static int do_numa_page(struct mm_struct *mm, struct vm_area_struct *vma,
3176 3177
		   unsigned long addr, pte_t pte, pte_t *ptep, pmd_t *pmd)
{
3178
	struct page *page = NULL;
3179
	spinlock_t *ptl;
3180
	int page_nid = -1;
3181
	int last_cpupid;
3182
	int target_nid;
3183
	bool migrated = false;
3184
	bool was_writable = pte_write(pte);
3185
	int flags = 0;
3186

3187 3188 3189
	/* A PROT_NONE fault should not end up here */
	BUG_ON(!(vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE)));

3190 3191 3192 3193 3194
	/*
	* The "pte" at this point cannot be used safely without
	* validation through pte_unmap_same(). It's of NUMA type but
	* the pfn may be screwed if the read is non atomic.
	*
3195 3196 3197
	* We can safely just do a "set_pte_at()", because the old
	* page table entry is not accessible, so there would be no
	* concurrent hardware modifications to the PTE.
3198 3199 3200
	*/
	ptl = pte_lockptr(mm, pmd);
	spin_lock(ptl);
3201 3202 3203 3204 3205
	if (unlikely(!pte_same(*ptep, pte))) {
		pte_unmap_unlock(ptep, ptl);
		goto out;
	}

3206 3207 3208
	/* Make it present again */
	pte = pte_modify(pte, vma->vm_page_prot);
	pte = pte_mkyoung(pte);
3209 3210
	if (was_writable)
		pte = pte_mkwrite(pte);
3211 3212 3213 3214 3215 3216 3217 3218 3219
	set_pte_at(mm, addr, ptep, pte);
	update_mmu_cache(vma, addr, ptep);

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

3220 3221 3222 3223 3224 3225
	/* TODO: handle PTE-mapped THP */
	if (PageCompound(page)) {
		pte_unmap_unlock(ptep, ptl);
		return 0;
	}

3226
	/*
3227 3228 3229 3230 3231 3232
	 * Avoid grouping on RO pages in general. RO pages shouldn't hurt as
	 * much anyway since they can be in shared cache state. This misses
	 * the case where a mapping is writable but the process never writes
	 * to it but pte_write gets cleared during protection updates and
	 * pte_dirty has unpredictable behaviour between PTE scan updates,
	 * background writeback, dirty balancing and application behaviour.
3233
	 */
3234
	if (!(vma->vm_flags & VM_WRITE))
3235 3236
		flags |= TNF_NO_GROUP;

3237 3238 3239 3240 3241 3242 3243
	/*
	 * Flag if the page is shared between multiple address spaces. This
	 * is later used when determining whether to group tasks together
	 */
	if (page_mapcount(page) > 1 && (vma->vm_flags & VM_SHARED))
		flags |= TNF_SHARED;

3244
	last_cpupid = page_cpupid_last(page);
3245
	page_nid = page_to_nid(page);
3246
	target_nid = numa_migrate_prep(page, vma, addr, page_nid, &flags);
3247
	pte_unmap_unlock(ptep, ptl);
3248 3249 3250 3251 3252 3253
	if (target_nid == -1) {
		put_page(page);
		goto out;
	}

	/* Migrate to the requested node */
3254
	migrated = migrate_misplaced_page(page, vma, target_nid);
3255
	if (migrated) {
3256
		page_nid = target_nid;
3257
		flags |= TNF_MIGRATED;
3258 3259
	} else
		flags |= TNF_MIGRATE_FAIL;
3260 3261

out:
3262
	if (page_nid != -1)
3263
		task_numa_fault(last_cpupid, page_nid, 1, flags);
3264 3265 3266
	return 0;
}

3267 3268 3269
static int create_huge_pmd(struct mm_struct *mm, struct vm_area_struct *vma,
			unsigned long address, pmd_t *pmd, unsigned int flags)
{
3270
	if (vma_is_anonymous(vma))
3271 3272 3273 3274 3275 3276 3277 3278 3279 3280
		return do_huge_pmd_anonymous_page(mm, vma, address, pmd, flags);
	if (vma->vm_ops->pmd_fault)
		return vma->vm_ops->pmd_fault(vma, address, pmd, flags);
	return VM_FAULT_FALLBACK;
}

static int wp_huge_pmd(struct mm_struct *mm, struct vm_area_struct *vma,
			unsigned long address, pmd_t *pmd, pmd_t orig_pmd,
			unsigned int flags)
{
3281
	if (vma_is_anonymous(vma))
3282 3283 3284 3285 3286 3287
		return do_huge_pmd_wp_page(mm, vma, address, pmd, orig_pmd);
	if (vma->vm_ops->pmd_fault)
		return vma->vm_ops->pmd_fault(vma, address, pmd, flags);
	return VM_FAULT_FALLBACK;
}

L
Linus Torvalds 已提交
3288 3289 3290 3291 3292 3293 3294 3295 3296
/*
 * 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).
 *
3297 3298
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
 * but allow concurrent faults), and pte mapped but not yet locked.
3299 3300 3301 3302
 * We return with pte unmapped and unlocked.
 *
 * The mmap_sem may have been released depending on flags and our
 * return value.  See filemap_fault() and __lock_page_or_retry().
L
Linus Torvalds 已提交
3303
 */
3304
static int handle_pte_fault(struct mm_struct *mm,
3305 3306
		     struct vm_area_struct *vma, unsigned long address,
		     pte_t *pte, pmd_t *pmd, unsigned int flags)
L
Linus Torvalds 已提交
3307 3308
{
	pte_t entry;
3309
	spinlock_t *ptl;
L
Linus Torvalds 已提交
3310

3311 3312 3313 3314 3315 3316 3317 3318 3319 3320
	/*
	 * some architectures can have larger ptes than wordsize,
	 * e.g.ppc44x-defconfig has CONFIG_PTE_64BIT=y and CONFIG_32BIT=y,
	 * so READ_ONCE or ACCESS_ONCE cannot guarantee atomic accesses.
	 * The code below just needs a consistent view for the ifs and
	 * we later double check anyway with the ptl lock held. So here
	 * a barrier will do.
	 */
	entry = *pte;
	barrier();
L
Linus Torvalds 已提交
3321
	if (!pte_present(entry)) {
3322
		if (pte_none(entry)) {
3323 3324 3325 3326
			if (vma_is_anonymous(vma))
				return do_anonymous_page(mm, vma, address,
							 pte, pmd, flags);
			else
3327 3328
				return do_fault(mm, vma, address, pte, pmd,
						flags, entry);
3329 3330
		}
		return do_swap_page(mm, vma, address,
3331
					pte, pmd, flags, entry);
L
Linus Torvalds 已提交
3332 3333
	}

3334
	if (pte_protnone(entry))
3335 3336
		return do_numa_page(mm, vma, address, entry, pte, pmd);

3337
	ptl = pte_lockptr(mm, pmd);
3338 3339 3340
	spin_lock(ptl);
	if (unlikely(!pte_same(*pte, entry)))
		goto unlock;
3341
	if (flags & FAULT_FLAG_WRITE) {
L
Linus Torvalds 已提交
3342
		if (!pte_write(entry))
3343 3344
			return do_wp_page(mm, vma, address,
					pte, pmd, ptl, entry);
L
Linus Torvalds 已提交
3345 3346 3347
		entry = pte_mkdirty(entry);
	}
	entry = pte_mkyoung(entry);
3348
	if (ptep_set_access_flags(vma, address, pte, entry, flags & FAULT_FLAG_WRITE)) {
3349
		update_mmu_cache(vma, address, pte);
3350 3351 3352 3353 3354 3355 3356
	} 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.
		 */
3357
		if (flags & FAULT_FLAG_WRITE)
3358
			flush_tlb_fix_spurious_fault(vma, address);
3359
	}
3360 3361
unlock:
	pte_unmap_unlock(pte, ptl);
N
Nick Piggin 已提交
3362
	return 0;
L
Linus Torvalds 已提交
3363 3364 3365 3366
}

/*
 * By the time we get here, we already hold the mm semaphore
3367 3368 3369
 *
 * The mmap_sem may have been released depending on flags and our
 * return value.  See filemap_fault() and __lock_page_or_retry().
L
Linus Torvalds 已提交
3370
 */
3371 3372
static int __handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
			     unsigned long address, unsigned int flags)
L
Linus Torvalds 已提交
3373 3374 3375 3376 3377 3378
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

3379
	if (!arch_vma_access_permitted(vma, flags & FAULT_FLAG_WRITE,
3380
					    flags & FAULT_FLAG_INSTRUCTION,
3381
					    flags & FAULT_FLAG_REMOTE))
3382 3383
		return VM_FAULT_SIGSEGV;

3384
	if (unlikely(is_vm_hugetlb_page(vma)))
3385
		return hugetlb_fault(mm, vma, address, flags);
L
Linus Torvalds 已提交
3386 3387 3388 3389

	pgd = pgd_offset(mm, address);
	pud = pud_alloc(mm, pgd, address);
	if (!pud)
3390
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
3391 3392
	pmd = pmd_alloc(mm, pud, address);
	if (!pmd)
3393
		return VM_FAULT_OOM;
3394
	if (pmd_none(*pmd) && transparent_hugepage_enabled(vma)) {
3395
		int ret = create_huge_pmd(mm, vma, address, pmd, flags);
3396 3397
		if (!(ret & VM_FAULT_FALLBACK))
			return ret;
3398 3399
	} else {
		pmd_t orig_pmd = *pmd;
3400 3401
		int ret;

3402
		barrier();
3403
		if (pmd_trans_huge(orig_pmd) || pmd_devmap(orig_pmd)) {
3404 3405
			unsigned int dirty = flags & FAULT_FLAG_WRITE;

3406
			if (pmd_protnone(orig_pmd))
3407
				return do_huge_pmd_numa_page(mm, vma, address,
3408 3409
							     orig_pmd, pmd);

3410
			if (dirty && !pmd_write(orig_pmd)) {
3411 3412
				ret = wp_huge_pmd(mm, vma, address, pmd,
							orig_pmd, flags);
3413 3414
				if (!(ret & VM_FAULT_FALLBACK))
					return ret;
3415 3416 3417
			} else {
				huge_pmd_set_accessed(mm, vma, address, pmd,
						      orig_pmd, dirty);
3418
				return 0;
3419
			}
3420 3421 3422 3423
		}
	}

	/*
3424
	 * Use pte_alloc() instead of pte_alloc_map, because we can't
3425 3426 3427
	 * run pte_offset_map on the pmd, if an huge pmd could
	 * materialize from under us from a different thread.
	 */
3428
	if (unlikely(pte_alloc(mm, pmd, address)))
3429
		return VM_FAULT_OOM;
3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441
	/*
	 * If a huge pmd materialized under us just retry later.  Use
	 * pmd_trans_unstable() instead of pmd_trans_huge() to ensure the pmd
	 * didn't become pmd_trans_huge under us and then back to pmd_none, as
	 * a result of MADV_DONTNEED running immediately after a huge pmd fault
	 * in a different thread of this mm, in turn leading to a misleading
	 * pmd_trans_huge() retval.  All we have to ensure is that it is a
	 * regular pmd that we can walk with pte_offset_map() and we can do that
	 * through an atomic read in C, which is what pmd_trans_unstable()
	 * provides.
	 */
	if (unlikely(pmd_trans_unstable(pmd) || pmd_devmap(*pmd)))
3442 3443 3444 3445 3446 3447 3448 3449
		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
3450

3451
	return handle_pte_fault(mm, vma, address, pte, pmd, flags);
L
Linus Torvalds 已提交
3452 3453
}

3454 3455 3456 3457 3458 3459
/*
 * By the time we get here, we already hold the mm semaphore
 *
 * The mmap_sem may have been released depending on flags and our
 * return value.  See filemap_fault() and __lock_page_or_retry().
 */
3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477
int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
		    unsigned long address, unsigned int flags)
{
	int ret;

	__set_current_state(TASK_RUNNING);

	count_vm_event(PGFAULT);
	mem_cgroup_count_vm_event(mm, PGFAULT);

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

	/*
	 * Enable the memcg OOM handling for faults triggered in user
	 * space.  Kernel faults are handled more gracefully.
	 */
	if (flags & FAULT_FLAG_USER)
3478
		mem_cgroup_oom_enable();
3479 3480 3481

	ret = __handle_mm_fault(mm, vma, address, flags);

3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492
	if (flags & FAULT_FLAG_USER) {
		mem_cgroup_oom_disable();
                /*
                 * The task may have entered a memcg OOM situation but
                 * if the allocation error was handled gracefully (no
                 * VM_FAULT_OOM), there is no need to kill anything.
                 * Just clean up the OOM state peacefully.
                 */
                if (task_in_memcg_oom(current) && !(ret & VM_FAULT_OOM))
                        mem_cgroup_oom_synchronize(false);
	}
3493

3494 3495
	return ret;
}
3496
EXPORT_SYMBOL_GPL(handle_mm_fault);
3497

L
Linus Torvalds 已提交
3498 3499 3500
#ifndef __PAGETABLE_PUD_FOLDED
/*
 * Allocate page upper directory.
3501
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
3502
 */
3503
int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
L
Linus Torvalds 已提交
3504
{
3505 3506
	pud_t *new = pud_alloc_one(mm, address);
	if (!new)
3507
		return -ENOMEM;
L
Linus Torvalds 已提交
3508

3509 3510
	smp_wmb(); /* See comment in __pte_alloc */

3511
	spin_lock(&mm->page_table_lock);
3512
	if (pgd_present(*pgd))		/* Another has populated it */
3513
		pud_free(mm, new);
3514 3515
	else
		pgd_populate(mm, pgd, new);
3516
	spin_unlock(&mm->page_table_lock);
3517
	return 0;
L
Linus Torvalds 已提交
3518 3519 3520 3521 3522 3523
}
#endif /* __PAGETABLE_PUD_FOLDED */

#ifndef __PAGETABLE_PMD_FOLDED
/*
 * Allocate page middle directory.
3524
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
3525
 */
3526
int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
L
Linus Torvalds 已提交
3527
{
3528 3529
	pmd_t *new = pmd_alloc_one(mm, address);
	if (!new)
3530
		return -ENOMEM;
L
Linus Torvalds 已提交
3531

3532 3533
	smp_wmb(); /* See comment in __pte_alloc */

3534
	spin_lock(&mm->page_table_lock);
L
Linus Torvalds 已提交
3535
#ifndef __ARCH_HAS_4LEVEL_HACK
3536 3537
	if (!pud_present(*pud)) {
		mm_inc_nr_pmds(mm);
3538
		pud_populate(mm, pud, new);
3539
	} else	/* Another has populated it */
3540
		pmd_free(mm, new);
3541 3542 3543
#else
	if (!pgd_present(*pud)) {
		mm_inc_nr_pmds(mm);
3544
		pgd_populate(mm, pud, new);
3545 3546
	} else /* Another has populated it */
		pmd_free(mm, new);
L
Linus Torvalds 已提交
3547
#endif /* __ARCH_HAS_4LEVEL_HACK */
3548
	spin_unlock(&mm->page_table_lock);
3549
	return 0;
3550
}
L
Linus Torvalds 已提交
3551 3552
#endif /* __PAGETABLE_PMD_FOLDED */

3553
static int __follow_pte(struct mm_struct *mm, unsigned long address,
3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569
		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);
3570
	VM_BUG_ON(pmd_trans_huge(*pmd));
3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590
	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;
}

3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601
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;
}

3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630
/**
 * 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);

3631
#ifdef CONFIG_HAVE_IOREMAP_PROT
3632 3633 3634
int follow_phys(struct vm_area_struct *vma,
		unsigned long address, unsigned int flags,
		unsigned long *prot, resource_size_t *phys)
3635
{
3636
	int ret = -EINVAL;
3637 3638 3639
	pte_t *ptep, pte;
	spinlock_t *ptl;

3640 3641
	if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
		goto out;
3642

3643
	if (follow_pte(vma->vm_mm, address, &ptep, &ptl))
3644
		goto out;
3645
	pte = *ptep;
3646

3647 3648 3649 3650
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;

	*prot = pgprot_val(pte_pgprot(pte));
3651
	*phys = (resource_size_t)pte_pfn(pte) << PAGE_SHIFT;
3652

3653
	ret = 0;
3654 3655 3656
unlock:
	pte_unmap_unlock(ptep, ptl);
out:
3657
	return ret;
3658 3659 3660 3661 3662 3663 3664
}

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
3665
	void __iomem *maddr;
3666 3667
	int offset = addr & (PAGE_SIZE-1);

3668
	if (follow_phys(vma, addr, write, &prot, &phys_addr))
3669 3670
		return -EINVAL;

3671
	maddr = ioremap_prot(phys_addr, PAGE_ALIGN(len + offset), prot);
3672 3673 3674 3675 3676 3677 3678 3679
	if (write)
		memcpy_toio(maddr + offset, buf, len);
	else
		memcpy_fromio(buf, maddr + offset, len);
	iounmap(maddr);

	return len;
}
3680
EXPORT_SYMBOL_GPL(generic_access_phys);
3681 3682
#endif

3683
/*
3684 3685
 * Access another process' address space as given in mm.  If non-NULL, use the
 * given task for page fault accounting.
3686
 */
3687 3688
static int __access_remote_vm(struct task_struct *tsk, struct mm_struct *mm,
		unsigned long addr, void *buf, int len, int write)
3689 3690 3691 3692 3693
{
	struct vm_area_struct *vma;
	void *old_buf = buf;

	down_read(&mm->mmap_sem);
S
Simon Arlott 已提交
3694
	/* ignore errors, just check how much was successfully transferred */
3695 3696 3697
	while (len) {
		int bytes, ret, offset;
		void *maddr;
3698
		struct page *page = NULL;
3699

3700
		ret = get_user_pages_remote(tsk, mm, addr, 1,
3701
				write, 1, &page, &vma);
3702
		if (ret <= 0) {
3703 3704 3705
#ifndef CONFIG_HAVE_IOREMAP_PROT
			break;
#else
3706 3707 3708 3709 3710
			/*
			 * Check if this is a VM_IO | VM_PFNMAP VMA, which
			 * we can access using slightly different code.
			 */
			vma = find_vma(mm, addr);
3711
			if (!vma || vma->vm_start > addr)
3712 3713 3714 3715 3716 3717 3718
				break;
			if (vma->vm_ops && vma->vm_ops->access)
				ret = vma->vm_ops->access(vma, addr, buf,
							  len, write);
			if (ret <= 0)
				break;
			bytes = ret;
3719
#endif
3720
		} else {
3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735
			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);
3736
			put_page(page);
3737 3738 3739 3740 3741 3742 3743 3744 3745
		}
		len -= bytes;
		buf += bytes;
		addr += bytes;
	}
	up_read(&mm->mmap_sem);

	return buf - old_buf;
}
3746

3747
/**
3748
 * access_remote_vm - access another process' address space
3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762
 * @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);
}

3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783
/*
 * 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;
}

3784 3785 3786 3787 3788 3789 3790 3791
/*
 * 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;

3792 3793 3794 3795 3796 3797 3798
	/*
	 * Do not print if we are in atomic
	 * contexts (in exception stacks, etc.):
	 */
	if (preempt_count())
		return;

3799 3800 3801 3802 3803 3804
	down_read(&mm->mmap_sem);
	vma = find_vma(mm, ip);
	if (vma && vma->vm_file) {
		struct file *f = vma->vm_file;
		char *buf = (char *)__get_free_page(GFP_KERNEL);
		if (buf) {
A
Andy Shevchenko 已提交
3805
			char *p;
3806

3807
			p = file_path(f, buf, PAGE_SIZE);
3808 3809
			if (IS_ERR(p))
				p = "?";
A
Andy Shevchenko 已提交
3810
			printk("%s%s[%lx+%lx]", prefix, kbasename(p),
3811 3812 3813 3814 3815
					vma->vm_start,
					vma->vm_end - vma->vm_start);
			free_page((unsigned long)buf);
		}
	}
3816
	up_read(&mm->mmap_sem);
3817
}
3818

3819
#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_DEBUG_ATOMIC_SLEEP)
3820
void __might_fault(const char *file, int line)
3821
{
3822 3823 3824 3825 3826 3827 3828 3829
	/*
	 * 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;
3830
	if (pagefault_disabled())
3831
		return;
3832 3833
	__might_sleep(file, line, 0);
#if defined(CONFIG_DEBUG_ATOMIC_SLEEP)
3834
	if (current->mm)
3835
		might_lock_read(&current->mm->mmap_sem);
3836
#endif
3837
}
3838
EXPORT_SYMBOL(__might_fault);
3839
#endif
3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910

#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 */
3911

3912
#if USE_SPLIT_PTE_PTLOCKS && ALLOC_SPLIT_PTLOCKS
3913 3914 3915 3916 3917 3918 3919 3920 3921

static struct kmem_cache *page_ptl_cachep;

void __init ptlock_cache_init(void)
{
	page_ptl_cachep = kmem_cache_create("page->ptl", sizeof(spinlock_t), 0,
			SLAB_PANIC, NULL);
}

3922
bool ptlock_alloc(struct page *page)
3923 3924 3925
{
	spinlock_t *ptl;

3926
	ptl = kmem_cache_alloc(page_ptl_cachep, GFP_KERNEL);
3927 3928
	if (!ptl)
		return false;
3929
	page->ptl = ptl;
3930 3931 3932
	return true;
}

3933
void ptlock_free(struct page *page)
3934
{
3935
	kmem_cache_free(page_ptl_cachep, page->ptl);
3936 3937
}
#endif