memory.c 102.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
/*
 *  linux/mm/memory.c
 *
 *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 */

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

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

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

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

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

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

68 69
#include "internal.h"

A
Andy Whitcroft 已提交
70
#ifndef CONFIG_NEED_MULTIPLE_NODES
L
Linus Torvalds 已提交
71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91
/* use the per-pgdat data instead for discontigmem - mbligh */
unsigned long max_mapnr;
struct page *mem_map;

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

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

EXPORT_SYMBOL(num_physpages);
EXPORT_SYMBOL(high_memory);

92 93 94 95 96 97 98 99 100 101 102 103
/*
 * Randomize the address space (stacks, mmaps, brk, etc.).
 *
 * ( When CONFIG_COMPAT_BRK=y we exclude brk from randomization,
 *   as ancient (libc5 based) binaries can segfault. )
 */
int randomize_va_space __read_mostly =
#ifdef CONFIG_COMPAT_BRK
					1;
#else
					2;
#endif
104 105 106 107

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

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

/*
 * CONFIG_MMU architectures set up ZERO_PAGE in their paging_init()
 */
static int __init init_zero_pfn(void)
{
	zero_pfn = page_to_pfn(ZERO_PAGE(0));
	return 0;
}
core_initcall(init_zero_pfn);
124

K
KAMEZAWA Hiroyuki 已提交
125

126 127
#if defined(SPLIT_RSS_COUNTING)

128
static void __sync_task_rss_stat(struct task_struct *task, struct mm_struct *mm)
129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
{
	int i;

	for (i = 0; i < NR_MM_COUNTERS; i++) {
		if (task->rss_stat.count[i]) {
			add_mm_counter(mm, i, task->rss_stat.count[i]);
			task->rss_stat.count[i] = 0;
		}
	}
	task->rss_stat.events = 0;
}

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))
		__sync_task_rss_stat(task, task->mm);
}

unsigned long get_mm_counter(struct mm_struct *mm, int member)
{
	long val = 0;

	/*
	 * Don't use task->mm here...for avoiding to use task_get_mm()..
	 * The caller must guarantee task->mm is not invalid.
	 */
	val = atomic_long_read(&mm->rss_stat.count[member]);
	/*
	 * counter is updated in asynchronous manner and may go to minus.
	 * But it's never be expected number for users.
	 */
	if (val < 0)
		return 0;
	return (unsigned long)val;
}

void sync_mm_rss(struct task_struct *task, struct mm_struct *mm)
{
	__sync_task_rss_stat(task, mm);
}
#else

#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)
{
}

#endif

L
Linus Torvalds 已提交
196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223
/*
 * If a p?d_bad entry is found while walking page tables, report
 * the error, before resetting entry to p?d_none.  Usually (but
 * very seldom) called out from the p?d_none_or_clear_bad macros.
 */

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

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

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

/*
 * Note: this doesn't free the actual pages themselves. That
 * has been handled earlier when unmapping all the memory regions.
 */
224 225
static void free_pte_range(struct mmu_gather *tlb, pmd_t *pmd,
			   unsigned long addr)
L
Linus Torvalds 已提交
226
{
227
	pgtable_t token = pmd_pgtable(*pmd);
228
	pmd_clear(pmd);
229
	pte_free_tlb(tlb, token, addr);
230
	tlb->mm->nr_ptes--;
L
Linus Torvalds 已提交
231 232
}

233 234 235
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
236 237 238
{
	pmd_t *pmd;
	unsigned long next;
239
	unsigned long start;
L
Linus Torvalds 已提交
240

241
	start = addr;
L
Linus Torvalds 已提交
242 243 244 245 246
	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
247
		free_pte_range(tlb, pmd, addr);
L
Linus Torvalds 已提交
248 249
	} while (pmd++, addr = next, addr != end);

250 251 252 253 254 255 256
	start &= PUD_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PUD_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
257
	}
258 259 260 261 262
	if (end - 1 > ceiling - 1)
		return;

	pmd = pmd_offset(pud, start);
	pud_clear(pud);
263
	pmd_free_tlb(tlb, pmd, start);
L
Linus Torvalds 已提交
264 265
}

266 267 268
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
269 270 271
{
	pud_t *pud;
	unsigned long next;
272
	unsigned long start;
L
Linus Torvalds 已提交
273

274
	start = addr;
L
Linus Torvalds 已提交
275 276 277 278 279
	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
280
		free_pmd_range(tlb, pud, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
281 282
	} while (pud++, addr = next, addr != end);

283 284 285 286 287 288 289
	start &= PGDIR_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PGDIR_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
290
	}
291 292 293 294 295
	if (end - 1 > ceiling - 1)
		return;

	pud = pud_offset(pgd, start);
	pgd_clear(pgd);
296
	pud_free_tlb(tlb, pud, start);
L
Linus Torvalds 已提交
297 298 299
}

/*
300 301
 * This function frees user-level page tables of a process.
 *
L
Linus Torvalds 已提交
302 303
 * Must be called with pagetable lock held.
 */
304
void free_pgd_range(struct mmu_gather *tlb,
305 306
			unsigned long addr, unsigned long end,
			unsigned long floor, unsigned long ceiling)
L
Linus Torvalds 已提交
307 308 309
{
	pgd_t *pgd;
	unsigned long next;
310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335

	/*
	 * 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
336

337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352
	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;

353
	pgd = pgd_offset(tlb->mm, addr);
L
Linus Torvalds 已提交
354 355 356 357
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
358
		free_pud_range(tlb, pgd, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
359
	} while (pgd++, addr = next, addr != end);
360 361
}

362
void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *vma,
363
		unsigned long floor, unsigned long ceiling)
364 365 366 367 368
{
	while (vma) {
		struct vm_area_struct *next = vma->vm_next;
		unsigned long addr = vma->vm_start;

369
		/*
N
npiggin@suse.de 已提交
370 371
		 * Hide vma from rmap and truncate_pagecache before freeing
		 * pgtables
372
		 */
373
		unlink_anon_vmas(vma);
374 375
		unlink_file_vma(vma);

376
		if (is_vm_hugetlb_page(vma)) {
377
			hugetlb_free_pgd_range(tlb, addr, vma->vm_end,
378
				floor, next? next->vm_start: ceiling);
379 380 381 382 383
		} else {
			/*
			 * Optimization: gather nearby vmas into one call down
			 */
			while (next && next->vm_start <= vma->vm_end + PMD_SIZE
384
			       && !is_vm_hugetlb_page(next)) {
385 386
				vma = next;
				next = vma->vm_next;
387
				unlink_anon_vmas(vma);
388
				unlink_file_vma(vma);
389 390 391 392
			}
			free_pgd_range(tlb, addr, vma->vm_end,
				floor, next? next->vm_start: ceiling);
		}
393 394
		vma = next;
	}
L
Linus Torvalds 已提交
395 396
}

397 398
int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma,
		pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
399
{
400
	pgtable_t new = pte_alloc_one(mm, address);
401
	int wait_split_huge_page;
402 403 404
	if (!new)
		return -ENOMEM;

405 406 407 408 409 410 411 412 413 414 415 416 417 418 419
	/*
	 * Ensure all pte setup (eg. pte page lock and page clearing) are
	 * visible before the pte is made visible to other CPUs by being
	 * put into page tables.
	 *
	 * The other side of the story is the pointer chasing in the page
	 * table walking code (when walking the page table without locking;
	 * ie. most of the time). Fortunately, these data accesses consist
	 * of a chain of data-dependent loads, meaning most CPUs (alpha
	 * being the notable exception) will already guarantee loads are
	 * seen in-order. See the alpha page table accessors for the
	 * smp_read_barrier_depends() barriers in page table walking code.
	 */
	smp_wmb(); /* Could be smp_wmb__xxx(before|after)_spin_lock */

H
Hugh Dickins 已提交
420
	spin_lock(&mm->page_table_lock);
421 422
	wait_split_huge_page = 0;
	if (likely(pmd_none(*pmd))) {	/* Has another populated it ? */
L
Linus Torvalds 已提交
423 424
		mm->nr_ptes++;
		pmd_populate(mm, pmd, new);
425
		new = NULL;
426 427
	} else if (unlikely(pmd_trans_splitting(*pmd)))
		wait_split_huge_page = 1;
H
Hugh Dickins 已提交
428
	spin_unlock(&mm->page_table_lock);
429 430
	if (new)
		pte_free(mm, new);
431 432
	if (wait_split_huge_page)
		wait_split_huge_page(vma->anon_vma, pmd);
433
	return 0;
L
Linus Torvalds 已提交
434 435
}

436
int __pte_alloc_kernel(pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
437
{
438 439 440 441
	pte_t *new = pte_alloc_one_kernel(&init_mm, address);
	if (!new)
		return -ENOMEM;

442 443
	smp_wmb(); /* See comment in __pte_alloc */

444
	spin_lock(&init_mm.page_table_lock);
445
	if (likely(pmd_none(*pmd))) {	/* Has another populated it ? */
446
		pmd_populate_kernel(&init_mm, pmd, new);
447
		new = NULL;
448 449
	} else
		VM_BUG_ON(pmd_trans_splitting(*pmd));
450
	spin_unlock(&init_mm.page_table_lock);
451 452
	if (new)
		pte_free_kernel(&init_mm, new);
453
	return 0;
L
Linus Torvalds 已提交
454 455
}

K
KAMEZAWA Hiroyuki 已提交
456 457 458 459 460 461
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)
462
{
K
KAMEZAWA Hiroyuki 已提交
463 464
	int i;

465 466
	if (current->mm == mm)
		sync_mm_rss(current, mm);
K
KAMEZAWA Hiroyuki 已提交
467 468 469
	for (i = 0; i < NR_MM_COUNTERS; i++)
		if (rss[i])
			add_mm_counter(mm, i, rss[i]);
470 471
}

N
Nick Piggin 已提交
472
/*
473 474 475
 * This function is called to print an error when a bad pte
 * is found. For example, we might have a PFN-mapped pte in
 * a region that doesn't allow it.
N
Nick Piggin 已提交
476 477 478
 *
 * The calling function must still handle the error.
 */
479 480
static void print_bad_pte(struct vm_area_struct *vma, unsigned long addr,
			  pte_t pte, struct page *page)
N
Nick Piggin 已提交
481
{
482 483 484 485 486
	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;
487 488 489 490 491 492 493 494 495 496 497 498 499 500
	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) {
501 502
			printk(KERN_ALERT
				"BUG: Bad page map: %lu messages suppressed\n",
503 504 505 506 507 508 509
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;
510 511 512 513

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

514 515
	printk(KERN_ALERT
		"BUG: Bad page map in process %s  pte:%08llx pmd:%08llx\n",
516 517
		current->comm,
		(long long)pte_val(pte), (long long)pmd_val(*pmd));
518 519
	if (page)
		dump_page(page);
520
	printk(KERN_ALERT
521 522 523 524 525 526
		"addr:%p vm_flags:%08lx anon_vma:%p mapping:%p index:%lx\n",
		(void *)addr, vma->vm_flags, vma->anon_vma, mapping, index);
	/*
	 * Choose text because data symbols depend on CONFIG_KALLSYMS_ALL=y
	 */
	if (vma->vm_ops)
527
		print_symbol(KERN_ALERT "vma->vm_ops->fault: %s\n",
528 529
				(unsigned long)vma->vm_ops->fault);
	if (vma->vm_file && vma->vm_file->f_op)
530
		print_symbol(KERN_ALERT "vma->vm_file->f_op->mmap: %s\n",
531
				(unsigned long)vma->vm_file->f_op->mmap);
N
Nick Piggin 已提交
532
	dump_stack();
533
	add_taint(TAINT_BAD_PAGE);
N
Nick Piggin 已提交
534 535
}

536 537 538 539 540
static inline int is_cow_mapping(unsigned int flags)
{
	return (flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
}

H
Hugh Dickins 已提交
541 542 543 544 545 546 547 548 549 550 551 552 553 554
#ifndef is_zero_pfn
static inline int is_zero_pfn(unsigned long pfn)
{
	return pfn == zero_pfn;
}
#endif

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

H
Hugh Dickins 已提交
555
/*
N
Nick Piggin 已提交
556
 * vm_normal_page -- This function gets the "struct page" associated with a pte.
557
 *
N
Nick Piggin 已提交
558 559 560
 * "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
561
 *
N
Nick Piggin 已提交
562 563 564 565 566 567 568 569
 * 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.
570
 *
J
Jared Hulbert 已提交
571 572
 * The way we recognize COWed pages within VM_PFNMAP mappings is through the
 * rules set up by "remap_pfn_range()": the vma will have the VM_PFNMAP bit
N
Nick Piggin 已提交
573 574
 * set, and the vm_pgoff will point to the first PFN mapped: thus every special
 * mapping will always honor the rule
575 576 577
 *
 *	pfn_of_page == vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT)
 *
N
Nick Piggin 已提交
578 579 580 581 582 583
 * 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
584 585
 *
 *
N
Nick Piggin 已提交
586
 * In order to support COW of arbitrary special mappings, we have VM_MIXEDMAP.
J
Jared Hulbert 已提交
587 588 589 590 591 592 593 594 595
 *
 * VM_MIXEDMAP mappings can likewise contain memory with or without "struct
 * page" backing, however the difference is that _all_ pages with a struct
 * page (that is, those where pfn_valid is true) are refcounted and considered
 * normal pages by the VM. The disadvantage is that pages are refcounted
 * (which can be slower and simply not an option for some PFNMAP users). The
 * advantage is that we don't have to follow the strict linearity rule of
 * PFNMAP mappings in order to support COWable mappings.
 *
H
Hugh Dickins 已提交
596
 */
N
Nick Piggin 已提交
597 598 599 600 601 602 603
#ifdef __HAVE_ARCH_PTE_SPECIAL
# define HAVE_PTE_SPECIAL 1
#else
# define HAVE_PTE_SPECIAL 0
#endif
struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
				pte_t pte)
H
Hugh Dickins 已提交
604
{
605
	unsigned long pfn = pte_pfn(pte);
N
Nick Piggin 已提交
606 607

	if (HAVE_PTE_SPECIAL) {
608 609
		if (likely(!pte_special(pte)))
			goto check_pfn;
H
Hugh Dickins 已提交
610 611
		if (vma->vm_flags & (VM_PFNMAP | VM_MIXEDMAP))
			return NULL;
H
Hugh Dickins 已提交
612
		if (!is_zero_pfn(pfn))
613
			print_bad_pte(vma, addr, pte, NULL);
N
Nick Piggin 已提交
614 615 616 617 618
		return NULL;
	}

	/* !HAVE_PTE_SPECIAL case follows: */

J
Jared Hulbert 已提交
619 620 621 622 623 624
	if (unlikely(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP))) {
		if (vma->vm_flags & VM_MIXEDMAP) {
			if (!pfn_valid(pfn))
				return NULL;
			goto out;
		} else {
N
Nick Piggin 已提交
625 626
			unsigned long off;
			off = (addr - vma->vm_start) >> PAGE_SHIFT;
J
Jared Hulbert 已提交
627 628 629 630 631
			if (pfn == vma->vm_pgoff + off)
				return NULL;
			if (!is_cow_mapping(vma->vm_flags))
				return NULL;
		}
632 633
	}

H
Hugh Dickins 已提交
634 635
	if (is_zero_pfn(pfn))
		return NULL;
636 637 638 639 640
check_pfn:
	if (unlikely(pfn > highest_memmap_pfn)) {
		print_bad_pte(vma, addr, pte, NULL);
		return NULL;
	}
641 642

	/*
N
Nick Piggin 已提交
643 644
	 * NOTE! We still have PageReserved() pages in the page tables.
	 * eg. VDSO mappings can cause them to exist.
645
	 */
J
Jared Hulbert 已提交
646
out:
647
	return pfn_to_page(pfn);
H
Hugh Dickins 已提交
648 649
}

L
Linus Torvalds 已提交
650 651 652 653 654 655
/*
 * copy one vm_area from one task to the other. Assumes the page tables
 * already present in the new task to be cleared in the whole range
 * covered by this vma.
 */

H
Hugh Dickins 已提交
656
static inline unsigned long
L
Linus Torvalds 已提交
657
copy_one_pte(struct mm_struct *dst_mm, struct mm_struct *src_mm,
N
Nick Piggin 已提交
658
		pte_t *dst_pte, pte_t *src_pte, struct vm_area_struct *vma,
H
Hugh Dickins 已提交
659
		unsigned long addr, int *rss)
L
Linus Torvalds 已提交
660
{
N
Nick Piggin 已提交
661
	unsigned long vm_flags = vma->vm_flags;
L
Linus Torvalds 已提交
662 663 664 665 666 667
	pte_t pte = *src_pte;
	struct page *page;

	/* pte contains position in swap or file, so copy. */
	if (unlikely(!pte_present(pte))) {
		if (!pte_file(pte)) {
668 669
			swp_entry_t entry = pte_to_swp_entry(pte);

H
Hugh Dickins 已提交
670 671 672
			if (swap_duplicate(entry) < 0)
				return entry.val;

L
Linus Torvalds 已提交
673 674 675
			/* make sure dst_mm is on swapoff's mmlist. */
			if (unlikely(list_empty(&dst_mm->mmlist))) {
				spin_lock(&mmlist_lock);
676 677 678
				if (list_empty(&dst_mm->mmlist))
					list_add(&dst_mm->mmlist,
						 &src_mm->mmlist);
L
Linus Torvalds 已提交
679 680
				spin_unlock(&mmlist_lock);
			}
K
KAMEZAWA Hiroyuki 已提交
681 682 683
			if (likely(!non_swap_entry(entry)))
				rss[MM_SWAPENTS]++;
			else if (is_write_migration_entry(entry) &&
684 685 686 687 688 689 690 691 692
					is_cow_mapping(vm_flags)) {
				/*
				 * COW mappings require pages in both parent
				 * and child to be set to read.
				 */
				make_migration_entry_read(&entry);
				pte = swp_entry_to_pte(entry);
				set_pte_at(src_mm, addr, src_pte, pte);
			}
L
Linus Torvalds 已提交
693
		}
694
		goto out_set_pte;
L
Linus Torvalds 已提交
695 696 697 698 699 700
	}

	/*
	 * If it's a COW mapping, write protect it both
	 * in the parent and the child
	 */
701
	if (is_cow_mapping(vm_flags)) {
L
Linus Torvalds 已提交
702
		ptep_set_wrprotect(src_mm, addr, src_pte);
703
		pte = pte_wrprotect(pte);
L
Linus Torvalds 已提交
704 705 706 707 708 709 710 711 712
	}

	/*
	 * 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);
713 714 715 716

	page = vm_normal_page(vma, addr, pte);
	if (page) {
		get_page(page);
H
Hugh Dickins 已提交
717
		page_dup_rmap(page);
K
KAMEZAWA Hiroyuki 已提交
718 719 720 721
		if (PageAnon(page))
			rss[MM_ANONPAGES]++;
		else
			rss[MM_FILEPAGES]++;
722
	}
723 724 725

out_set_pte:
	set_pte_at(dst_mm, addr, dst_pte, pte);
H
Hugh Dickins 已提交
726
	return 0;
L
Linus Torvalds 已提交
727 728
}

729 730 731
int copy_pte_range(struct mm_struct *dst_mm, struct mm_struct *src_mm,
		   pmd_t *dst_pmd, pmd_t *src_pmd, struct vm_area_struct *vma,
		   unsigned long addr, unsigned long end)
L
Linus Torvalds 已提交
732
{
733
	pte_t *orig_src_pte, *orig_dst_pte;
L
Linus Torvalds 已提交
734
	pte_t *src_pte, *dst_pte;
H
Hugh Dickins 已提交
735
	spinlock_t *src_ptl, *dst_ptl;
736
	int progress = 0;
K
KAMEZAWA Hiroyuki 已提交
737
	int rss[NR_MM_COUNTERS];
H
Hugh Dickins 已提交
738
	swp_entry_t entry = (swp_entry_t){0};
L
Linus Torvalds 已提交
739 740

again:
K
KAMEZAWA Hiroyuki 已提交
741 742
	init_rss_vec(rss);

H
Hugh Dickins 已提交
743
	dst_pte = pte_alloc_map_lock(dst_mm, dst_pmd, addr, &dst_ptl);
L
Linus Torvalds 已提交
744 745
	if (!dst_pte)
		return -ENOMEM;
P
Peter Zijlstra 已提交
746
	src_pte = pte_offset_map(src_pmd, addr);
H
Hugh Dickins 已提交
747
	src_ptl = pte_lockptr(src_mm, src_pmd);
I
Ingo Molnar 已提交
748
	spin_lock_nested(src_ptl, SINGLE_DEPTH_NESTING);
749 750
	orig_src_pte = src_pte;
	orig_dst_pte = dst_pte;
751
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
752 753 754 755 756 757

	do {
		/*
		 * We are holding two locks at this point - either of them
		 * could generate latencies in another task on another CPU.
		 */
758 759 760
		if (progress >= 32) {
			progress = 0;
			if (need_resched() ||
N
Nick Piggin 已提交
761
			    spin_needbreak(src_ptl) || spin_needbreak(dst_ptl))
762 763
				break;
		}
L
Linus Torvalds 已提交
764 765 766 767
		if (pte_none(*src_pte)) {
			progress++;
			continue;
		}
H
Hugh Dickins 已提交
768 769 770 771
		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
772 773 774
		progress += 8;
	} while (dst_pte++, src_pte++, addr += PAGE_SIZE, addr != end);

775
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
776
	spin_unlock(src_ptl);
P
Peter Zijlstra 已提交
777
	pte_unmap(orig_src_pte);
K
KAMEZAWA Hiroyuki 已提交
778
	add_mm_rss_vec(dst_mm, rss);
779
	pte_unmap_unlock(orig_dst_pte, dst_ptl);
H
Hugh Dickins 已提交
780
	cond_resched();
H
Hugh Dickins 已提交
781 782 783 784 785 786

	if (entry.val) {
		if (add_swap_count_continuation(entry, GFP_KERNEL) < 0)
			return -ENOMEM;
		progress = 0;
	}
L
Linus Torvalds 已提交
787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804
	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);
805 806
		if (pmd_trans_huge(*src_pmd)) {
			int err;
807
			VM_BUG_ON(next-addr != HPAGE_PMD_SIZE);
808 809 810 811 812 813 814 815
			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
816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853
		if (pmd_none_or_clear_bad(src_pmd))
			continue;
		if (copy_pte_range(dst_mm, src_mm, dst_pmd, src_pmd,
						vma, addr, next))
			return -ENOMEM;
	} while (dst_pmd++, src_pmd++, addr = next, addr != end);
	return 0;
}

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

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

int copy_page_range(struct mm_struct *dst_mm, struct mm_struct *src_mm,
		struct vm_area_struct *vma)
{
	pgd_t *src_pgd, *dst_pgd;
	unsigned long next;
	unsigned long addr = vma->vm_start;
	unsigned long end = vma->vm_end;
A
Andrea Arcangeli 已提交
854
	int ret;
L
Linus Torvalds 已提交
855

856 857 858 859 860 861
	/*
	 * 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.
	 */
862
	if (!(vma->vm_flags & (VM_HUGETLB|VM_NONLINEAR|VM_PFNMAP|VM_INSERTPAGE))) {
863 864 865 866
		if (!vma->anon_vma)
			return 0;
	}

L
Linus Torvalds 已提交
867 868 869
	if (is_vm_hugetlb_page(vma))
		return copy_hugetlb_page_range(dst_mm, src_mm, vma);

870
	if (unlikely(is_pfn_mapping(vma))) {
871 872 873 874 875 876 877 878 879
		/*
		 * We do not free on error cases below as remove_vma
		 * gets called on error from higher level routine
		 */
		ret = track_pfn_vma_copy(vma);
		if (ret)
			return ret;
	}

A
Andrea Arcangeli 已提交
880 881 882 883 884 885 886 887 888 889
	/*
	 * We need to invalidate the secondary MMU mappings only when
	 * there could be a permission downgrade on the ptes of the
	 * parent mm. And a permission downgrade will only happen if
	 * is_cow_mapping() returns true.
	 */
	if (is_cow_mapping(vma->vm_flags))
		mmu_notifier_invalidate_range_start(src_mm, addr, end);

	ret = 0;
L
Linus Torvalds 已提交
890 891 892 893 894 895
	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
896 897 898 899 900
		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
901
	} while (dst_pgd++, src_pgd++, addr = next, addr != end);
A
Andrea Arcangeli 已提交
902 903 904 905 906

	if (is_cow_mapping(vma->vm_flags))
		mmu_notifier_invalidate_range_end(src_mm,
						  vma->vm_start, end);
	return ret;
L
Linus Torvalds 已提交
907 908
}

909
static unsigned long zap_pte_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
910
				struct vm_area_struct *vma, pmd_t *pmd,
L
Linus Torvalds 已提交
911
				unsigned long addr, unsigned long end,
912
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
913
{
N
Nick Piggin 已提交
914
	struct mm_struct *mm = tlb->mm;
L
Linus Torvalds 已提交
915
	pte_t *pte;
916
	spinlock_t *ptl;
K
KAMEZAWA Hiroyuki 已提交
917 918 919
	int rss[NR_MM_COUNTERS];

	init_rss_vec(rss);
L
Linus Torvalds 已提交
920

921
	pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
922
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
923 924
	do {
		pte_t ptent = *pte;
925 926
		if (pte_none(ptent)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
927
			continue;
928
		}
929 930 931

		(*zap_work) -= PAGE_SIZE;

L
Linus Torvalds 已提交
932
		if (pte_present(ptent)) {
H
Hugh Dickins 已提交
933
			struct page *page;
934

935
			page = vm_normal_page(vma, addr, ptent);
L
Linus Torvalds 已提交
936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953
			if (unlikely(details) && page) {
				/*
				 * unmap_shared_mapping_pages() wants to
				 * invalidate cache without truncating:
				 * unmap shared but keep private pages.
				 */
				if (details->check_mapping &&
				    details->check_mapping != page->mapping)
					continue;
				/*
				 * Each page->index must be checked when
				 * invalidating or truncating nonlinear.
				 */
				if (details->nonlinear_vma &&
				    (page->index < details->first_index ||
				     page->index > details->last_index))
					continue;
			}
N
Nick Piggin 已提交
954
			ptent = ptep_get_and_clear_full(mm, addr, pte,
955
							tlb->fullmm);
L
Linus Torvalds 已提交
956 957 958 959 960 961
			tlb_remove_tlb_entry(tlb, pte, addr);
			if (unlikely(!page))
				continue;
			if (unlikely(details) && details->nonlinear_vma
			    && linear_page_index(details->nonlinear_vma,
						addr) != page->index)
N
Nick Piggin 已提交
962
				set_pte_at(mm, addr, pte,
L
Linus Torvalds 已提交
963 964
					   pgoff_to_pte(page->index));
			if (PageAnon(page))
K
KAMEZAWA Hiroyuki 已提交
965
				rss[MM_ANONPAGES]--;
966 967 968
			else {
				if (pte_dirty(ptent))
					set_page_dirty(page);
969 970
				if (pte_young(ptent) &&
				    likely(!VM_SequentialReadHint(vma)))
971
					mark_page_accessed(page);
K
KAMEZAWA Hiroyuki 已提交
972
				rss[MM_FILEPAGES]--;
973
			}
974
			page_remove_rmap(page);
975 976
			if (unlikely(page_mapcount(page) < 0))
				print_bad_pte(vma, addr, ptent, page);
L
Linus Torvalds 已提交
977 978 979 980 981 982 983 984 985
			tlb_remove_page(tlb, page);
			continue;
		}
		/*
		 * If details->check_mapping, we leave swap entries;
		 * if details->nonlinear_vma, we leave file entries.
		 */
		if (unlikely(details))
			continue;
986 987 988
		if (pte_file(ptent)) {
			if (unlikely(!(vma->vm_flags & VM_NONLINEAR)))
				print_bad_pte(vma, addr, ptent, NULL);
K
KAMEZAWA Hiroyuki 已提交
989 990 991 992 993 994 995 996
		} else {
			swp_entry_t entry = pte_to_swp_entry(ptent);

			if (!non_swap_entry(entry))
				rss[MM_SWAPENTS]--;
			if (unlikely(!free_swap_and_cache(entry)))
				print_bad_pte(vma, addr, ptent, NULL);
		}
997
		pte_clear_not_present_full(mm, addr, pte, tlb->fullmm);
998
	} while (pte++, addr += PAGE_SIZE, (addr != end && *zap_work > 0));
999

K
KAMEZAWA Hiroyuki 已提交
1000
	add_mm_rss_vec(mm, rss);
1001
	arch_leave_lazy_mmu_mode();
1002
	pte_unmap_unlock(pte - 1, ptl);
1003 1004

	return addr;
L
Linus Torvalds 已提交
1005 1006
}

1007
static inline unsigned long zap_pmd_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
1008
				struct vm_area_struct *vma, pud_t *pud,
L
Linus Torvalds 已提交
1009
				unsigned long addr, unsigned long end,
1010
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
1011 1012 1013 1014 1015 1016 1017
{
	pmd_t *pmd;
	unsigned long next;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
1018
		if (pmd_trans_huge(*pmd)) {
1019 1020
			if (next-addr != HPAGE_PMD_SIZE) {
				VM_BUG_ON(!rwsem_is_locked(&tlb->mm->mmap_sem));
1021
				split_huge_page_pmd(vma->vm_mm, pmd);
1022
			} else if (zap_huge_pmd(tlb, vma, pmd)) {
1023 1024 1025 1026 1027
				(*zap_work)--;
				continue;
			}
			/* fall through */
		}
1028 1029
		if (pmd_none_or_clear_bad(pmd)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
1030
			continue;
1031 1032 1033 1034 1035 1036
		}
		next = zap_pte_range(tlb, vma, pmd, addr, next,
						zap_work, details);
	} while (pmd++, addr = next, (addr != end && *zap_work > 0));

	return addr;
L
Linus Torvalds 已提交
1037 1038
}

1039
static inline unsigned long zap_pud_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
1040
				struct vm_area_struct *vma, pgd_t *pgd,
L
Linus Torvalds 已提交
1041
				unsigned long addr, unsigned long end,
1042
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
1043 1044 1045 1046 1047 1048 1049
{
	pud_t *pud;
	unsigned long next;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
1050 1051
		if (pud_none_or_clear_bad(pud)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
1052
			continue;
1053 1054 1055 1056 1057 1058
		}
		next = zap_pmd_range(tlb, vma, pud, addr, next,
						zap_work, details);
	} while (pud++, addr = next, (addr != end && *zap_work > 0));

	return addr;
L
Linus Torvalds 已提交
1059 1060
}

1061 1062
static unsigned long unmap_page_range(struct mmu_gather *tlb,
				struct vm_area_struct *vma,
L
Linus Torvalds 已提交
1063
				unsigned long addr, unsigned long end,
1064
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
1065 1066 1067 1068 1069 1070 1071 1072
{
	pgd_t *pgd;
	unsigned long next;

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

	BUG_ON(addr >= end);
1073
	mem_cgroup_uncharge_start();
L
Linus Torvalds 已提交
1074 1075 1076 1077
	tlb_start_vma(tlb, vma);
	pgd = pgd_offset(vma->vm_mm, addr);
	do {
		next = pgd_addr_end(addr, end);
1078 1079
		if (pgd_none_or_clear_bad(pgd)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
1080
			continue;
1081 1082 1083 1084
		}
		next = zap_pud_range(tlb, vma, pgd, addr, next,
						zap_work, details);
	} while (pgd++, addr = next, (addr != end && *zap_work > 0));
L
Linus Torvalds 已提交
1085
	tlb_end_vma(tlb, vma);
1086
	mem_cgroup_uncharge_end();
1087 1088

	return addr;
L
Linus Torvalds 已提交
1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106
}

#ifdef CONFIG_PREEMPT
# define ZAP_BLOCK_SIZE	(8 * PAGE_SIZE)
#else
/* No preempt: go for improved straight-line efficiency */
# define ZAP_BLOCK_SIZE	(1024 * PAGE_SIZE)
#endif

/**
 * unmap_vmas - unmap a range of memory covered by a list of vma's
 * @tlbp: address of the caller's struct mmu_gather
 * @vma: the starting vma
 * @start_addr: virtual address at which to start unmapping
 * @end_addr: virtual address at which to end unmapping
 * @nr_accounted: Place number of unmapped pages in vm-accountable vma's here
 * @details: details of nonlinear truncation or shared cache invalidation
 *
1107
 * Returns the end address of the unmapping (restart addr if interrupted).
L
Linus Torvalds 已提交
1108
 *
1109
 * Unmap all pages in the vma list.
L
Linus Torvalds 已提交
1110
 *
1111 1112
 * We aim to not hold locks for too long (for scheduling latency reasons).
 * So zap pages in ZAP_BLOCK_SIZE bytecounts.  This means we need to
L
Linus Torvalds 已提交
1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123
 * return the ending mmu_gather to the caller.
 *
 * 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.
 */
1124
unsigned long unmap_vmas(struct mmu_gather **tlbp,
L
Linus Torvalds 已提交
1125 1126 1127 1128
		struct vm_area_struct *vma, unsigned long start_addr,
		unsigned long end_addr, unsigned long *nr_accounted,
		struct zap_details *details)
{
1129
	long zap_work = ZAP_BLOCK_SIZE;
L
Linus Torvalds 已提交
1130 1131
	unsigned long tlb_start = 0;	/* For tlb_finish_mmu */
	int tlb_start_valid = 0;
1132
	unsigned long start = start_addr;
L
Linus Torvalds 已提交
1133
	spinlock_t *i_mmap_lock = details? details->i_mmap_lock: NULL;
1134
	int fullmm = (*tlbp)->fullmm;
A
Andrea Arcangeli 已提交
1135
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
1136

A
Andrea Arcangeli 已提交
1137
	mmu_notifier_invalidate_range_start(mm, start_addr, end_addr);
L
Linus Torvalds 已提交
1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150
	for ( ; vma && vma->vm_start < end_addr; vma = vma->vm_next) {
		unsigned long end;

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

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

1151
		if (unlikely(is_pfn_mapping(vma)))
1152 1153
			untrack_pfn_vma(vma, 0, 0);

L
Linus Torvalds 已提交
1154 1155 1156 1157 1158 1159
		while (start != end) {
			if (!tlb_start_valid) {
				tlb_start = start;
				tlb_start_valid = 1;
			}

1160
			if (unlikely(is_vm_hugetlb_page(vma))) {
1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174
				/*
				 * It is undesirable to test vma->vm_file as it
				 * should be non-null for valid hugetlb area.
				 * However, vm_file will be NULL in the error
				 * cleanup path of do_mmap_pgoff. When
				 * hugetlbfs ->mmap method fails,
				 * do_mmap_pgoff() nullifies vma->vm_file
				 * before calling this function to clean up.
				 * Since no pte has actually been setup, it is
				 * safe to do nothing in this case.
				 */
				if (vma->vm_file) {
					unmap_hugepage_range(vma, start, end, NULL);
					zap_work -= (end - start) /
1175
					pages_per_huge_page(hstate_vma(vma));
1176 1177
				}

1178 1179 1180 1181 1182 1183 1184 1185
				start = end;
			} else
				start = unmap_page_range(*tlbp, vma,
						start, end, &zap_work, details);

			if (zap_work > 0) {
				BUG_ON(start != end);
				break;
L
Linus Torvalds 已提交
1186 1187 1188 1189 1190
			}

			tlb_finish_mmu(*tlbp, tlb_start, start);

			if (need_resched() ||
N
Nick Piggin 已提交
1191
				(i_mmap_lock && spin_needbreak(i_mmap_lock))) {
L
Linus Torvalds 已提交
1192
				if (i_mmap_lock) {
1193
					*tlbp = NULL;
L
Linus Torvalds 已提交
1194 1195 1196 1197 1198
					goto out;
				}
				cond_resched();
			}

1199
			*tlbp = tlb_gather_mmu(vma->vm_mm, fullmm);
L
Linus Torvalds 已提交
1200
			tlb_start_valid = 0;
1201
			zap_work = ZAP_BLOCK_SIZE;
L
Linus Torvalds 已提交
1202 1203 1204
		}
	}
out:
A
Andrea Arcangeli 已提交
1205
	mmu_notifier_invalidate_range_end(mm, start_addr, end_addr);
1206
	return start;	/* which is now the end (or restart) address */
L
Linus Torvalds 已提交
1207 1208 1209 1210 1211 1212 1213 1214 1215
}

/**
 * zap_page_range - remove user pages in a given range
 * @vma: vm_area_struct holding the applicable pages
 * @address: starting address of pages to zap
 * @size: number of bytes to zap
 * @details: details of nonlinear truncation or shared cache invalidation
 */
1216
unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
L
Linus Torvalds 已提交
1217 1218 1219 1220 1221 1222 1223 1224 1225
		unsigned long size, struct zap_details *details)
{
	struct mm_struct *mm = vma->vm_mm;
	struct mmu_gather *tlb;
	unsigned long end = address + size;
	unsigned long nr_accounted = 0;

	lru_add_drain();
	tlb = tlb_gather_mmu(mm, 0);
1226
	update_hiwater_rss(mm);
1227 1228 1229
	end = unmap_vmas(&tlb, vma, address, end, &nr_accounted, details);
	if (tlb)
		tlb_finish_mmu(tlb, address, end);
1230
	return end;
L
Linus Torvalds 已提交
1231 1232
}

1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255
/**
 * zap_vma_ptes - remove ptes mapping the vma
 * @vma: vm_area_struct holding ptes to be zapped
 * @address: starting address of pages to zap
 * @size: number of bytes to zap
 *
 * This function only unmaps ptes assigned to VM_PFNMAP vmas.
 *
 * The entire address range must be fully contained within the vma.
 *
 * Returns 0 if successful.
 */
int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
		unsigned long size)
{
	if (address < vma->vm_start || address + size > vma->vm_end ||
	    		!(vma->vm_flags & VM_PFNMAP))
		return -1;
	zap_page_range(vma, address, size, NULL);
	return 0;
}
EXPORT_SYMBOL_GPL(zap_vma_ptes);

J
Johannes Weiner 已提交
1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
/**
 * follow_page - look up a page descriptor from a user-virtual address
 * @vma: vm_area_struct mapping @address
 * @address: virtual address to look up
 * @flags: flags modifying lookup behaviour
 *
 * @flags can have FOLL_ flags set, defined in <linux/mm.h>
 *
 * Returns the mapped (struct page *), %NULL if no mapping exists, or
 * an error pointer if there is a mapping to something not represented
 * by a page descriptor (see also vm_normal_page()).
L
Linus Torvalds 已提交
1267
 */
1268
struct page *follow_page(struct vm_area_struct *vma, unsigned long address,
1269
			unsigned int flags)
L
Linus Torvalds 已提交
1270 1271 1272 1273 1274
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *ptep, pte;
1275
	spinlock_t *ptl;
L
Linus Torvalds 已提交
1276
	struct page *page;
1277
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
1278

1279 1280 1281 1282 1283
	page = follow_huge_addr(mm, address, flags & FOLL_WRITE);
	if (!IS_ERR(page)) {
		BUG_ON(flags & FOLL_GET);
		goto out;
	}
L
Linus Torvalds 已提交
1284

1285
	page = NULL;
L
Linus Torvalds 已提交
1286 1287
	pgd = pgd_offset(mm, address);
	if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
1288
		goto no_page_table;
L
Linus Torvalds 已提交
1289 1290

	pud = pud_offset(pgd, address);
A
Andi Kleen 已提交
1291
	if (pud_none(*pud))
1292
		goto no_page_table;
1293
	if (pud_huge(*pud) && vma->vm_flags & VM_HUGETLB) {
A
Andi Kleen 已提交
1294 1295 1296 1297 1298 1299 1300
		BUG_ON(flags & FOLL_GET);
		page = follow_huge_pud(mm, address, pud, flags & FOLL_WRITE);
		goto out;
	}
	if (unlikely(pud_bad(*pud)))
		goto no_page_table;

L
Linus Torvalds 已提交
1301
	pmd = pmd_offset(pud, address);
1302
	if (pmd_none(*pmd))
1303
		goto no_page_table;
1304
	if (pmd_huge(*pmd) && vma->vm_flags & VM_HUGETLB) {
1305 1306
		BUG_ON(flags & FOLL_GET);
		page = follow_huge_pmd(mm, address, pmd, flags & FOLL_WRITE);
L
Linus Torvalds 已提交
1307
		goto out;
1308
	}
1309
	if (pmd_trans_huge(*pmd)) {
1310 1311 1312 1313
		if (flags & FOLL_SPLIT) {
			split_huge_page_pmd(mm, pmd);
			goto split_fallthrough;
		}
1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
		spin_lock(&mm->page_table_lock);
		if (likely(pmd_trans_huge(*pmd))) {
			if (unlikely(pmd_trans_splitting(*pmd))) {
				spin_unlock(&mm->page_table_lock);
				wait_split_huge_page(vma->anon_vma, pmd);
			} else {
				page = follow_trans_huge_pmd(mm, address,
							     pmd, flags);
				spin_unlock(&mm->page_table_lock);
				goto out;
			}
		} else
			spin_unlock(&mm->page_table_lock);
		/* fall through */
	}
1329
split_fallthrough:
1330 1331 1332
	if (unlikely(pmd_bad(*pmd)))
		goto no_page_table;

1333
	ptep = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
1334 1335

	pte = *ptep;
1336
	if (!pte_present(pte))
1337
		goto no_page;
1338 1339
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;
H
Hugh Dickins 已提交
1340

1341
	page = vm_normal_page(vma, address, pte);
H
Hugh Dickins 已提交
1342 1343
	if (unlikely(!page)) {
		if ((flags & FOLL_DUMP) ||
H
Hugh Dickins 已提交
1344
		    !is_zero_pfn(pte_pfn(pte)))
H
Hugh Dickins 已提交
1345 1346 1347
			goto bad_page;
		page = pte_page(pte);
	}
L
Linus Torvalds 已提交
1348

1349 1350 1351 1352 1353 1354
	if (flags & FOLL_GET)
		get_page(page);
	if (flags & FOLL_TOUCH) {
		if ((flags & FOLL_WRITE) &&
		    !pte_dirty(pte) && !PageDirty(page))
			set_page_dirty(page);
1355 1356 1357 1358 1359
		/*
		 * pte_mkyoung() would be more correct here, but atomic care
		 * is needed to avoid losing the dirty bit: it is easier to use
		 * mark_page_accessed().
		 */
1360 1361
		mark_page_accessed(page);
	}
1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383
	if (flags & FOLL_MLOCK) {
		/*
		 * The preliminary mapping check is mainly to avoid the
		 * pointless overhead of lock_page on the ZERO_PAGE
		 * which might bounce very badly if there is contention.
		 *
		 * If the page is already locked, we don't need to
		 * handle it now - vmscan will handle it later if and
		 * when it attempts to reclaim the page.
		 */
		if (page->mapping && trylock_page(page)) {
			lru_add_drain();  /* push cached pages to LRU */
			/*
			 * Because we lock page here and migration is
			 * blocked by the pte's page reference, we need
			 * only check for file-cache page truncation.
			 */
			if (page->mapping)
				mlock_vma_page(page);
			unlock_page(page);
		}
	}
1384 1385
unlock:
	pte_unmap_unlock(ptep, ptl);
L
Linus Torvalds 已提交
1386
out:
1387
	return page;
L
Linus Torvalds 已提交
1388

1389 1390 1391 1392 1393 1394 1395 1396
bad_page:
	pte_unmap_unlock(ptep, ptl);
	return ERR_PTR(-EFAULT);

no_page:
	pte_unmap_unlock(ptep, ptl);
	if (!pte_none(pte))
		return page;
H
Hugh Dickins 已提交
1397

1398 1399 1400
no_page_table:
	/*
	 * When core dumping an enormous anonymous area that nobody
H
Hugh Dickins 已提交
1401 1402 1403 1404 1405
	 * has touched so far, we don't want to allocate unnecessary pages or
	 * page tables.  Return error instead of NULL to skip handle_mm_fault,
	 * then get_dump_page() will return NULL to leave a hole in the dump.
	 * But we can only make this optimization where a hole would surely
	 * be zero-filled if handle_mm_fault() actually did handle it.
1406
	 */
H
Hugh Dickins 已提交
1407 1408 1409
	if ((flags & FOLL_DUMP) &&
	    (!vma->vm_ops || !vma->vm_ops->fault))
		return ERR_PTR(-EFAULT);
1410
	return page;
L
Linus Torvalds 已提交
1411 1412
}

N
Nick Piggin 已提交
1413
int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
H
Hugh Dickins 已提交
1414
		     unsigned long start, int nr_pages, unsigned int gup_flags,
1415 1416
		     struct page **pages, struct vm_area_struct **vmas,
		     int *nonblocking)
L
Linus Torvalds 已提交
1417 1418
{
	int i;
H
Hugh Dickins 已提交
1419
	unsigned long vm_flags;
L
Linus Torvalds 已提交
1420

1421
	if (nr_pages <= 0)
1422
		return 0;
H
Hugh Dickins 已提交
1423 1424 1425

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

L
Linus Torvalds 已提交
1426 1427
	/* 
	 * Require read or write permissions.
H
Hugh Dickins 已提交
1428
	 * If FOLL_FORCE is set, we only require the "MAY" flags.
L
Linus Torvalds 已提交
1429
	 */
H
Hugh Dickins 已提交
1430 1431 1432 1433
	vm_flags  = (gup_flags & FOLL_WRITE) ?
			(VM_WRITE | VM_MAYWRITE) : (VM_READ | VM_MAYREAD);
	vm_flags &= (gup_flags & FOLL_FORCE) ?
			(VM_MAYREAD | VM_MAYWRITE) : (VM_READ | VM_WRITE);
L
Linus Torvalds 已提交
1434 1435 1436
	i = 0;

	do {
1437
		struct vm_area_struct *vma;
L
Linus Torvalds 已提交
1438 1439 1440 1441 1442 1443 1444 1445 1446

		vma = find_extend_vma(mm, start);
		if (!vma && in_gate_area(tsk, start)) {
			unsigned long pg = start & PAGE_MASK;
			struct vm_area_struct *gate_vma = get_gate_vma(tsk);
			pgd_t *pgd;
			pud_t *pud;
			pmd_t *pmd;
			pte_t *pte;
N
Nick Piggin 已提交
1447 1448

			/* user gate pages are read-only */
H
Hugh Dickins 已提交
1449
			if (gup_flags & FOLL_WRITE)
L
Linus Torvalds 已提交
1450 1451 1452 1453 1454 1455 1456 1457 1458
				return i ? : -EFAULT;
			if (pg > TASK_SIZE)
				pgd = pgd_offset_k(pg);
			else
				pgd = pgd_offset_gate(mm, pg);
			BUG_ON(pgd_none(*pgd));
			pud = pud_offset(pgd, pg);
			BUG_ON(pud_none(*pud));
			pmd = pmd_offset(pud, pg);
1459 1460
			if (pmd_none(*pmd))
				return i ? : -EFAULT;
1461
			VM_BUG_ON(pmd_trans_huge(*pmd));
L
Linus Torvalds 已提交
1462
			pte = pte_offset_map(pmd, pg);
1463 1464 1465 1466
			if (pte_none(*pte)) {
				pte_unmap(pte);
				return i ? : -EFAULT;
			}
L
Linus Torvalds 已提交
1467
			if (pages) {
1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479
				struct page *page;

				page = vm_normal_page(gate_vma, start, *pte);
				if (!page) {
					if (!(gup_flags & FOLL_DUMP) &&
					     is_zero_pfn(pte_pfn(*pte)))
						page = pte_page(*pte);
					else {
						pte_unmap(pte);
						return i ? : -EFAULT;
					}
				}
1480
				pages[i] = page;
1481
				get_page(page);
L
Linus Torvalds 已提交
1482 1483 1484 1485 1486 1487
			}
			pte_unmap(pte);
			if (vmas)
				vmas[i] = gate_vma;
			i++;
			start += PAGE_SIZE;
1488
			nr_pages--;
L
Linus Torvalds 已提交
1489 1490 1491
			continue;
		}

N
Nick Piggin 已提交
1492 1493
		if (!vma ||
		    (vma->vm_flags & (VM_IO | VM_PFNMAP)) ||
H
Hugh Dickins 已提交
1494
		    !(vm_flags & vma->vm_flags))
L
Linus Torvalds 已提交
1495 1496
			return i ? : -EFAULT;

H
Hugh Dickins 已提交
1497 1498
		if (is_vm_hugetlb_page(vma)) {
			i = follow_hugetlb_page(mm, vma, pages, vmas,
H
Hugh Dickins 已提交
1499
					&start, &nr_pages, i, gup_flags);
H
Hugh Dickins 已提交
1500 1501
			continue;
		}
1502

L
Linus Torvalds 已提交
1503
		do {
1504
			struct page *page;
H
Hugh Dickins 已提交
1505
			unsigned int foll_flags = gup_flags;
L
Linus Torvalds 已提交
1506

1507
			/*
1508
			 * If we have a pending SIGKILL, don't keep faulting
H
Hugh Dickins 已提交
1509
			 * pages and potentially allocating memory.
1510
			 */
H
Hugh Dickins 已提交
1511
			if (unlikely(fatal_signal_pending(current)))
1512
				return i ? i : -ERESTARTSYS;
1513

1514
			cond_resched();
1515
			while (!(page = follow_page(vma, start, foll_flags))) {
1516
				int ret;
1517 1518 1519 1520 1521 1522
				unsigned int fault_flags = 0;

				if (foll_flags & FOLL_WRITE)
					fault_flags |= FAULT_FLAG_WRITE;
				if (nonblocking)
					fault_flags |= FAULT_FLAG_ALLOW_RETRY;
1523

1524
				ret = handle_mm_fault(mm, vma, start,
1525
							fault_flags);
1526

N
Nick Piggin 已提交
1527 1528 1529
				if (ret & VM_FAULT_ERROR) {
					if (ret & VM_FAULT_OOM)
						return i ? i : -ENOMEM;
1530
					if (ret &
1531 1532
					    (VM_FAULT_HWPOISON|VM_FAULT_HWPOISON_LARGE|
					     VM_FAULT_SIGBUS))
N
Nick Piggin 已提交
1533 1534 1535 1536 1537 1538 1539 1540
						return i ? i : -EFAULT;
					BUG();
				}
				if (ret & VM_FAULT_MAJOR)
					tsk->maj_flt++;
				else
					tsk->min_flt++;

1541 1542 1543 1544 1545
				if (ret & VM_FAULT_RETRY) {
					*nonblocking = 0;
					return i;
				}

1546
				/*
N
Nick Piggin 已提交
1547 1548 1549 1550
				 * The VM_FAULT_WRITE bit tells us that
				 * do_wp_page has broken COW when necessary,
				 * even if maybe_mkwrite decided not to set
				 * pte_write. We can thus safely do subsequent
1551 1552 1553 1554 1555 1556
				 * page lookups as if they were reads. But only
				 * do so when looping for pte_write is futile:
				 * in some cases userspace may also be wanting
				 * to write to the gotten user page, which a
				 * read fault here might prevent (a readonly
				 * page might get reCOWed by userspace write).
1557
				 */
1558 1559
				if ((ret & VM_FAULT_WRITE) &&
				    !(vma->vm_flags & VM_WRITE))
1560
					foll_flags &= ~FOLL_WRITE;
N
Nick Piggin 已提交
1561

1562
				cond_resched();
L
Linus Torvalds 已提交
1563
			}
1564 1565
			if (IS_ERR(page))
				return i ? i : PTR_ERR(page);
L
Linus Torvalds 已提交
1566
			if (pages) {
1567
				pages[i] = page;
1568

1569
				flush_anon_page(vma, page, start);
1570
				flush_dcache_page(page);
L
Linus Torvalds 已提交
1571 1572 1573 1574 1575
			}
			if (vmas)
				vmas[i] = vma;
			i++;
			start += PAGE_SIZE;
1576 1577 1578
			nr_pages--;
		} while (nr_pages && start < vma->vm_end);
	} while (nr_pages);
L
Linus Torvalds 已提交
1579 1580
	return i;
}
N
Nick Piggin 已提交
1581

1582 1583 1584 1585 1586
/**
 * get_user_pages() - pin user pages in memory
 * @tsk:	task_struct of target task
 * @mm:		mm_struct of target mm
 * @start:	starting user address
1587
 * @nr_pages:	number of pages from start to pin
1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598
 * @write:	whether pages will be written to by the caller
 * @force:	whether to force write access even if user mapping is
 *		readonly. This will result in the page being COWed even
 *		in MAP_SHARED mappings. You do not want this.
 * @pages:	array that receives pointers to the pages pinned.
 *		Should be at least nr_pages long. Or NULL, if caller
 *		only intends to ensure the pages are faulted in.
 * @vmas:	array of pointers to vmas corresponding to each page.
 *		Or NULL if the caller does not require them.
 *
 * Returns number of pages pinned. This may be fewer than the number
1599
 * requested. If nr_pages is 0 or negative, returns 0. If no pages
1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631
 * were pinned, returns -errno. Each page returned must be released
 * with a put_page() call when it is finished with. vmas will only
 * remain valid while mmap_sem is held.
 *
 * Must be called with mmap_sem held for read or write.
 *
 * get_user_pages walks a process's page tables and takes a reference to
 * each struct page that each user address corresponds to at a given
 * instant. That is, it takes the page that would be accessed if a user
 * thread accesses the given user virtual address at that instant.
 *
 * This does not guarantee that the page exists in the user mappings when
 * get_user_pages returns, and there may even be a completely different
 * page there in some cases (eg. if mmapped pagecache has been invalidated
 * and subsequently re faulted). However it does guarantee that the page
 * won't be freed completely. And mostly callers simply care that the page
 * contains data that was valid *at some point in time*. Typically, an IO
 * or similar operation cannot guarantee anything stronger anyway because
 * locks can't be held over the syscall boundary.
 *
 * If write=0, the page must not be written to. If the page is written to,
 * set_page_dirty (or set_page_dirty_lock, as appropriate) must be called
 * after the page is finished with, and before put_page is called.
 *
 * get_user_pages is typically used for fewer-copy IO operations, to get a
 * handle on the memory by some means other than accesses via the user virtual
 * addresses. The pages may be submitted for DMA to devices or accessed via
 * their kernel linear mapping (via the kmap APIs). Care should be taken to
 * use the correct cache flushing APIs.
 *
 * See also get_user_pages_fast, for performance critical applications.
 */
N
Nick Piggin 已提交
1632
int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
1633
		unsigned long start, int nr_pages, int write, int force,
N
Nick Piggin 已提交
1634 1635
		struct page **pages, struct vm_area_struct **vmas)
{
H
Hugh Dickins 已提交
1636
	int flags = FOLL_TOUCH;
N
Nick Piggin 已提交
1637

H
Hugh Dickins 已提交
1638 1639
	if (pages)
		flags |= FOLL_GET;
N
Nick Piggin 已提交
1640
	if (write)
H
Hugh Dickins 已提交
1641
		flags |= FOLL_WRITE;
N
Nick Piggin 已提交
1642
	if (force)
H
Hugh Dickins 已提交
1643
		flags |= FOLL_FORCE;
N
Nick Piggin 已提交
1644

1645 1646
	return __get_user_pages(tsk, mm, start, nr_pages, flags, pages, vmas,
				NULL);
N
Nick Piggin 已提交
1647
}
L
Linus Torvalds 已提交
1648 1649
EXPORT_SYMBOL(get_user_pages);

H
Hugh Dickins 已提交
1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670
/**
 * get_dump_page() - pin user page in memory while writing it to core dump
 * @addr: user address
 *
 * Returns struct page pointer of user page pinned for dump,
 * to be freed afterwards by page_cache_release() or put_page().
 *
 * Returns NULL on any kind of failure - a hole must then be inserted into
 * the corefile, to preserve alignment with its headers; and also returns
 * NULL wherever the ZERO_PAGE, or an anonymous pte_none, has been found -
 * allowing a hole to be left in the corefile to save diskspace.
 *
 * Called without mmap_sem, but after all other threads have been killed.
 */
#ifdef CONFIG_ELF_CORE
struct page *get_dump_page(unsigned long addr)
{
	struct vm_area_struct *vma;
	struct page *page;

	if (__get_user_pages(current, current->mm, addr, 1,
1671 1672
			     FOLL_FORCE | FOLL_DUMP | FOLL_GET, &page, &vma,
			     NULL) < 1)
H
Hugh Dickins 已提交
1673 1674 1675 1676 1677 1678
		return NULL;
	flush_cache_page(vma, addr, page_to_pfn(page));
	return page;
}
#endif /* CONFIG_ELF_CORE */

1679
pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
H
Harvey Harrison 已提交
1680
			spinlock_t **ptl)
1681 1682 1683 1684
{
	pgd_t * pgd = pgd_offset(mm, addr);
	pud_t * pud = pud_alloc(mm, pgd, addr);
	if (pud) {
1685
		pmd_t * pmd = pmd_alloc(mm, pud, addr);
1686 1687
		if (pmd) {
			VM_BUG_ON(pmd_trans_huge(*pmd));
1688
			return pte_alloc_map_lock(mm, pmd, addr, ptl);
1689
		}
1690 1691 1692 1693
	}
	return NULL;
}

1694 1695 1696 1697 1698 1699 1700
/*
 * 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
1701 1702
static int insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page, pgprot_t prot)
1703
{
N
Nick Piggin 已提交
1704
	struct mm_struct *mm = vma->vm_mm;
1705
	int retval;
1706
	pte_t *pte;
1707 1708
	spinlock_t *ptl;

1709
	retval = -EINVAL;
1710
	if (PageAnon(page))
1711
		goto out;
1712 1713
	retval = -ENOMEM;
	flush_dcache_page(page);
1714
	pte = get_locked_pte(mm, addr, &ptl);
1715
	if (!pte)
1716
		goto out;
1717 1718 1719 1720 1721 1722
	retval = -EBUSY;
	if (!pte_none(*pte))
		goto out_unlock;

	/* Ok, finally just insert the thing.. */
	get_page(page);
1723
	inc_mm_counter_fast(mm, MM_FILEPAGES);
1724 1725 1726 1727
	page_add_file_rmap(page);
	set_pte_at(mm, addr, pte, mk_pte(page, prot));

	retval = 0;
1728 1729
	pte_unmap_unlock(pte, ptl);
	return retval;
1730 1731 1732 1733 1734 1735
out_unlock:
	pte_unmap_unlock(pte, ptl);
out:
	return retval;
}

1736 1737 1738 1739 1740 1741
/**
 * 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
 *
1742 1743 1744 1745 1746 1747
 * This allows drivers to insert individual pages they've allocated
 * into a user vma.
 *
 * The page has to be a nice clean _individual_ kernel allocation.
 * If you allocate a compound page, you need to have marked it as
 * such (__GFP_COMP), or manually just split the page up yourself
N
Nick Piggin 已提交
1748
 * (see split_page()).
1749 1750 1751 1752 1753 1754 1755 1756 1757
 *
 * NOTE! Traditionally this was done with "remap_pfn_range()" which
 * took an arbitrary page protection parameter. This doesn't allow
 * that. Your vma protection will have to be set up correctly, which
 * means that if you want a shared writable mapping, you'd better
 * ask for a shared writable mapping!
 *
 * The page does not need to be reserved.
 */
N
Nick Piggin 已提交
1758 1759
int vm_insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page)
1760 1761 1762 1763 1764
{
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
	if (!page_count(page))
		return -EINVAL;
1765
	vma->vm_flags |= VM_INSERTPAGE;
N
Nick Piggin 已提交
1766
	return insert_page(vma, addr, page, vma->vm_page_prot);
1767
}
1768
EXPORT_SYMBOL(vm_insert_page);
1769

N
Nick Piggin 已提交
1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788
static int insert_pfn(struct vm_area_struct *vma, unsigned long addr,
			unsigned long pfn, pgprot_t prot)
{
	struct mm_struct *mm = vma->vm_mm;
	int retval;
	pte_t *pte, entry;
	spinlock_t *ptl;

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

	/* Ok, finally just insert the thing.. */
	entry = pte_mkspecial(pfn_pte(pfn, prot));
	set_pte_at(mm, addr, pte, entry);
1789
	update_mmu_cache(vma, addr, pte); /* XXX: why not for insert_page? */
N
Nick Piggin 已提交
1790 1791 1792 1793 1794 1795 1796 1797

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

N
Nick Piggin 已提交
1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808
/**
 * vm_insert_pfn - insert single pfn into user vma
 * @vma: user vma to map to
 * @addr: target user address of this page
 * @pfn: source kernel pfn
 *
 * Similar to vm_inert_page, this allows drivers to insert individual pages
 * they've allocated into a user vma. Same comments apply.
 *
 * This function should only be called from a vm_ops->fault handler, and
 * in that case the handler should return NULL.
N
Nick Piggin 已提交
1809 1810 1811 1812 1813
 *
 * 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
1814 1815
 */
int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
N
Nick Piggin 已提交
1816
			unsigned long pfn)
N
Nick Piggin 已提交
1817
{
1818
	int ret;
1819
	pgprot_t pgprot = vma->vm_page_prot;
N
Nick Piggin 已提交
1820 1821 1822 1823 1824 1825
	/*
	 * 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
1826 1827 1828 1829 1830
	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
1831

N
Nick Piggin 已提交
1832 1833
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
1834
	if (track_pfn_vma_new(vma, &pgprot, pfn, PAGE_SIZE))
1835 1836
		return -EINVAL;

1837
	ret = insert_pfn(vma, addr, pfn, pgprot);
1838 1839 1840 1841 1842

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

	return ret;
N
Nick Piggin 已提交
1843 1844
}
EXPORT_SYMBOL(vm_insert_pfn);
N
Nick Piggin 已提交
1845

N
Nick Piggin 已提交
1846 1847 1848 1849
int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
			unsigned long pfn)
{
	BUG_ON(!(vma->vm_flags & VM_MIXEDMAP));
N
Nick Piggin 已提交
1850

N
Nick Piggin 已提交
1851 1852
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
N
Nick Piggin 已提交
1853

N
Nick Piggin 已提交
1854 1855 1856 1857
	/*
	 * If we don't have pte special, then we have to use the pfn_valid()
	 * based VM_MIXEDMAP scheme (see vm_normal_page), and thus we *must*
	 * refcount the page if pfn_valid is true (hence insert_page rather
H
Hugh Dickins 已提交
1858 1859
	 * 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
1860 1861 1862 1863 1864 1865 1866 1867
	 */
	if (!HAVE_PTE_SPECIAL && pfn_valid(pfn)) {
		struct page *page;

		page = pfn_to_page(pfn);
		return insert_page(vma, addr, page, vma->vm_page_prot);
	}
	return insert_pfn(vma, addr, pfn, vma->vm_page_prot);
N
Nick Piggin 已提交
1868
}
N
Nick Piggin 已提交
1869
EXPORT_SYMBOL(vm_insert_mixed);
N
Nick Piggin 已提交
1870

L
Linus Torvalds 已提交
1871 1872 1873 1874 1875 1876 1877 1878 1879 1880
/*
 * maps a range of physical memory into the requested pages. the old
 * mappings are removed. any references to nonexistent pages results
 * in null mappings (currently treated as "copy-on-access")
 */
static int remap_pte_range(struct mm_struct *mm, pmd_t *pmd,
			unsigned long addr, unsigned long end,
			unsigned long pfn, pgprot_t prot)
{
	pte_t *pte;
H
Hugh Dickins 已提交
1881
	spinlock_t *ptl;
L
Linus Torvalds 已提交
1882

H
Hugh Dickins 已提交
1883
	pte = pte_alloc_map_lock(mm, pmd, addr, &ptl);
L
Linus Torvalds 已提交
1884 1885
	if (!pte)
		return -ENOMEM;
1886
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
1887 1888
	do {
		BUG_ON(!pte_none(*pte));
N
Nick Piggin 已提交
1889
		set_pte_at(mm, addr, pte, pte_mkspecial(pfn_pte(pfn, prot)));
L
Linus Torvalds 已提交
1890 1891
		pfn++;
	} while (pte++, addr += PAGE_SIZE, addr != end);
1892
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
1893
	pte_unmap_unlock(pte - 1, ptl);
L
Linus Torvalds 已提交
1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907
	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;
1908
	VM_BUG_ON(pmd_trans_huge(*pmd));
L
Linus Torvalds 已提交
1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937
	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;
}

1938 1939 1940 1941 1942 1943 1944 1945 1946 1947
/**
 * 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
1948 1949 1950 1951 1952
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;
1953
	unsigned long end = addr + PAGE_ALIGN(size);
L
Linus Torvalds 已提交
1954 1955 1956 1957 1958 1959 1960 1961
	struct mm_struct *mm = vma->vm_mm;
	int err;

	/*
	 * Physically remapped pages are special. Tell the
	 * rest of the world about it:
	 *   VM_IO tells people not to look at these pages
	 *	(accesses can have side effects).
H
Hugh Dickins 已提交
1962 1963 1964 1965 1966
	 *   VM_RESERVED is specified all over the place, because
	 *	in 2.4 it kept swapout's vma scan off this vma; but
	 *	in 2.6 the LRU scan won't even find its pages, so this
	 *	flag means no more than count its pages in reserved_vm,
	 * 	and omit it from core dump, even when VM_IO turned off.
1967 1968 1969
	 *   VM_PFNMAP tells the core MM that the base pages are just
	 *	raw PFN mappings, and do not have a "struct page" associated
	 *	with them.
L
Linus Torvalds 已提交
1970 1971 1972 1973
	 *
	 * There's a horrible special case to handle copy-on-write
	 * behaviour that some programs depend on. We mark the "original"
	 * un-COW'ed pages by matching them up with "vma->vm_pgoff".
L
Linus Torvalds 已提交
1974
	 */
1975
	if (addr == vma->vm_start && end == vma->vm_end) {
L
Linus Torvalds 已提交
1976
		vma->vm_pgoff = pfn;
1977
		vma->vm_flags |= VM_PFN_AT_MMAP;
1978
	} else if (is_cow_mapping(vma->vm_flags))
1979
		return -EINVAL;
L
Linus Torvalds 已提交
1980

1981
	vma->vm_flags |= VM_IO | VM_RESERVED | VM_PFNMAP;
L
Linus Torvalds 已提交
1982

1983
	err = track_pfn_vma_new(vma, &prot, pfn, PAGE_ALIGN(size));
1984 1985 1986 1987 1988 1989
	if (err) {
		/*
		 * To indicate that track_pfn related cleanup is not
		 * needed from higher level routine calling unmap_vmas
		 */
		vma->vm_flags &= ~(VM_IO | VM_RESERVED | VM_PFNMAP);
1990
		vma->vm_flags &= ~VM_PFN_AT_MMAP;
1991
		return -EINVAL;
1992
	}
1993

L
Linus Torvalds 已提交
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
	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);
2005 2006 2007 2008

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

L
Linus Torvalds 已提交
2009 2010 2011 2012
	return err;
}
EXPORT_SYMBOL(remap_pfn_range);

2013 2014 2015 2016 2017 2018
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;
2019
	pgtable_t token;
2020
	spinlock_t *uninitialized_var(ptl);
2021 2022 2023 2024 2025 2026 2027 2028 2029

	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));

2030 2031
	arch_enter_lazy_mmu_mode();

2032
	token = pmd_pgtable(*pmd);
2033 2034

	do {
2035
		err = fn(pte++, token, addr, data);
2036 2037
		if (err)
			break;
2038
	} while (addr += PAGE_SIZE, addr != end);
2039

2040 2041
	arch_leave_lazy_mmu_mode();

2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054
	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
2055 2056
	BUG_ON(pud_huge(*pud));

2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097
	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;
2098
	unsigned long end = addr + size;
2099 2100 2101 2102 2103 2104 2105 2106 2107 2108
	int err;

	BUG_ON(addr >= end);
	pgd = pgd_offset(mm, addr);
	do {
		next = pgd_addr_end(addr, end);
		err = apply_to_pud_range(mm, pgd, addr, next, fn, data);
		if (err)
			break;
	} while (pgd++, addr = next, addr != end);
2109

2110 2111 2112 2113
	return err;
}
EXPORT_SYMBOL_GPL(apply_to_page_range);

2114 2115 2116 2117 2118 2119 2120 2121 2122
/*
 * 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 and do_file_page
 * must check under lock before unmapping the pte and proceeding
 * (but do_wp_page is only called after already making such a check;
 * and do_anonymous_page and do_no_page can safely check later on).
 */
H
Hugh Dickins 已提交
2123
static inline int pte_unmap_same(struct mm_struct *mm, pmd_t *pmd,
2124 2125 2126 2127 2128
				pte_t *page_table, pte_t orig_pte)
{
	int same = 1;
#if defined(CONFIG_SMP) || defined(CONFIG_PREEMPT)
	if (sizeof(pte_t) > sizeof(unsigned long)) {
H
Hugh Dickins 已提交
2129 2130
		spinlock_t *ptl = pte_lockptr(mm, pmd);
		spin_lock(ptl);
2131
		same = pte_same(*page_table, orig_pte);
H
Hugh Dickins 已提交
2132
		spin_unlock(ptl);
2133 2134 2135 2136 2137 2138
	}
#endif
	pte_unmap(page_table);
	return same;
}

2139
static inline void cow_user_page(struct page *dst, struct page *src, unsigned long va, struct vm_area_struct *vma)
2140 2141 2142 2143 2144 2145 2146 2147 2148
{
	/*
	 * 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)) {
		void *kaddr = kmap_atomic(dst, KM_USER0);
L
Linus Torvalds 已提交
2149 2150 2151 2152 2153 2154 2155 2156 2157
		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))
2158
			clear_page(kaddr);
2159
		kunmap_atomic(kaddr, KM_USER0);
2160
		flush_dcache_page(dst);
N
Nick Piggin 已提交
2161 2162
	} else
		copy_user_highpage(dst, src, va, vma);
2163 2164
}

L
Linus Torvalds 已提交
2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178
/*
 * 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.
 *
2179 2180 2181
 * 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
2182
 */
2183 2184
static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2185
		spinlock_t *ptl, pte_t orig_pte)
2186
	__releases(ptl)
L
Linus Torvalds 已提交
2187
{
2188
	struct page *old_page, *new_page;
L
Linus Torvalds 已提交
2189
	pte_t entry;
2190
	int ret = 0;
2191
	int page_mkwrite = 0;
2192
	struct page *dirty_page = NULL;
L
Linus Torvalds 已提交
2193

2194
	old_page = vm_normal_page(vma, address, orig_pte);
2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205
	if (!old_page) {
		/*
		 * VM_MIXEDMAP !pfn_valid() case
		 *
		 * We should not cow pages in a shared writeable mapping.
		 * Just mark the pages writable as we can't do any dirty
		 * accounting on raw pfn maps.
		 */
		if ((vma->vm_flags & (VM_WRITE|VM_SHARED)) ==
				     (VM_WRITE|VM_SHARED))
			goto reuse;
2206
		goto gotten;
2207
	}
L
Linus Torvalds 已提交
2208

2209
	/*
P
Peter Zijlstra 已提交
2210 2211
	 * Take out anonymous pages first, anonymous shared vmas are
	 * not dirty accountable.
2212
	 */
H
Hugh Dickins 已提交
2213
	if (PageAnon(old_page) && !PageKsm(old_page)) {
2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224
		if (!trylock_page(old_page)) {
			page_cache_get(old_page);
			pte_unmap_unlock(page_table, ptl);
			lock_page(old_page);
			page_table = pte_offset_map_lock(mm, pmd, address,
							 &ptl);
			if (!pte_same(*page_table, orig_pte)) {
				unlock_page(old_page);
				goto unlock;
			}
			page_cache_release(old_page);
P
Peter Zijlstra 已提交
2225
		}
2226
		if (reuse_swap_page(old_page)) {
2227 2228 2229 2230 2231 2232
			/*
			 * 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);
2233 2234 2235
			unlock_page(old_page);
			goto reuse;
		}
2236
		unlock_page(old_page);
P
Peter Zijlstra 已提交
2237
	} else if (unlikely((vma->vm_flags & (VM_WRITE|VM_SHARED)) ==
2238
					(VM_WRITE|VM_SHARED))) {
P
Peter Zijlstra 已提交
2239 2240 2241 2242 2243
		/*
		 * Only catch write-faults on shared writable pages,
		 * read-only shared pages can get COWed by
		 * get_user_pages(.write=1, .force=1).
		 */
2244
		if (vma->vm_ops && vma->vm_ops->page_mkwrite) {
2245 2246 2247 2248 2249 2250 2251 2252 2253
			struct vm_fault vmf;
			int tmp;

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

2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264
			/*
			 * Notify the address space that the page is about to
			 * become writable so that it can prohibit this or wait
			 * for the page to get into an appropriate state.
			 *
			 * We do this without the lock held, so that it can
			 * sleep if it needs to.
			 */
			page_cache_get(old_page);
			pte_unmap_unlock(page_table, ptl);

2265 2266 2267 2268
			tmp = vma->vm_ops->page_mkwrite(vma, &vmf);
			if (unlikely(tmp &
					(VM_FAULT_ERROR | VM_FAULT_NOPAGE))) {
				ret = tmp;
2269
				goto unwritable_page;
2270
			}
N
Nick Piggin 已提交
2271 2272 2273 2274 2275 2276 2277 2278 2279
			if (unlikely(!(tmp & VM_FAULT_LOCKED))) {
				lock_page(old_page);
				if (!old_page->mapping) {
					ret = 0; /* retry the fault */
					unlock_page(old_page);
					goto unwritable_page;
				}
			} else
				VM_BUG_ON(!PageLocked(old_page));
2280 2281 2282 2283 2284 2285 2286 2287 2288

			/*
			 * Since we dropped the lock we need to revalidate
			 * the PTE as someone else may have changed it.  If
			 * they did, we just return, as we can count on the
			 * MMU to tell us if they didn't also make it writable.
			 */
			page_table = pte_offset_map_lock(mm, pmd, address,
							 &ptl);
N
Nick Piggin 已提交
2289 2290
			if (!pte_same(*page_table, orig_pte)) {
				unlock_page(old_page);
2291
				goto unlock;
N
Nick Piggin 已提交
2292
			}
2293 2294

			page_mkwrite = 1;
L
Linus Torvalds 已提交
2295
		}
2296 2297
		dirty_page = old_page;
		get_page(dirty_page);
2298

2299
reuse:
2300 2301 2302
		flush_cache_page(vma, address, pte_pfn(orig_pte));
		entry = pte_mkyoung(orig_pte);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
2303
		if (ptep_set_access_flags(vma, address, page_table, entry,1))
2304
			update_mmu_cache(vma, address, page_table);
2305
		pte_unmap_unlock(page_table, ptl);
2306
		ret |= VM_FAULT_WRITE;
2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343

		if (!dirty_page)
			return ret;

		/*
		 * Yes, Virginia, this is actually required to prevent a race
		 * with clear_page_dirty_for_io() from clearing the page dirty
		 * bit after it clear all dirty ptes, but before a racing
		 * do_wp_page installs a dirty pte.
		 *
		 * do_no_page is protected similarly.
		 */
		if (!page_mkwrite) {
			wait_on_page_locked(dirty_page);
			set_page_dirty_balance(dirty_page, page_mkwrite);
		}
		put_page(dirty_page);
		if (page_mkwrite) {
			struct address_space *mapping = dirty_page->mapping;

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

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

		return ret;
L
Linus Torvalds 已提交
2344 2345 2346 2347 2348
	}

	/*
	 * Ok, we need to copy. Oh, well..
	 */
N
Nick Piggin 已提交
2349
	page_cache_get(old_page);
H
Hugh Dickins 已提交
2350
gotten:
2351
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
2352 2353

	if (unlikely(anon_vma_prepare(vma)))
2354
		goto oom;
H
Hugh Dickins 已提交
2355

H
Hugh Dickins 已提交
2356
	if (is_zero_pfn(pte_pfn(orig_pte))) {
H
Hugh Dickins 已提交
2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367
		new_page = alloc_zeroed_user_highpage_movable(vma, address);
		if (!new_page)
			goto oom;
	} else {
		new_page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address);
		if (!new_page)
			goto oom;
		cow_user_page(new_page, old_page, address, vma);
	}
	__SetPageUptodate(new_page);

K
KAMEZAWA Hiroyuki 已提交
2368
	if (mem_cgroup_newpage_charge(new_page, mm, GFP_KERNEL))
2369 2370
		goto oom_free_new;

L
Linus Torvalds 已提交
2371 2372 2373
	/*
	 * Re-check the pte - we dropped the lock
	 */
2374
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
2375
	if (likely(pte_same(*page_table, orig_pte))) {
H
Hugh Dickins 已提交
2376 2377
		if (old_page) {
			if (!PageAnon(old_page)) {
2378 2379
				dec_mm_counter_fast(mm, MM_FILEPAGES);
				inc_mm_counter_fast(mm, MM_ANONPAGES);
H
Hugh Dickins 已提交
2380 2381
			}
		} else
2382
			inc_mm_counter_fast(mm, MM_ANONPAGES);
2383
		flush_cache_page(vma, address, pte_pfn(orig_pte));
2384 2385
		entry = mk_pte(new_page, vma->vm_page_prot);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
2386 2387 2388 2389 2390 2391
		/*
		 * 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.
		 */
2392
		ptep_clear_flush(vma, address, page_table);
N
Nick Piggin 已提交
2393
		page_add_new_anon_rmap(new_page, vma, address);
2394 2395 2396 2397 2398 2399
		/*
		 * 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);
2400
		update_mmu_cache(vma, address, page_table);
N
Nick Piggin 已提交
2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423
		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.
			 */
2424
			page_remove_rmap(old_page);
N
Nick Piggin 已提交
2425 2426
		}

L
Linus Torvalds 已提交
2427 2428
		/* Free the old page.. */
		new_page = old_page;
N
Nick Piggin 已提交
2429
		ret |= VM_FAULT_WRITE;
2430 2431 2432
	} else
		mem_cgroup_uncharge_page(new_page);

H
Hugh Dickins 已提交
2433 2434
	if (new_page)
		page_cache_release(new_page);
2435
unlock:
2436
	pte_unmap_unlock(page_table, ptl);
2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448
	if (old_page) {
		/*
		 * Don't let another task, with possibly unlocked vma,
		 * keep the mlocked page.
		 */
		if ((ret & VM_FAULT_WRITE) && (vma->vm_flags & VM_LOCKED)) {
			lock_page(old_page);	/* LRU manipulation */
			munlock_vma_page(old_page);
			unlock_page(old_page);
		}
		page_cache_release(old_page);
	}
N
Nick Piggin 已提交
2449
	return ret;
2450
oom_free_new:
2451
	page_cache_release(new_page);
2452
oom:
N
Nick Piggin 已提交
2453 2454 2455 2456 2457
	if (old_page) {
		if (page_mkwrite) {
			unlock_page(old_page);
			page_cache_release(old_page);
		}
H
Hugh Dickins 已提交
2458
		page_cache_release(old_page);
N
Nick Piggin 已提交
2459
	}
L
Linus Torvalds 已提交
2460
	return VM_FAULT_OOM;
2461 2462 2463

unwritable_page:
	page_cache_release(old_page);
2464
	return ret;
L
Linus Torvalds 已提交
2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491
}

/*
 * Helper functions for unmap_mapping_range().
 *
 * __ Notes on dropping i_mmap_lock to reduce latency while unmapping __
 *
 * We have to restart searching the prio_tree whenever we drop the lock,
 * since the iterator is only valid while the lock is held, and anyway
 * a later vma might be split and reinserted earlier while lock dropped.
 *
 * The list of nonlinear vmas could be handled more efficiently, using
 * a placeholder, but handle it in the same way until a need is shown.
 * It is important to search the prio_tree before nonlinear list: a vma
 * may become nonlinear and be shifted from prio_tree to nonlinear list
 * while the lock is dropped; but never shifted from list to prio_tree.
 *
 * In order to make forward progress despite restarting the search,
 * vm_truncate_count is used to mark a vma as now dealt with, so we can
 * quickly skip it next time around.  Since the prio_tree search only
 * shows us those vmas affected by unmapping the range in question, we
 * can't efficiently keep all vmas in step with mapping->truncate_count:
 * so instead reset them all whenever it wraps back to 0 (then go to 1).
 * mapping->truncate_count and vma->vm_truncate_count are protected by
 * i_mmap_lock.
 *
 * In order to make forward progress despite repeatedly restarting some
2492
 * large vma, note the restart_addr from unmap_vmas when it breaks out:
L
Linus Torvalds 已提交
2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518
 * and restart from that address when we reach that vma again.  It might
 * have been split or merged, shrunk or extended, but never shifted: so
 * restart_addr remains valid so long as it remains in the vma's range.
 * unmap_mapping_range forces truncate_count to leap over page-aligned
 * values so we can save vma's restart_addr in its truncate_count field.
 */
#define is_restart_addr(truncate_count) (!((truncate_count) & ~PAGE_MASK))

static void reset_vma_truncate_counts(struct address_space *mapping)
{
	struct vm_area_struct *vma;
	struct prio_tree_iter iter;

	vma_prio_tree_foreach(vma, &iter, &mapping->i_mmap, 0, ULONG_MAX)
		vma->vm_truncate_count = 0;
	list_for_each_entry(vma, &mapping->i_mmap_nonlinear, shared.vm_set.list)
		vma->vm_truncate_count = 0;
}

static int unmap_mapping_range_vma(struct vm_area_struct *vma,
		unsigned long start_addr, unsigned long end_addr,
		struct zap_details *details)
{
	unsigned long restart_addr;
	int need_break;

2519 2520
	/*
	 * files that support invalidating or truncating portions of the
N
Nick Piggin 已提交
2521
	 * file from under mmaped areas must have their ->fault function
N
Nick Piggin 已提交
2522 2523
	 * return a locked page (and set VM_FAULT_LOCKED in the return).
	 * This provides synchronisation against concurrent unmapping here.
2524 2525
	 */

L
Linus Torvalds 已提交
2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536
again:
	restart_addr = vma->vm_truncate_count;
	if (is_restart_addr(restart_addr) && start_addr < restart_addr) {
		start_addr = restart_addr;
		if (start_addr >= end_addr) {
			/* Top of vma has been split off since last time */
			vma->vm_truncate_count = details->truncate_count;
			return 0;
		}
	}

2537 2538
	restart_addr = zap_page_range(vma, start_addr,
					end_addr - start_addr, details);
N
Nick Piggin 已提交
2539
	need_break = need_resched() || spin_needbreak(details->i_mmap_lock);
L
Linus Torvalds 已提交
2540

2541
	if (restart_addr >= end_addr) {
L
Linus Torvalds 已提交
2542 2543 2544 2545 2546 2547
		/* We have now completed this vma: mark it so */
		vma->vm_truncate_count = details->truncate_count;
		if (!need_break)
			return 0;
	} else {
		/* Note restart_addr in vma's truncate_count field */
2548
		vma->vm_truncate_count = restart_addr;
L
Linus Torvalds 已提交
2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614
		if (!need_break)
			goto again;
	}

	spin_unlock(details->i_mmap_lock);
	cond_resched();
	spin_lock(details->i_mmap_lock);
	return -EINTR;
}

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

restart:
	vma_prio_tree_foreach(vma, &iter, root,
			details->first_index, details->last_index) {
		/* Skip quickly over those we have already dealt with */
		if (vma->vm_truncate_count == details->truncate_count)
			continue;

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

		if (unmap_mapping_range_vma(vma,
			((zba - vba) << PAGE_SHIFT) + vma->vm_start,
			((zea - vba + 1) << PAGE_SHIFT) + vma->vm_start,
				details) < 0)
			goto restart;
	}
}

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

	/*
	 * In nonlinear VMAs there is no correspondence between virtual address
	 * offset and file offset.  So we must perform an exhaustive search
	 * across *all* the pages in each nonlinear VMA, not just the pages
	 * whose virtual address lies outside the file truncation point.
	 */
restart:
	list_for_each_entry(vma, head, shared.vm_set.list) {
		/* Skip quickly over those we have already dealt with */
		if (vma->vm_truncate_count == details->truncate_count)
			continue;
		details->nonlinear_vma = vma;
		if (unmap_mapping_range_vma(vma, vma->vm_start,
					vma->vm_end, details) < 0)
			goto restart;
	}
}

/**
2615
 * unmap_mapping_range - unmap the portion of all mmaps in the specified address_space corresponding to the specified page range in the underlying file.
M
Martin Waitz 已提交
2616
 * @mapping: the address space containing mmaps to be unmapped.
L
Linus Torvalds 已提交
2617 2618
 * @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
2619
 * boundary.  Note that this is different from truncate_pagecache(), which
L
Linus Torvalds 已提交
2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652
 * must keep the partial page.  In contrast, we must get rid of
 * partial pages.
 * @holelen: size of prospective hole in bytes.  This will be rounded
 * up to a PAGE_SIZE boundary.  A holelen of zero truncates to the
 * end of the file.
 * @even_cows: 1 when truncating a file, unmap even private COWed pages;
 * but 0 when invalidating pagecache, don't throw away private data.
 */
void unmap_mapping_range(struct address_space *mapping,
		loff_t const holebegin, loff_t const holelen, int even_cows)
{
	struct zap_details details;
	pgoff_t hba = holebegin >> PAGE_SHIFT;
	pgoff_t hlen = (holelen + PAGE_SIZE - 1) >> PAGE_SHIFT;

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

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

	spin_lock(&mapping->i_mmap_lock);

2653
	/* Protect against endless unmapping loops */
L
Linus Torvalds 已提交
2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669
	mapping->truncate_count++;
	if (unlikely(is_restart_addr(mapping->truncate_count))) {
		if (mapping->truncate_count == 0)
			reset_vma_truncate_counts(mapping);
		mapping->truncate_count++;
	}
	details.truncate_count = mapping->truncate_count;

	if (unlikely(!prio_tree_empty(&mapping->i_mmap)))
		unmap_mapping_range_tree(&mapping->i_mmap, &details);
	if (unlikely(!list_empty(&mapping->i_mmap_nonlinear)))
		unmap_mapping_range_list(&mapping->i_mmap_nonlinear, &details);
	spin_unlock(&mapping->i_mmap_lock);
}
EXPORT_SYMBOL(unmap_mapping_range);

2670 2671 2672 2673 2674 2675 2676 2677 2678
int vmtruncate_range(struct inode *inode, loff_t offset, loff_t end)
{
	struct address_space *mapping = inode->i_mapping;

	/*
	 * If the underlying filesystem is not going to provide
	 * a way to truncate a range of blocks (punch a hole) -
	 * we should return failure right now.
	 */
A
Al Viro 已提交
2679
	if (!inode->i_op->truncate_range)
2680 2681
		return -ENOSYS;

2682
	mutex_lock(&inode->i_mutex);
2683 2684 2685
	down_write(&inode->i_alloc_sem);
	unmap_mapping_range(mapping, offset, (end - offset), 1);
	truncate_inode_pages_range(mapping, offset, end);
2686
	unmap_mapping_range(mapping, offset, (end - offset), 1);
2687 2688
	inode->i_op->truncate_range(inode, offset, end);
	up_write(&inode->i_alloc_sem);
2689
	mutex_unlock(&inode->i_mutex);
2690 2691 2692 2693

	return 0;
}

L
Linus Torvalds 已提交
2694
/*
2695 2696 2697
 * 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
2698
 */
2699 2700
static int do_swap_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2701
		unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
2702
{
2703
	spinlock_t *ptl;
A
Andrea Arcangeli 已提交
2704
	struct page *page, *swapcache = NULL;
2705
	swp_entry_t entry;
L
Linus Torvalds 已提交
2706
	pte_t pte;
2707
	int locked;
2708
	struct mem_cgroup *ptr = NULL;
2709
	int exclusive = 0;
N
Nick Piggin 已提交
2710
	int ret = 0;
L
Linus Torvalds 已提交
2711

H
Hugh Dickins 已提交
2712
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
2713
		goto out;
2714 2715

	entry = pte_to_swp_entry(orig_pte);
2716 2717 2718 2719 2720 2721 2722
	if (unlikely(non_swap_entry(entry))) {
		if (is_migration_entry(entry)) {
			migration_entry_wait(mm, pmd, address);
		} else if (is_hwpoison_entry(entry)) {
			ret = VM_FAULT_HWPOISON;
		} else {
			print_bad_pte(vma, address, orig_pte, NULL);
H
Hugh Dickins 已提交
2723
			ret = VM_FAULT_SIGBUS;
2724
		}
2725 2726
		goto out;
	}
2727
	delayacct_set_flag(DELAYACCT_PF_SWAPIN);
L
Linus Torvalds 已提交
2728 2729
	page = lookup_swap_cache(entry);
	if (!page) {
H
Hugh Dickins 已提交
2730
		grab_swap_token(mm); /* Contend for token _before_ read-in */
2731 2732
		page = swapin_readahead(entry,
					GFP_HIGHUSER_MOVABLE, vma, address);
L
Linus Torvalds 已提交
2733 2734
		if (!page) {
			/*
2735 2736
			 * Back out if somebody else faulted in this pte
			 * while we released the pte lock.
L
Linus Torvalds 已提交
2737
			 */
2738
			page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
2739 2740
			if (likely(pte_same(*page_table, orig_pte)))
				ret = VM_FAULT_OOM;
2741
			delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2742
			goto unlock;
L
Linus Torvalds 已提交
2743 2744 2745 2746
		}

		/* Had to read the page from swap area: Major fault */
		ret = VM_FAULT_MAJOR;
2747
		count_vm_event(PGMAJFAULT);
2748
	} else if (PageHWPoison(page)) {
2749 2750 2751 2752
		/*
		 * hwpoisoned dirty swapcache pages are kept for killing
		 * owner processes (which may be unknown at hwpoison time)
		 */
2753 2754
		ret = VM_FAULT_HWPOISON;
		delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2755
		goto out_release;
L
Linus Torvalds 已提交
2756 2757
	}

2758
	locked = lock_page_or_retry(page, mm, flags);
2759
	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2760 2761 2762 2763
	if (!locked) {
		ret |= VM_FAULT_RETRY;
		goto out_release;
	}
2764

A
Andrea Arcangeli 已提交
2765
	/*
2766 2767 2768 2769
	 * Make sure try_to_free_swap or reuse_swap_page or swapoff did not
	 * release the swapcache from under us.  The page pin, and pte_same
	 * test below, are not enough to exclude that.  Even if it is still
	 * swapcache, we need to check that the page's swap has not changed.
A
Andrea Arcangeli 已提交
2770
	 */
2771
	if (unlikely(!PageSwapCache(page) || page_private(page) != entry.val))
A
Andrea Arcangeli 已提交
2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783
		goto out_page;

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

		if (unlikely(!page)) {
			ret = VM_FAULT_OOM;
			page = swapcache;
			swapcache = NULL;
			goto out_page;
		}
H
Hugh Dickins 已提交
2784 2785
	}

K
KAMEZAWA Hiroyuki 已提交
2786
	if (mem_cgroup_try_charge_swapin(mm, page, GFP_KERNEL, &ptr)) {
2787
		ret = VM_FAULT_OOM;
2788
		goto out_page;
2789 2790
	}

L
Linus Torvalds 已提交
2791
	/*
2792
	 * Back out if somebody else already faulted in this pte.
L
Linus Torvalds 已提交
2793
	 */
2794
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
H
Hugh Dickins 已提交
2795
	if (unlikely(!pte_same(*page_table, orig_pte)))
2796 2797 2798 2799 2800
		goto out_nomap;

	if (unlikely(!PageUptodate(page))) {
		ret = VM_FAULT_SIGBUS;
		goto out_nomap;
L
Linus Torvalds 已提交
2801 2802
	}

2803 2804 2805 2806 2807 2808 2809 2810
	/*
	 * 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.
2811 2812 2813 2814
	 * Because delete_from_swap_page() may be called by reuse_swap_page(),
	 * mem_cgroup_commit_charge_swapin() may not be able to find swp_entry
	 * in page->private. In this case, a record in swap_cgroup  is silently
	 * discarded at swap_free().
2815
	 */
L
Linus Torvalds 已提交
2816

2817
	inc_mm_counter_fast(mm, MM_ANONPAGES);
K
KAMEZAWA Hiroyuki 已提交
2818
	dec_mm_counter_fast(mm, MM_SWAPENTS);
L
Linus Torvalds 已提交
2819
	pte = mk_pte(page, vma->vm_page_prot);
2820
	if ((flags & FAULT_FLAG_WRITE) && reuse_swap_page(page)) {
L
Linus Torvalds 已提交
2821
		pte = maybe_mkwrite(pte_mkdirty(pte), vma);
2822
		flags &= ~FAULT_FLAG_WRITE;
2823
		ret |= VM_FAULT_WRITE;
2824
		exclusive = 1;
L
Linus Torvalds 已提交
2825 2826 2827
	}
	flush_icache_page(vma, page);
	set_pte_at(mm, address, page_table, pte);
2828
	do_page_add_anon_rmap(page, vma, address, exclusive);
2829 2830
	/* It's better to call commit-charge after rmap is established */
	mem_cgroup_commit_charge_swapin(page, ptr);
L
Linus Torvalds 已提交
2831

2832
	swap_free(entry);
N
Nick Piggin 已提交
2833
	if (vm_swap_full() || (vma->vm_flags & VM_LOCKED) || PageMlocked(page))
2834
		try_to_free_swap(page);
2835
	unlock_page(page);
A
Andrea Arcangeli 已提交
2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847
	if (swapcache) {
		/*
		 * Hold the lock to avoid the swap entry to be reused
		 * until we take the PT lock for the pte_same() check
		 * (to avoid false positives from pte_same). For
		 * further safety release the lock after the swap_free
		 * so that the swap count won't change under a
		 * parallel locked swapcache.
		 */
		unlock_page(swapcache);
		page_cache_release(swapcache);
	}
2848

2849
	if (flags & FAULT_FLAG_WRITE) {
2850 2851 2852
		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
2853 2854 2855 2856
		goto out;
	}

	/* No need to invalidate - it was non-present before */
2857
	update_mmu_cache(vma, address, page_table);
2858
unlock:
2859
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
2860 2861
out:
	return ret;
2862
out_nomap:
2863
	mem_cgroup_cancel_charge_swapin(ptr);
2864
	pte_unmap_unlock(page_table, ptl);
2865
out_page:
2866
	unlock_page(page);
2867
out_release:
2868
	page_cache_release(page);
A
Andrea Arcangeli 已提交
2869 2870 2871 2872
	if (swapcache) {
		unlock_page(swapcache);
		page_cache_release(swapcache);
	}
2873
	return ret;
L
Linus Torvalds 已提交
2874 2875
}

2876
/*
2877 2878
 * This is like a special single-page "expand_{down|up}wards()",
 * except we must first make sure that 'address{-|+}PAGE_SIZE'
2879 2880 2881 2882 2883 2884
 * 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) {
2885 2886 2887 2888 2889 2890 2891 2892 2893 2894
		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;
2895

2896
		expand_stack(vma, address - PAGE_SIZE);
2897
	}
2898 2899 2900 2901 2902 2903 2904 2905 2906
	if ((vma->vm_flags & VM_GROWSUP) && address + PAGE_SIZE == vma->vm_end) {
		struct vm_area_struct *next = vma->vm_next;

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

		expand_upwards(vma, address + PAGE_SIZE);
	}
2907 2908 2909
	return 0;
}

L
Linus Torvalds 已提交
2910
/*
2911 2912 2913
 * 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
2914
 */
2915 2916
static int do_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2917
		unsigned int flags)
L
Linus Torvalds 已提交
2918
{
2919 2920
	struct page *page;
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2921 2922
	pte_t entry;

2923 2924 2925 2926
	pte_unmap(page_table);

	/* Check if we need to add a guard page to the stack */
	if (check_stack_guard_page(vma, address) < 0)
2927 2928
		return VM_FAULT_SIGBUS;

2929
	/* Use the zero-page for reads */
H
Hugh Dickins 已提交
2930 2931 2932
	if (!(flags & FAULT_FLAG_WRITE)) {
		entry = pte_mkspecial(pfn_pte(my_zero_pfn(address),
						vma->vm_page_prot));
2933
		page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
H
Hugh Dickins 已提交
2934 2935 2936 2937 2938
		if (!pte_none(*page_table))
			goto unlock;
		goto setpte;
	}

N
Nick Piggin 已提交
2939 2940 2941 2942 2943 2944
	/* Allocate our own private page. */
	if (unlikely(anon_vma_prepare(vma)))
		goto oom;
	page = alloc_zeroed_user_highpage_movable(vma, address);
	if (!page)
		goto oom;
N
Nick Piggin 已提交
2945
	__SetPageUptodate(page);
2946

K
KAMEZAWA Hiroyuki 已提交
2947
	if (mem_cgroup_newpage_charge(page, mm, GFP_KERNEL))
2948 2949
		goto oom_free_page;

N
Nick Piggin 已提交
2950
	entry = mk_pte(page, vma->vm_page_prot);
H
Hugh Dickins 已提交
2951 2952
	if (vma->vm_flags & VM_WRITE)
		entry = pte_mkwrite(pte_mkdirty(entry));
L
Linus Torvalds 已提交
2953

N
Nick Piggin 已提交
2954
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
2955
	if (!pte_none(*page_table))
N
Nick Piggin 已提交
2956
		goto release;
H
Hugh Dickins 已提交
2957

2958
	inc_mm_counter_fast(mm, MM_ANONPAGES);
N
Nick Piggin 已提交
2959
	page_add_new_anon_rmap(page, vma, address);
H
Hugh Dickins 已提交
2960
setpte:
2961
	set_pte_at(mm, address, page_table, entry);
L
Linus Torvalds 已提交
2962 2963

	/* No need to invalidate - it was non-present before */
2964
	update_mmu_cache(vma, address, page_table);
2965
unlock:
2966
	pte_unmap_unlock(page_table, ptl);
N
Nick Piggin 已提交
2967
	return 0;
2968
release:
2969
	mem_cgroup_uncharge_page(page);
2970 2971
	page_cache_release(page);
	goto unlock;
2972
oom_free_page:
2973
	page_cache_release(page);
2974
oom:
L
Linus Torvalds 已提交
2975 2976 2977 2978
	return VM_FAULT_OOM;
}

/*
2979
 * __do_fault() tries to create a new page mapping. It aggressively
L
Linus Torvalds 已提交
2980
 * tries to share with existing pages, but makes a separate copy if
2981 2982
 * the FAULT_FLAG_WRITE is set in the flags parameter in order to avoid
 * the next page fault.
L
Linus Torvalds 已提交
2983 2984 2985 2986
 *
 * As this is called only for pages that do not currently exist, we
 * do not need to flush old virtual caches or the TLB.
 *
2987
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
2988
 * but allow concurrent faults), and pte neither mapped nor locked.
2989
 * We return with mmap_sem still held, but pte unmapped and unlocked.
L
Linus Torvalds 已提交
2990
 */
2991
static int __do_fault(struct mm_struct *mm, struct vm_area_struct *vma,
2992
		unsigned long address, pmd_t *pmd,
2993
		pgoff_t pgoff, unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
2994
{
2995
	pte_t *page_table;
2996
	spinlock_t *ptl;
N
Nick Piggin 已提交
2997
	struct page *page;
L
Linus Torvalds 已提交
2998 2999
	pte_t entry;
	int anon = 0;
3000
	int charged = 0;
3001
	struct page *dirty_page = NULL;
N
Nick Piggin 已提交
3002 3003
	struct vm_fault vmf;
	int ret;
3004
	int page_mkwrite = 0;
3005

N
Nick Piggin 已提交
3006 3007 3008 3009
	vmf.virtual_address = (void __user *)(address & PAGE_MASK);
	vmf.pgoff = pgoff;
	vmf.flags = flags;
	vmf.page = NULL;
L
Linus Torvalds 已提交
3010

N
Nick Piggin 已提交
3011
	ret = vma->vm_ops->fault(vma, &vmf);
3012 3013
	if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE |
			    VM_FAULT_RETRY)))
N
Nick Piggin 已提交
3014
		return ret;
L
Linus Torvalds 已提交
3015

3016 3017 3018 3019 3020 3021
	if (unlikely(PageHWPoison(vmf.page))) {
		if (ret & VM_FAULT_LOCKED)
			unlock_page(vmf.page);
		return VM_FAULT_HWPOISON;
	}

3022
	/*
N
Nick Piggin 已提交
3023
	 * For consistency in subsequent calls, make the faulted page always
3024 3025
	 * locked.
	 */
N
Nick Piggin 已提交
3026
	if (unlikely(!(ret & VM_FAULT_LOCKED)))
N
Nick Piggin 已提交
3027
		lock_page(vmf.page);
3028
	else
N
Nick Piggin 已提交
3029
		VM_BUG_ON(!PageLocked(vmf.page));
3030

L
Linus Torvalds 已提交
3031 3032 3033
	/*
	 * Should we do an early C-O-W break?
	 */
N
Nick Piggin 已提交
3034
	page = vmf.page;
3035
	if (flags & FAULT_FLAG_WRITE) {
3036
		if (!(vma->vm_flags & VM_SHARED)) {
3037
			anon = 1;
3038
			if (unlikely(anon_vma_prepare(vma))) {
N
Nick Piggin 已提交
3039
				ret = VM_FAULT_OOM;
3040
				goto out;
3041
			}
N
Nick Piggin 已提交
3042 3043
			page = alloc_page_vma(GFP_HIGHUSER_MOVABLE,
						vma, address);
3044
			if (!page) {
N
Nick Piggin 已提交
3045
				ret = VM_FAULT_OOM;
3046
				goto out;
3047
			}
K
KAMEZAWA Hiroyuki 已提交
3048
			if (mem_cgroup_newpage_charge(page, mm, GFP_KERNEL)) {
3049 3050 3051 3052 3053
				ret = VM_FAULT_OOM;
				page_cache_release(page);
				goto out;
			}
			charged = 1;
N
Nick Piggin 已提交
3054
			copy_user_highpage(page, vmf.page, address, vma);
N
Nick Piggin 已提交
3055
			__SetPageUptodate(page);
3056
		} else {
3057 3058
			/*
			 * If the page will be shareable, see if the backing
3059
			 * address space wants to know that the page is about
3060 3061
			 * to become writable
			 */
3062
			if (vma->vm_ops->page_mkwrite) {
3063 3064
				int tmp;

3065
				unlock_page(page);
N
Nick Piggin 已提交
3066
				vmf.flags = FAULT_FLAG_WRITE|FAULT_FLAG_MKWRITE;
3067 3068 3069 3070
				tmp = vma->vm_ops->page_mkwrite(vma, &vmf);
				if (unlikely(tmp &
					  (VM_FAULT_ERROR | VM_FAULT_NOPAGE))) {
					ret = tmp;
N
Nick Piggin 已提交
3071
					goto unwritable_page;
N
Nick Piggin 已提交
3072
				}
N
Nick Piggin 已提交
3073 3074 3075 3076 3077 3078 3079 3080 3081
				if (unlikely(!(tmp & VM_FAULT_LOCKED))) {
					lock_page(page);
					if (!page->mapping) {
						ret = 0; /* retry the fault */
						unlock_page(page);
						goto unwritable_page;
					}
				} else
					VM_BUG_ON(!PageLocked(page));
3082
				page_mkwrite = 1;
3083 3084
			}
		}
3085

L
Linus Torvalds 已提交
3086 3087
	}

3088
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
3089 3090 3091 3092 3093 3094

	/*
	 * This silly early PAGE_DIRTY setting removes a race
	 * due to the bad i386 page protection. But it's valid
	 * for other architectures too.
	 *
3095
	 * Note that if FAULT_FLAG_WRITE is set, we either now have
L
Linus Torvalds 已提交
3096 3097 3098 3099 3100
	 * an exclusive copy of the page, or this is a shared mapping,
	 * so we can make it writable and dirty to avoid having to
	 * handle that later.
	 */
	/* Only go through if we didn't race with anybody else... */
3101
	if (likely(pte_same(*page_table, orig_pte))) {
3102 3103
		flush_icache_page(vma, page);
		entry = mk_pte(page, vma->vm_page_prot);
3104
		if (flags & FAULT_FLAG_WRITE)
L
Linus Torvalds 已提交
3105 3106
			entry = maybe_mkwrite(pte_mkdirty(entry), vma);
		if (anon) {
3107
			inc_mm_counter_fast(mm, MM_ANONPAGES);
3108
			page_add_new_anon_rmap(page, vma, address);
3109
		} else {
3110
			inc_mm_counter_fast(mm, MM_FILEPAGES);
3111
			page_add_file_rmap(page);
3112
			if (flags & FAULT_FLAG_WRITE) {
3113
				dirty_page = page;
3114 3115
				get_page(dirty_page);
			}
3116
		}
3117
		set_pte_at(mm, address, page_table, entry);
3118 3119

		/* no need to invalidate: a not-present page won't be cached */
3120
		update_mmu_cache(vma, address, page_table);
L
Linus Torvalds 已提交
3121
	} else {
3122 3123
		if (charged)
			mem_cgroup_uncharge_page(page);
3124 3125 3126
		if (anon)
			page_cache_release(page);
		else
3127
			anon = 1; /* no anon but release faulted_page */
L
Linus Torvalds 已提交
3128 3129
	}

3130
	pte_unmap_unlock(page_table, ptl);
3131 3132

out:
N
Nick Piggin 已提交
3133 3134
	if (dirty_page) {
		struct address_space *mapping = page->mapping;
3135

N
Nick Piggin 已提交
3136 3137 3138
		if (set_page_dirty(dirty_page))
			page_mkwrite = 1;
		unlock_page(dirty_page);
3139
		put_page(dirty_page);
N
Nick Piggin 已提交
3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154
		if (page_mkwrite && mapping) {
			/*
			 * Some device drivers do not set page.mapping but still
			 * dirty their pages
			 */
			balance_dirty_pages_ratelimited(mapping);
		}

		/* file_update_time outside page_lock */
		if (vma->vm_file)
			file_update_time(vma->vm_file);
	} else {
		unlock_page(vmf.page);
		if (anon)
			page_cache_release(vmf.page);
3155
	}
3156

N
Nick Piggin 已提交
3157
	return ret;
N
Nick Piggin 已提交
3158 3159 3160 3161

unwritable_page:
	page_cache_release(page);
	return ret;
3162
}
3163

3164 3165
static int do_linear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
3166
		unsigned int flags, pte_t orig_pte)
3167 3168
{
	pgoff_t pgoff = (((address & PAGE_MASK)
3169
			- vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
3170

3171 3172
	pte_unmap(page_table);
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
3173 3174
}

L
Linus Torvalds 已提交
3175 3176 3177 3178
/*
 * Fault of a previously existing named mapping. Repopulate the pte
 * from the encoded file_pte if possible. This enables swappable
 * nonlinear vmas.
3179 3180 3181 3182
 *
 * 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
3183
 */
N
Nick Piggin 已提交
3184
static int do_nonlinear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3185
		unsigned long address, pte_t *page_table, pmd_t *pmd,
3186
		unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
3187
{
3188
	pgoff_t pgoff;
L
Linus Torvalds 已提交
3189

3190 3191
	flags |= FAULT_FLAG_NONLINEAR;

H
Hugh Dickins 已提交
3192
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
N
Nick Piggin 已提交
3193
		return 0;
L
Linus Torvalds 已提交
3194

3195
	if (unlikely(!(vma->vm_flags & VM_NONLINEAR))) {
3196 3197 3198
		/*
		 * Page table corrupted: show pte and kill process.
		 */
3199
		print_bad_pte(vma, address, orig_pte, NULL);
H
Hugh Dickins 已提交
3200
		return VM_FAULT_SIGBUS;
3201 3202 3203
	}

	pgoff = pte_to_pgoff(orig_pte);
3204
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
L
Linus Torvalds 已提交
3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215
}

/*
 * These routines also need to handle stuff like marking pages dirty
 * and/or accessed for architectures that don't do it in hardware (most
 * RISC architectures).  The early dirtying is also good on the i386.
 *
 * There is also a hook called "update_mmu_cache()" that architectures
 * with external mmu caches can use to update those (ie the Sparc or
 * PowerPC hashed page tables that act as extended TLBs).
 *
H
Hugh Dickins 已提交
3216 3217 3218
 * 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
3219
 */
3220 3221 3222
int handle_pte_fault(struct mm_struct *mm,
		     struct vm_area_struct *vma, unsigned long address,
		     pte_t *pte, pmd_t *pmd, unsigned int flags)
L
Linus Torvalds 已提交
3223 3224
{
	pte_t entry;
3225
	spinlock_t *ptl;
L
Linus Torvalds 已提交
3226

3227
	entry = *pte;
L
Linus Torvalds 已提交
3228
	if (!pte_present(entry)) {
3229
		if (pte_none(entry)) {
J
Jes Sorensen 已提交
3230
			if (vma->vm_ops) {
N
Nick Piggin 已提交
3231
				if (likely(vma->vm_ops->fault))
3232
					return do_linear_fault(mm, vma, address,
3233
						pte, pmd, flags, entry);
J
Jes Sorensen 已提交
3234 3235
			}
			return do_anonymous_page(mm, vma, address,
3236
						 pte, pmd, flags);
3237
		}
L
Linus Torvalds 已提交
3238
		if (pte_file(entry))
N
Nick Piggin 已提交
3239
			return do_nonlinear_fault(mm, vma, address,
3240
					pte, pmd, flags, entry);
3241
		return do_swap_page(mm, vma, address,
3242
					pte, pmd, flags, entry);
L
Linus Torvalds 已提交
3243 3244
	}

H
Hugh Dickins 已提交
3245
	ptl = pte_lockptr(mm, pmd);
3246 3247 3248
	spin_lock(ptl);
	if (unlikely(!pte_same(*pte, entry)))
		goto unlock;
3249
	if (flags & FAULT_FLAG_WRITE) {
L
Linus Torvalds 已提交
3250
		if (!pte_write(entry))
3251 3252
			return do_wp_page(mm, vma, address,
					pte, pmd, ptl, entry);
L
Linus Torvalds 已提交
3253 3254 3255
		entry = pte_mkdirty(entry);
	}
	entry = pte_mkyoung(entry);
3256
	if (ptep_set_access_flags(vma, address, pte, entry, flags & FAULT_FLAG_WRITE)) {
3257
		update_mmu_cache(vma, address, pte);
3258 3259 3260 3261 3262 3263 3264
	} 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.
		 */
3265
		if (flags & FAULT_FLAG_WRITE)
3266
			flush_tlb_fix_spurious_fault(vma, address);
3267
	}
3268 3269
unlock:
	pte_unmap_unlock(pte, ptl);
N
Nick Piggin 已提交
3270
	return 0;
L
Linus Torvalds 已提交
3271 3272 3273 3274 3275
}

/*
 * By the time we get here, we already hold the mm semaphore
 */
N
Nick Piggin 已提交
3276
int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3277
		unsigned long address, unsigned int flags)
L
Linus Torvalds 已提交
3278 3279 3280 3281 3282 3283 3284 3285
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

	__set_current_state(TASK_RUNNING);

3286
	count_vm_event(PGFAULT);
L
Linus Torvalds 已提交
3287

3288 3289 3290
	/* do counter updates before entering really critical section. */
	check_sync_rss_stat(current);

3291
	if (unlikely(is_vm_hugetlb_page(vma)))
3292
		return hugetlb_fault(mm, vma, address, flags);
L
Linus Torvalds 已提交
3293 3294 3295 3296

	pgd = pgd_offset(mm, address);
	pud = pud_alloc(mm, pgd, address);
	if (!pud)
H
Hugh Dickins 已提交
3297
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
3298 3299
	pmd = pmd_alloc(mm, pud, address);
	if (!pmd)
H
Hugh Dickins 已提交
3300
		return VM_FAULT_OOM;
3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323
	if (pmd_none(*pmd) && transparent_hugepage_enabled(vma)) {
		if (!vma->vm_ops)
			return do_huge_pmd_anonymous_page(mm, vma, address,
							  pmd, flags);
	} else {
		pmd_t orig_pmd = *pmd;
		barrier();
		if (pmd_trans_huge(orig_pmd)) {
			if (flags & FAULT_FLAG_WRITE &&
			    !pmd_write(orig_pmd) &&
			    !pmd_trans_splitting(orig_pmd))
				return do_huge_pmd_wp_page(mm, vma, address,
							   pmd, orig_pmd);
			return 0;
		}
	}

	/*
	 * Use __pte_alloc instead of pte_alloc_map, because we can't
	 * run pte_offset_map on the pmd, if an huge pmd could
	 * materialize from under us from a different thread.
	 */
	if (unlikely(__pte_alloc(mm, vma, pmd, address)))
H
Hugh Dickins 已提交
3324
		return VM_FAULT_OOM;
3325 3326 3327 3328 3329 3330 3331 3332 3333 3334
	/* if an huge pmd materialized from under us just retry later */
	if (unlikely(pmd_trans_huge(*pmd)))
		return 0;
	/*
	 * A regular pmd is established and it can't morph into a huge pmd
	 * from under us anymore at this point because we hold the mmap_sem
	 * read mode and khugepaged takes it in write mode. So now it's
	 * safe to run pte_offset_map().
	 */
	pte = pte_offset_map(pmd, address);
L
Linus Torvalds 已提交
3335

3336
	return handle_pte_fault(mm, vma, address, pte, pmd, flags);
L
Linus Torvalds 已提交
3337 3338 3339 3340 3341
}

#ifndef __PAGETABLE_PUD_FOLDED
/*
 * Allocate page upper directory.
H
Hugh Dickins 已提交
3342
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
3343
 */
3344
int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
L
Linus Torvalds 已提交
3345
{
H
Hugh Dickins 已提交
3346 3347
	pud_t *new = pud_alloc_one(mm, address);
	if (!new)
3348
		return -ENOMEM;
L
Linus Torvalds 已提交
3349

3350 3351
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
3352
	spin_lock(&mm->page_table_lock);
3353
	if (pgd_present(*pgd))		/* Another has populated it */
3354
		pud_free(mm, new);
3355 3356
	else
		pgd_populate(mm, pgd, new);
H
Hugh Dickins 已提交
3357
	spin_unlock(&mm->page_table_lock);
3358
	return 0;
L
Linus Torvalds 已提交
3359 3360 3361 3362 3363 3364
}
#endif /* __PAGETABLE_PUD_FOLDED */

#ifndef __PAGETABLE_PMD_FOLDED
/*
 * Allocate page middle directory.
H
Hugh Dickins 已提交
3365
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
3366
 */
3367
int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
L
Linus Torvalds 已提交
3368
{
H
Hugh Dickins 已提交
3369 3370
	pmd_t *new = pmd_alloc_one(mm, address);
	if (!new)
3371
		return -ENOMEM;
L
Linus Torvalds 已提交
3372

3373 3374
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
3375
	spin_lock(&mm->page_table_lock);
L
Linus Torvalds 已提交
3376
#ifndef __ARCH_HAS_4LEVEL_HACK
3377
	if (pud_present(*pud))		/* Another has populated it */
3378
		pmd_free(mm, new);
3379 3380
	else
		pud_populate(mm, pud, new);
L
Linus Torvalds 已提交
3381
#else
3382
	if (pgd_present(*pud))		/* Another has populated it */
3383
		pmd_free(mm, new);
3384 3385
	else
		pgd_populate(mm, pud, new);
L
Linus Torvalds 已提交
3386
#endif /* __ARCH_HAS_4LEVEL_HACK */
H
Hugh Dickins 已提交
3387
	spin_unlock(&mm->page_table_lock);
3388
	return 0;
3389
}
L
Linus Torvalds 已提交
3390 3391 3392 3393 3394 3395 3396 3397 3398
#endif /* __PAGETABLE_PMD_FOLDED */

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

	vma = find_vma(current->mm, addr);
	if (!vma)
K
KOSAKI Motohiro 已提交
3399
		return -ENOMEM;
3400 3401 3402 3403 3404 3405
	/*
	 * We want to touch writable mappings with a write fault in order
	 * to break COW, except for shared mappings because these don't COW
	 * and we would not want to dirty them for nothing.
	 */
	write = (vma->vm_flags & (VM_WRITE | VM_SHARED)) == VM_WRITE;
3406 3407
	BUG_ON(addr >= end);
	BUG_ON(end > vma->vm_end);
3408
	len = DIV_ROUND_UP(end, PAGE_SIZE) - addr/PAGE_SIZE;
L
Linus Torvalds 已提交
3409 3410
	ret = get_user_pages(current, current->mm, addr,
			len, write, 0, NULL, NULL);
3411
	if (ret < 0)
L
Linus Torvalds 已提交
3412
		return ret;
3413
	return ret == len ? 0 : -EFAULT;
L
Linus Torvalds 已提交
3414 3415 3416 3417 3418
}

#if !defined(__HAVE_ARCH_GATE_AREA)

#if defined(AT_SYSINFO_EHDR)
3419
static struct vm_area_struct gate_vma;
L
Linus Torvalds 已提交
3420 3421 3422 3423 3424 3425

static int __init gate_vma_init(void)
{
	gate_vma.vm_mm = NULL;
	gate_vma.vm_start = FIXADDR_USER_START;
	gate_vma.vm_end = FIXADDR_USER_END;
R
Roland McGrath 已提交
3426 3427
	gate_vma.vm_flags = VM_READ | VM_MAYREAD | VM_EXEC | VM_MAYEXEC;
	gate_vma.vm_page_prot = __P101;
3428 3429 3430 3431 3432 3433 3434
	/*
	 * Make sure the vDSO gets into every core dump.
	 * Dumping its contents makes post-mortem fully interpretable later
	 * without matching up the same kernel and hardware config to see
	 * what PC values meant.
	 */
	gate_vma.vm_flags |= VM_ALWAYSDUMP;
L
Linus Torvalds 已提交
3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458
	return 0;
}
__initcall(gate_vma_init);
#endif

struct vm_area_struct *get_gate_vma(struct task_struct *tsk)
{
#ifdef AT_SYSINFO_EHDR
	return &gate_vma;
#else
	return NULL;
#endif
}

int in_gate_area_no_task(unsigned long addr)
{
#ifdef AT_SYSINFO_EHDR
	if ((addr >= FIXADDR_USER_START) && (addr < FIXADDR_USER_END))
		return 1;
#endif
	return 0;
}

#endif	/* __HAVE_ARCH_GATE_AREA */
3459

3460
static int __follow_pte(struct mm_struct *mm, unsigned long address,
J
Johannes Weiner 已提交
3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476
		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);
3477
	VM_BUG_ON(pmd_trans_huge(*pmd));
J
Johannes Weiner 已提交
3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497
	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;
}

3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508
static inline int follow_pte(struct mm_struct *mm, unsigned long address,
			     pte_t **ptepp, spinlock_t **ptlp)
{
	int res;

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

J
Johannes Weiner 已提交
3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537
/**
 * 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);

3538
#ifdef CONFIG_HAVE_IOREMAP_PROT
3539 3540 3541
int follow_phys(struct vm_area_struct *vma,
		unsigned long address, unsigned int flags,
		unsigned long *prot, resource_size_t *phys)
3542
{
3543
	int ret = -EINVAL;
3544 3545 3546
	pte_t *ptep, pte;
	spinlock_t *ptl;

3547 3548
	if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
		goto out;
3549

3550
	if (follow_pte(vma->vm_mm, address, &ptep, &ptl))
3551
		goto out;
3552
	pte = *ptep;
3553

3554 3555 3556 3557
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;

	*prot = pgprot_val(pte_pgprot(pte));
3558
	*phys = (resource_size_t)pte_pfn(pte) << PAGE_SHIFT;
3559

3560
	ret = 0;
3561 3562 3563
unlock:
	pte_unmap_unlock(ptep, ptl);
out:
3564
	return ret;
3565 3566 3567 3568 3569 3570 3571
}

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
3572
	void __iomem *maddr;
3573 3574
	int offset = addr & (PAGE_SIZE-1);

3575
	if (follow_phys(vma, addr, write, &prot, &phys_addr))
3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588
		return -EINVAL;

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

	return len;
}
#endif

3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604
/*
 * 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;
	struct vm_area_struct *vma;
	void *old_buf = buf;

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

	down_read(&mm->mmap_sem);
S
Simon Arlott 已提交
3605
	/* ignore errors, just check how much was successfully transferred */
3606 3607 3608
	while (len) {
		int bytes, ret, offset;
		void *maddr;
3609
		struct page *page = NULL;
3610 3611 3612

		ret = get_user_pages(tsk, mm, addr, 1,
				write, 1, &page, &vma);
3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628
		if (ret <= 0) {
			/*
			 * Check if this is a VM_IO | VM_PFNMAP VMA, which
			 * we can access using slightly different code.
			 */
#ifdef CONFIG_HAVE_IOREMAP_PROT
			vma = find_vma(mm, addr);
			if (!vma)
				break;
			if (vma->vm_ops && vma->vm_ops->access)
				ret = vma->vm_ops->access(vma, addr, buf,
							  len, write);
			if (ret <= 0)
#endif
				break;
			bytes = ret;
3629
		} else {
3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645
			bytes = len;
			offset = addr & (PAGE_SIZE-1);
			if (bytes > PAGE_SIZE-offset)
				bytes = PAGE_SIZE-offset;

			maddr = kmap(page);
			if (write) {
				copy_to_user_page(vma, page, addr,
						  maddr + offset, buf, bytes);
				set_page_dirty_lock(page);
			} else {
				copy_from_user_page(vma, page, addr,
						    buf, maddr + offset, bytes);
			}
			kunmap(page);
			page_cache_release(page);
3646 3647 3648 3649 3650 3651 3652 3653 3654 3655
		}
		len -= bytes;
		buf += bytes;
		addr += bytes;
	}
	up_read(&mm->mmap_sem);
	mmput(mm);

	return buf - old_buf;
}
3656 3657 3658 3659 3660 3661 3662 3663 3664

/*
 * 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;

3665 3666 3667 3668 3669 3670 3671
	/*
	 * Do not print if we are in atomic
	 * contexts (in exception stacks, etc.):
	 */
	if (preempt_count())
		return;

3672 3673 3674 3675 3676 3677 3678 3679
	down_read(&mm->mmap_sem);
	vma = find_vma(mm, ip);
	if (vma && vma->vm_file) {
		struct file *f = vma->vm_file;
		char *buf = (char *)__get_free_page(GFP_KERNEL);
		if (buf) {
			char *p, *s;

3680
			p = d_path(&f->f_path, buf, PAGE_SIZE);
3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693
			if (IS_ERR(p))
				p = "?";
			s = strrchr(p, '/');
			if (s)
				p = s+1;
			printk("%s%s[%lx+%lx]", prefix, p,
					vma->vm_start,
					vma->vm_end - vma->vm_start);
			free_page((unsigned long)buf);
		}
	}
	up_read(&current->mm->mmap_sem);
}
3694 3695 3696 3697

#ifdef CONFIG_PROVE_LOCKING
void might_fault(void)
{
3698 3699 3700 3701 3702 3703 3704 3705 3706
	/*
	 * 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;

3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717
	might_sleep();
	/*
	 * it would be nicer only to annotate paths which are not under
	 * pagefault_disable, however that requires a larger audit and
	 * providing helpers like get_user_atomic.
	 */
	if (!in_atomic() && current->mm)
		might_lock_read(&current->mm->mmap_sem);
}
EXPORT_SYMBOL(might_fault);
#endif
A
Andrea Arcangeli 已提交
3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788

#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 */