memory.c 91.9 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>
L
Linus Torvalds 已提交
59

A
Alexey Dobriyan 已提交
60
#include <asm/io.h>
L
Linus Torvalds 已提交
61 62 63 64 65 66
#include <asm/pgalloc.h>
#include <asm/uaccess.h>
#include <asm/tlb.h>
#include <asm/tlbflush.h>
#include <asm/pgtable.h>

67 68
#include "internal.h"

A
Andy Whitcroft 已提交
69
#ifndef CONFIG_NEED_MULTIPLE_NODES
L
Linus Torvalds 已提交
70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90
/* 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);

91 92 93 94 95 96 97 98 99 100 101 102
/*
 * 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
103 104 105 106

static int __init disable_randmaps(char *s)
{
	randomize_va_space = 0;
107
	return 1;
108 109 110
}
__setup("norandmaps", disable_randmaps);

H
Hugh Dickins 已提交
111
unsigned long zero_pfn __read_mostly;
H
Hugh Dickins 已提交
112
unsigned long highest_memmap_pfn __read_mostly;
H
Hugh Dickins 已提交
113 114 115 116 117 118 119 120 121 122

/*
 * 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);
123

L
Linus Torvalds 已提交
124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151
/*
 * 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.
 */
152 153
static void free_pte_range(struct mmu_gather *tlb, pmd_t *pmd,
			   unsigned long addr)
L
Linus Torvalds 已提交
154
{
155
	pgtable_t token = pmd_pgtable(*pmd);
156
	pmd_clear(pmd);
157
	pte_free_tlb(tlb, token, addr);
158
	tlb->mm->nr_ptes--;
L
Linus Torvalds 已提交
159 160
}

161 162 163
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
164 165 166
{
	pmd_t *pmd;
	unsigned long next;
167
	unsigned long start;
L
Linus Torvalds 已提交
168

169
	start = addr;
L
Linus Torvalds 已提交
170 171 172 173 174
	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
175
		free_pte_range(tlb, pmd, addr);
L
Linus Torvalds 已提交
176 177
	} while (pmd++, addr = next, addr != end);

178 179 180 181 182 183 184
	start &= PUD_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PUD_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
185
	}
186 187 188 189 190
	if (end - 1 > ceiling - 1)
		return;

	pmd = pmd_offset(pud, start);
	pud_clear(pud);
191
	pmd_free_tlb(tlb, pmd, start);
L
Linus Torvalds 已提交
192 193
}

194 195 196
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
197 198 199
{
	pud_t *pud;
	unsigned long next;
200
	unsigned long start;
L
Linus Torvalds 已提交
201

202
	start = addr;
L
Linus Torvalds 已提交
203 204 205 206 207
	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
208
		free_pmd_range(tlb, pud, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
209 210
	} while (pud++, addr = next, addr != end);

211 212 213 214 215 216 217
	start &= PGDIR_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PGDIR_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
218
	}
219 220 221 222 223
	if (end - 1 > ceiling - 1)
		return;

	pud = pud_offset(pgd, start);
	pgd_clear(pgd);
224
	pud_free_tlb(tlb, pud, start);
L
Linus Torvalds 已提交
225 226 227
}

/*
228 229
 * This function frees user-level page tables of a process.
 *
L
Linus Torvalds 已提交
230 231
 * Must be called with pagetable lock held.
 */
232
void free_pgd_range(struct mmu_gather *tlb,
233 234
			unsigned long addr, unsigned long end,
			unsigned long floor, unsigned long ceiling)
L
Linus Torvalds 已提交
235 236 237
{
	pgd_t *pgd;
	unsigned long next;
238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
	unsigned long start;

	/*
	 * 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
265

266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282
	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;

	start = addr;
283
	pgd = pgd_offset(tlb->mm, addr);
L
Linus Torvalds 已提交
284 285 286 287
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
288
		free_pud_range(tlb, pgd, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
289
	} while (pgd++, addr = next, addr != end);
290 291
}

292
void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *vma,
293
		unsigned long floor, unsigned long ceiling)
294 295 296 297 298
{
	while (vma) {
		struct vm_area_struct *next = vma->vm_next;
		unsigned long addr = vma->vm_start;

299
		/*
N
npiggin@suse.de 已提交
300 301
		 * Hide vma from rmap and truncate_pagecache before freeing
		 * pgtables
302 303 304 305
		 */
		anon_vma_unlink(vma);
		unlink_file_vma(vma);

306
		if (is_vm_hugetlb_page(vma)) {
307
			hugetlb_free_pgd_range(tlb, addr, vma->vm_end,
308
				floor, next? next->vm_start: ceiling);
309 310 311 312 313
		} else {
			/*
			 * Optimization: gather nearby vmas into one call down
			 */
			while (next && next->vm_start <= vma->vm_end + PMD_SIZE
314
			       && !is_vm_hugetlb_page(next)) {
315 316
				vma = next;
				next = vma->vm_next;
317 318
				anon_vma_unlink(vma);
				unlink_file_vma(vma);
319 320 321 322
			}
			free_pgd_range(tlb, addr, vma->vm_end,
				floor, next? next->vm_start: ceiling);
		}
323 324
		vma = next;
	}
L
Linus Torvalds 已提交
325 326
}

327
int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
328
{
329
	pgtable_t new = pte_alloc_one(mm, address);
330 331 332
	if (!new)
		return -ENOMEM;

333 334 335 336 337 338 339 340 341 342 343 344 345 346 347
	/*
	 * 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
348
	spin_lock(&mm->page_table_lock);
349
	if (!pmd_present(*pmd)) {	/* Has another populated it ? */
L
Linus Torvalds 已提交
350 351
		mm->nr_ptes++;
		pmd_populate(mm, pmd, new);
352
		new = NULL;
L
Linus Torvalds 已提交
353
	}
H
Hugh Dickins 已提交
354
	spin_unlock(&mm->page_table_lock);
355 356
	if (new)
		pte_free(mm, new);
357
	return 0;
L
Linus Torvalds 已提交
358 359
}

360
int __pte_alloc_kernel(pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
361
{
362 363 364 365
	pte_t *new = pte_alloc_one_kernel(&init_mm, address);
	if (!new)
		return -ENOMEM;

366 367
	smp_wmb(); /* See comment in __pte_alloc */

368
	spin_lock(&init_mm.page_table_lock);
369
	if (!pmd_present(*pmd)) {	/* Has another populated it ? */
370
		pmd_populate_kernel(&init_mm, pmd, new);
371 372
		new = NULL;
	}
373
	spin_unlock(&init_mm.page_table_lock);
374 375
	if (new)
		pte_free_kernel(&init_mm, new);
376
	return 0;
L
Linus Torvalds 已提交
377 378
}

379 380 381 382 383 384 385 386
static inline void add_mm_rss(struct mm_struct *mm, int file_rss, int anon_rss)
{
	if (file_rss)
		add_mm_counter(mm, file_rss, file_rss);
	if (anon_rss)
		add_mm_counter(mm, anon_rss, anon_rss);
}

N
Nick Piggin 已提交
387
/*
388 389 390
 * 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
391 392 393
 *
 * The calling function must still handle the error.
 */
394 395
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
396
{
397 398 399 400 401
	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;
402 403 404 405 406 407 408 409 410 411 412 413 414 415
	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) {
416 417
			printk(KERN_ALERT
				"BUG: Bad page map: %lu messages suppressed\n",
418 419 420 421 422 423 424
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;
425 426 427 428

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

429 430
	printk(KERN_ALERT
		"BUG: Bad page map in process %s  pte:%08llx pmd:%08llx\n",
431 432 433
		current->comm,
		(long long)pte_val(pte), (long long)pmd_val(*pmd));
	if (page) {
434
		printk(KERN_ALERT
435 436 437 438
		"page:%p flags:%p count:%d mapcount:%d mapping:%p index:%lx\n",
		page, (void *)page->flags, page_count(page),
		page_mapcount(page), page->mapping, page->index);
	}
439
	printk(KERN_ALERT
440 441 442 443 444 445
		"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)
446
		print_symbol(KERN_ALERT "vma->vm_ops->fault: %s\n",
447 448
				(unsigned long)vma->vm_ops->fault);
	if (vma->vm_file && vma->vm_file->f_op)
449
		print_symbol(KERN_ALERT "vma->vm_file->f_op->mmap: %s\n",
450
				(unsigned long)vma->vm_file->f_op->mmap);
N
Nick Piggin 已提交
451
	dump_stack();
452
	add_taint(TAINT_BAD_PAGE);
N
Nick Piggin 已提交
453 454
}

455 456 457 458 459
static inline int is_cow_mapping(unsigned int flags)
{
	return (flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
}

H
Hugh Dickins 已提交
460 461 462 463 464 465 466 467 468 469 470 471 472 473
#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
474
/*
N
Nick Piggin 已提交
475
 * vm_normal_page -- This function gets the "struct page" associated with a pte.
476
 *
N
Nick Piggin 已提交
477 478 479
 * "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
480
 *
N
Nick Piggin 已提交
481 482 483 484 485 486 487 488
 * 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.
489
 *
J
Jared Hulbert 已提交
490 491
 * 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
492 493
 * set, and the vm_pgoff will point to the first PFN mapped: thus every special
 * mapping will always honor the rule
494 495 496
 *
 *	pfn_of_page == vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT)
 *
N
Nick Piggin 已提交
497 498 499 500 501 502
 * 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
503 504
 *
 *
N
Nick Piggin 已提交
505
 * In order to support COW of arbitrary special mappings, we have VM_MIXEDMAP.
J
Jared Hulbert 已提交
506 507 508 509 510 511 512 513 514
 *
 * 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
515
 */
N
Nick Piggin 已提交
516 517 518 519 520 521 522
#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
523
{
524
	unsigned long pfn = pte_pfn(pte);
N
Nick Piggin 已提交
525 526

	if (HAVE_PTE_SPECIAL) {
527 528
		if (likely(!pte_special(pte)))
			goto check_pfn;
H
Hugh Dickins 已提交
529 530
		if (vma->vm_flags & (VM_PFNMAP | VM_MIXEDMAP))
			return NULL;
H
Hugh Dickins 已提交
531
		if (!is_zero_pfn(pfn))
532
			print_bad_pte(vma, addr, pte, NULL);
N
Nick Piggin 已提交
533 534 535 536 537
		return NULL;
	}

	/* !HAVE_PTE_SPECIAL case follows: */

J
Jared Hulbert 已提交
538 539 540 541 542 543
	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
544 545
			unsigned long off;
			off = (addr - vma->vm_start) >> PAGE_SHIFT;
J
Jared Hulbert 已提交
546 547 548 549 550
			if (pfn == vma->vm_pgoff + off)
				return NULL;
			if (!is_cow_mapping(vma->vm_flags))
				return NULL;
		}
551 552
	}

H
Hugh Dickins 已提交
553 554
	if (is_zero_pfn(pfn))
		return NULL;
555 556 557 558 559
check_pfn:
	if (unlikely(pfn > highest_memmap_pfn)) {
		print_bad_pte(vma, addr, pte, NULL);
		return NULL;
	}
560 561

	/*
N
Nick Piggin 已提交
562 563
	 * NOTE! We still have PageReserved() pages in the page tables.
	 * eg. VDSO mappings can cause them to exist.
564
	 */
J
Jared Hulbert 已提交
565
out:
566
	return pfn_to_page(pfn);
H
Hugh Dickins 已提交
567 568
}

L
Linus Torvalds 已提交
569 570 571 572 573 574
/*
 * 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
575
static inline unsigned long
L
Linus Torvalds 已提交
576
copy_one_pte(struct mm_struct *dst_mm, struct mm_struct *src_mm,
N
Nick Piggin 已提交
577
		pte_t *dst_pte, pte_t *src_pte, struct vm_area_struct *vma,
H
Hugh Dickins 已提交
578
		unsigned long addr, int *rss)
L
Linus Torvalds 已提交
579
{
N
Nick Piggin 已提交
580
	unsigned long vm_flags = vma->vm_flags;
L
Linus Torvalds 已提交
581 582 583 584 585 586
	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)) {
587 588
			swp_entry_t entry = pte_to_swp_entry(pte);

H
Hugh Dickins 已提交
589 590 591
			if (swap_duplicate(entry) < 0)
				return entry.val;

L
Linus Torvalds 已提交
592 593 594
			/* make sure dst_mm is on swapoff's mmlist. */
			if (unlikely(list_empty(&dst_mm->mmlist))) {
				spin_lock(&mmlist_lock);
595 596 597
				if (list_empty(&dst_mm->mmlist))
					list_add(&dst_mm->mmlist,
						 &src_mm->mmlist);
L
Linus Torvalds 已提交
598 599
				spin_unlock(&mmlist_lock);
			}
600 601 602 603 604 605 606 607 608 609
			if (is_write_migration_entry(entry) &&
					is_cow_mapping(vm_flags)) {
				/*
				 * COW mappings require pages in both parent
				 * and child to be set to read.
				 */
				make_migration_entry_read(&entry);
				pte = swp_entry_to_pte(entry);
				set_pte_at(src_mm, addr, src_pte, pte);
			}
L
Linus Torvalds 已提交
610
		}
611
		goto out_set_pte;
L
Linus Torvalds 已提交
612 613 614 615 616 617
	}

	/*
	 * If it's a COW mapping, write protect it both
	 * in the parent and the child
	 */
618
	if (is_cow_mapping(vm_flags)) {
L
Linus Torvalds 已提交
619
		ptep_set_wrprotect(src_mm, addr, src_pte);
620
		pte = pte_wrprotect(pte);
L
Linus Torvalds 已提交
621 622 623 624 625 626 627 628 629
	}

	/*
	 * 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);
630 631 632 633

	page = vm_normal_page(vma, addr, pte);
	if (page) {
		get_page(page);
H
Hugh Dickins 已提交
634
		page_dup_rmap(page);
635
		rss[PageAnon(page)]++;
636
	}
637 638 639

out_set_pte:
	set_pte_at(dst_mm, addr, dst_pte, pte);
H
Hugh Dickins 已提交
640
	return 0;
L
Linus Torvalds 已提交
641 642 643 644 645 646
}

static 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)
{
647
	pte_t *orig_src_pte, *orig_dst_pte;
L
Linus Torvalds 已提交
648
	pte_t *src_pte, *dst_pte;
H
Hugh Dickins 已提交
649
	spinlock_t *src_ptl, *dst_ptl;
650
	int progress = 0;
H
Hugh Dickins 已提交
651
	int rss[2];
H
Hugh Dickins 已提交
652
	swp_entry_t entry = (swp_entry_t){0};
L
Linus Torvalds 已提交
653 654

again:
655
	rss[1] = rss[0] = 0;
H
Hugh Dickins 已提交
656
	dst_pte = pte_alloc_map_lock(dst_mm, dst_pmd, addr, &dst_ptl);
L
Linus Torvalds 已提交
657 658 659
	if (!dst_pte)
		return -ENOMEM;
	src_pte = pte_offset_map_nested(src_pmd, addr);
H
Hugh Dickins 已提交
660
	src_ptl = pte_lockptr(src_mm, src_pmd);
I
Ingo Molnar 已提交
661
	spin_lock_nested(src_ptl, SINGLE_DEPTH_NESTING);
662 663
	orig_src_pte = src_pte;
	orig_dst_pte = dst_pte;
664
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
665 666 667 668 669 670

	do {
		/*
		 * We are holding two locks at this point - either of them
		 * could generate latencies in another task on another CPU.
		 */
671 672 673
		if (progress >= 32) {
			progress = 0;
			if (need_resched() ||
N
Nick Piggin 已提交
674
			    spin_needbreak(src_ptl) || spin_needbreak(dst_ptl))
675 676
				break;
		}
L
Linus Torvalds 已提交
677 678 679 680
		if (pte_none(*src_pte)) {
			progress++;
			continue;
		}
H
Hugh Dickins 已提交
681 682 683 684
		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
685 686 687
		progress += 8;
	} while (dst_pte++, src_pte++, addr += PAGE_SIZE, addr != end);

688
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
689
	spin_unlock(src_ptl);
690
	pte_unmap_nested(orig_src_pte);
691
	add_mm_rss(dst_mm, rss[0], rss[1]);
692
	pte_unmap_unlock(orig_dst_pte, dst_ptl);
H
Hugh Dickins 已提交
693
	cond_resched();
H
Hugh Dickins 已提交
694 695 696 697 698 699

	if (entry.val) {
		if (add_swap_count_continuation(entry, GFP_KERNEL) < 0)
			return -ENOMEM;
		progress = 0;
	}
L
Linus Torvalds 已提交
700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755
	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);
		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
756
	int ret;
L
Linus Torvalds 已提交
757

758 759 760 761 762 763
	/*
	 * 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.
	 */
764
	if (!(vma->vm_flags & (VM_HUGETLB|VM_NONLINEAR|VM_PFNMAP|VM_INSERTPAGE))) {
765 766 767 768
		if (!vma->anon_vma)
			return 0;
	}

L
Linus Torvalds 已提交
769 770 771
	if (is_vm_hugetlb_page(vma))
		return copy_hugetlb_page_range(dst_mm, src_mm, vma);

772
	if (unlikely(is_pfn_mapping(vma))) {
773 774 775 776 777 778 779 780 781
		/*
		 * 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
782 783 784 785 786 787 788 789 790 791
	/*
	 * 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
792 793 794 795 796 797
	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
798 799 800 801 802
		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
803
	} while (dst_pgd++, src_pgd++, addr = next, addr != end);
A
Andrea Arcangeli 已提交
804 805 806 807 808

	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
809 810
}

811
static unsigned long zap_pte_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
812
				struct vm_area_struct *vma, pmd_t *pmd,
L
Linus Torvalds 已提交
813
				unsigned long addr, unsigned long end,
814
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
815
{
N
Nick Piggin 已提交
816
	struct mm_struct *mm = tlb->mm;
L
Linus Torvalds 已提交
817
	pte_t *pte;
818
	spinlock_t *ptl;
819 820
	int file_rss = 0;
	int anon_rss = 0;
L
Linus Torvalds 已提交
821

822
	pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
823
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
824 825
	do {
		pte_t ptent = *pte;
826 827
		if (pte_none(ptent)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
828
			continue;
829
		}
830 831 832

		(*zap_work) -= PAGE_SIZE;

L
Linus Torvalds 已提交
833
		if (pte_present(ptent)) {
H
Hugh Dickins 已提交
834
			struct page *page;
835

836
			page = vm_normal_page(vma, addr, ptent);
L
Linus Torvalds 已提交
837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854
			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
855
			ptent = ptep_get_and_clear_full(mm, addr, pte,
856
							tlb->fullmm);
L
Linus Torvalds 已提交
857 858 859 860 861 862
			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
863
				set_pte_at(mm, addr, pte,
L
Linus Torvalds 已提交
864 865
					   pgoff_to_pte(page->index));
			if (PageAnon(page))
H
Hugh Dickins 已提交
866
				anon_rss--;
867 868 869
			else {
				if (pte_dirty(ptent))
					set_page_dirty(page);
870 871
				if (pte_young(ptent) &&
				    likely(!VM_SequentialReadHint(vma)))
872
					mark_page_accessed(page);
H
Hugh Dickins 已提交
873
				file_rss--;
874
			}
875
			page_remove_rmap(page);
876 877
			if (unlikely(page_mapcount(page) < 0))
				print_bad_pte(vma, addr, ptent, page);
L
Linus Torvalds 已提交
878 879 880 881 882 883 884 885 886
			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;
887 888 889 890 891 892
		if (pte_file(ptent)) {
			if (unlikely(!(vma->vm_flags & VM_NONLINEAR)))
				print_bad_pte(vma, addr, ptent, NULL);
		} else if
		  (unlikely(!free_swap_and_cache(pte_to_swp_entry(ptent))))
			print_bad_pte(vma, addr, ptent, NULL);
893
		pte_clear_not_present_full(mm, addr, pte, tlb->fullmm);
894
	} while (pte++, addr += PAGE_SIZE, (addr != end && *zap_work > 0));
895

H
Hugh Dickins 已提交
896
	add_mm_rss(mm, file_rss, anon_rss);
897
	arch_leave_lazy_mmu_mode();
898
	pte_unmap_unlock(pte - 1, ptl);
899 900

	return addr;
L
Linus Torvalds 已提交
901 902
}

903
static inline unsigned long zap_pmd_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
904
				struct vm_area_struct *vma, pud_t *pud,
L
Linus Torvalds 已提交
905
				unsigned long addr, unsigned long end,
906
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
907 908 909 910 911 912 913
{
	pmd_t *pmd;
	unsigned long next;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
914 915
		if (pmd_none_or_clear_bad(pmd)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
916
			continue;
917 918 919 920 921 922
		}
		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
923 924
}

925
static inline unsigned long zap_pud_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
926
				struct vm_area_struct *vma, pgd_t *pgd,
L
Linus Torvalds 已提交
927
				unsigned long addr, unsigned long end,
928
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
929 930 931 932 933 934 935
{
	pud_t *pud;
	unsigned long next;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
936 937
		if (pud_none_or_clear_bad(pud)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
938
			continue;
939 940 941 942 943 944
		}
		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
945 946
}

947 948
static unsigned long unmap_page_range(struct mmu_gather *tlb,
				struct vm_area_struct *vma,
L
Linus Torvalds 已提交
949
				unsigned long addr, unsigned long end,
950
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
951 952 953 954 955 956 957 958 959 960 961 962
{
	pgd_t *pgd;
	unsigned long next;

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

	BUG_ON(addr >= end);
	tlb_start_vma(tlb, vma);
	pgd = pgd_offset(vma->vm_mm, addr);
	do {
		next = pgd_addr_end(addr, end);
963 964
		if (pgd_none_or_clear_bad(pgd)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
965
			continue;
966 967 968 969
		}
		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
970
	tlb_end_vma(tlb, vma);
971 972

	return addr;
L
Linus Torvalds 已提交
973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990
}

#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
 *
991
 * Returns the end address of the unmapping (restart addr if interrupted).
L
Linus Torvalds 已提交
992
 *
993
 * Unmap all pages in the vma list.
L
Linus Torvalds 已提交
994
 *
995 996
 * 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
997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
 * 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.
 */
1008
unsigned long unmap_vmas(struct mmu_gather **tlbp,
L
Linus Torvalds 已提交
1009 1010 1011 1012
		struct vm_area_struct *vma, unsigned long start_addr,
		unsigned long end_addr, unsigned long *nr_accounted,
		struct zap_details *details)
{
1013
	long zap_work = ZAP_BLOCK_SIZE;
L
Linus Torvalds 已提交
1014 1015
	unsigned long tlb_start = 0;	/* For tlb_finish_mmu */
	int tlb_start_valid = 0;
1016
	unsigned long start = start_addr;
L
Linus Torvalds 已提交
1017
	spinlock_t *i_mmap_lock = details? details->i_mmap_lock: NULL;
1018
	int fullmm = (*tlbp)->fullmm;
A
Andrea Arcangeli 已提交
1019
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
1020

A
Andrea Arcangeli 已提交
1021
	mmu_notifier_invalidate_range_start(mm, start_addr, end_addr);
L
Linus Torvalds 已提交
1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034
	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;

1035
		if (unlikely(is_pfn_mapping(vma)))
1036 1037
			untrack_pfn_vma(vma, 0, 0);

L
Linus Torvalds 已提交
1038 1039 1040 1041 1042 1043
		while (start != end) {
			if (!tlb_start_valid) {
				tlb_start = start;
				tlb_start_valid = 1;
			}

1044
			if (unlikely(is_vm_hugetlb_page(vma))) {
1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058
				/*
				 * 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) /
1059
					pages_per_huge_page(hstate_vma(vma));
1060 1061
				}

1062 1063 1064 1065 1066 1067 1068 1069
				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
1070 1071 1072 1073 1074
			}

			tlb_finish_mmu(*tlbp, tlb_start, start);

			if (need_resched() ||
N
Nick Piggin 已提交
1075
				(i_mmap_lock && spin_needbreak(i_mmap_lock))) {
L
Linus Torvalds 已提交
1076
				if (i_mmap_lock) {
1077
					*tlbp = NULL;
L
Linus Torvalds 已提交
1078 1079 1080 1081 1082
					goto out;
				}
				cond_resched();
			}

1083
			*tlbp = tlb_gather_mmu(vma->vm_mm, fullmm);
L
Linus Torvalds 已提交
1084
			tlb_start_valid = 0;
1085
			zap_work = ZAP_BLOCK_SIZE;
L
Linus Torvalds 已提交
1086 1087 1088
		}
	}
out:
A
Andrea Arcangeli 已提交
1089
	mmu_notifier_invalidate_range_end(mm, start_addr, end_addr);
1090
	return start;	/* which is now the end (or restart) address */
L
Linus Torvalds 已提交
1091 1092 1093 1094 1095 1096 1097 1098 1099
}

/**
 * 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
 */
1100
unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
L
Linus Torvalds 已提交
1101 1102 1103 1104 1105 1106 1107 1108 1109
		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);
1110
	update_hiwater_rss(mm);
1111 1112 1113
	end = unmap_vmas(&tlb, vma, address, end, &nr_accounted, details);
	if (tlb)
		tlb_finish_mmu(tlb, address, end);
1114
	return end;
L
Linus Torvalds 已提交
1115 1116
}

1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139
/**
 * 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);

L
Linus Torvalds 已提交
1140 1141 1142
/*
 * Do a quick page-table lookup for a single page.
 */
1143
struct page *follow_page(struct vm_area_struct *vma, unsigned long address,
1144
			unsigned int flags)
L
Linus Torvalds 已提交
1145 1146 1147 1148 1149
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *ptep, pte;
1150
	spinlock_t *ptl;
L
Linus Torvalds 已提交
1151
	struct page *page;
1152
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
1153

1154 1155 1156 1157 1158
	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
1159

1160
	page = NULL;
L
Linus Torvalds 已提交
1161 1162
	pgd = pgd_offset(mm, address);
	if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
1163
		goto no_page_table;
L
Linus Torvalds 已提交
1164 1165

	pud = pud_offset(pgd, address);
A
Andi Kleen 已提交
1166
	if (pud_none(*pud))
1167
		goto no_page_table;
A
Andi Kleen 已提交
1168 1169 1170 1171 1172 1173 1174 1175
	if (pud_huge(*pud)) {
		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
1176
	pmd = pmd_offset(pud, address);
1177
	if (pmd_none(*pmd))
1178 1179 1180 1181
		goto no_page_table;
	if (pmd_huge(*pmd)) {
		BUG_ON(flags & FOLL_GET);
		page = follow_huge_pmd(mm, address, pmd, flags & FOLL_WRITE);
L
Linus Torvalds 已提交
1182
		goto out;
1183
	}
1184 1185 1186
	if (unlikely(pmd_bad(*pmd)))
		goto no_page_table;

1187
	ptep = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
1188 1189

	pte = *ptep;
1190
	if (!pte_present(pte))
1191
		goto no_page;
1192 1193
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;
H
Hugh Dickins 已提交
1194

1195
	page = vm_normal_page(vma, address, pte);
H
Hugh Dickins 已提交
1196 1197
	if (unlikely(!page)) {
		if ((flags & FOLL_DUMP) ||
H
Hugh Dickins 已提交
1198
		    !is_zero_pfn(pte_pfn(pte)))
H
Hugh Dickins 已提交
1199 1200 1201
			goto bad_page;
		page = pte_page(pte);
	}
L
Linus Torvalds 已提交
1202

1203 1204 1205 1206 1207 1208
	if (flags & FOLL_GET)
		get_page(page);
	if (flags & FOLL_TOUCH) {
		if ((flags & FOLL_WRITE) &&
		    !pte_dirty(pte) && !PageDirty(page))
			set_page_dirty(page);
1209 1210 1211 1212 1213
		/*
		 * 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().
		 */
1214 1215 1216 1217
		mark_page_accessed(page);
	}
unlock:
	pte_unmap_unlock(ptep, ptl);
L
Linus Torvalds 已提交
1218
out:
1219
	return page;
L
Linus Torvalds 已提交
1220

1221 1222 1223 1224 1225 1226 1227 1228
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
1229

1230 1231 1232
no_page_table:
	/*
	 * When core dumping an enormous anonymous area that nobody
H
Hugh Dickins 已提交
1233 1234 1235 1236 1237
	 * 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.
1238
	 */
H
Hugh Dickins 已提交
1239 1240 1241
	if ((flags & FOLL_DUMP) &&
	    (!vma->vm_ops || !vma->vm_ops->fault))
		return ERR_PTR(-EFAULT);
1242
	return page;
L
Linus Torvalds 已提交
1243 1244
}

N
Nick Piggin 已提交
1245
int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
H
Hugh Dickins 已提交
1246
		     unsigned long start, int nr_pages, unsigned int gup_flags,
1247
		     struct page **pages, struct vm_area_struct **vmas)
L
Linus Torvalds 已提交
1248 1249
{
	int i;
H
Hugh Dickins 已提交
1250
	unsigned long vm_flags;
L
Linus Torvalds 已提交
1251

1252
	if (nr_pages <= 0)
1253
		return 0;
H
Hugh Dickins 已提交
1254 1255 1256

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

L
Linus Torvalds 已提交
1257 1258
	/* 
	 * Require read or write permissions.
H
Hugh Dickins 已提交
1259
	 * If FOLL_FORCE is set, we only require the "MAY" flags.
L
Linus Torvalds 已提交
1260
	 */
H
Hugh Dickins 已提交
1261 1262 1263 1264
	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
1265 1266 1267
	i = 0;

	do {
1268
		struct vm_area_struct *vma;
L
Linus Torvalds 已提交
1269 1270 1271 1272 1273 1274 1275 1276 1277

		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
1278 1279

			/* user gate pages are read-only */
H
Hugh Dickins 已提交
1280
			if (gup_flags & FOLL_WRITE)
L
Linus Torvalds 已提交
1281 1282 1283 1284 1285 1286 1287 1288 1289
				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);
1290 1291
			if (pmd_none(*pmd))
				return i ? : -EFAULT;
L
Linus Torvalds 已提交
1292
			pte = pte_offset_map(pmd, pg);
1293 1294 1295 1296
			if (pte_none(*pte)) {
				pte_unmap(pte);
				return i ? : -EFAULT;
			}
L
Linus Torvalds 已提交
1297
			if (pages) {
1298
				struct page *page = vm_normal_page(gate_vma, start, *pte);
1299 1300 1301
				pages[i] = page;
				if (page)
					get_page(page);
L
Linus Torvalds 已提交
1302 1303 1304 1305 1306 1307
			}
			pte_unmap(pte);
			if (vmas)
				vmas[i] = gate_vma;
			i++;
			start += PAGE_SIZE;
1308
			nr_pages--;
L
Linus Torvalds 已提交
1309 1310 1311
			continue;
		}

N
Nick Piggin 已提交
1312 1313
		if (!vma ||
		    (vma->vm_flags & (VM_IO | VM_PFNMAP)) ||
H
Hugh Dickins 已提交
1314
		    !(vm_flags & vma->vm_flags))
L
Linus Torvalds 已提交
1315 1316
			return i ? : -EFAULT;

H
Hugh Dickins 已提交
1317 1318
		if (is_vm_hugetlb_page(vma)) {
			i = follow_hugetlb_page(mm, vma, pages, vmas,
H
Hugh Dickins 已提交
1319
					&start, &nr_pages, i, gup_flags);
H
Hugh Dickins 已提交
1320 1321
			continue;
		}
1322

L
Linus Torvalds 已提交
1323
		do {
1324
			struct page *page;
H
Hugh Dickins 已提交
1325
			unsigned int foll_flags = gup_flags;
L
Linus Torvalds 已提交
1326

1327
			/*
1328
			 * If we have a pending SIGKILL, don't keep faulting
H
Hugh Dickins 已提交
1329
			 * pages and potentially allocating memory.
1330
			 */
H
Hugh Dickins 已提交
1331
			if (unlikely(fatal_signal_pending(current)))
1332
				return i ? i : -ERESTARTSYS;
1333

1334
			cond_resched();
1335
			while (!(page = follow_page(vma, start, foll_flags))) {
1336
				int ret;
1337

1338 1339 1340 1341
				ret = handle_mm_fault(mm, vma, start,
					(foll_flags & FOLL_WRITE) ?
					FAULT_FLAG_WRITE : 0);

N
Nick Piggin 已提交
1342 1343 1344
				if (ret & VM_FAULT_ERROR) {
					if (ret & VM_FAULT_OOM)
						return i ? i : -ENOMEM;
1345 1346
					if (ret &
					    (VM_FAULT_HWPOISON|VM_FAULT_SIGBUS))
N
Nick Piggin 已提交
1347 1348 1349 1350 1351 1352 1353 1354
						return i ? i : -EFAULT;
					BUG();
				}
				if (ret & VM_FAULT_MAJOR)
					tsk->maj_flt++;
				else
					tsk->min_flt++;

1355
				/*
N
Nick Piggin 已提交
1356 1357 1358 1359
				 * 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
1360 1361 1362 1363 1364 1365
				 * 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).
1366
				 */
1367 1368
				if ((ret & VM_FAULT_WRITE) &&
				    !(vma->vm_flags & VM_WRITE))
1369
					foll_flags &= ~FOLL_WRITE;
N
Nick Piggin 已提交
1370

1371
				cond_resched();
L
Linus Torvalds 已提交
1372
			}
1373 1374
			if (IS_ERR(page))
				return i ? i : PTR_ERR(page);
L
Linus Torvalds 已提交
1375
			if (pages) {
1376
				pages[i] = page;
1377

1378
				flush_anon_page(vma, page, start);
1379
				flush_dcache_page(page);
L
Linus Torvalds 已提交
1380 1381 1382 1383 1384
			}
			if (vmas)
				vmas[i] = vma;
			i++;
			start += PAGE_SIZE;
1385 1386 1387
			nr_pages--;
		} while (nr_pages && start < vma->vm_end);
	} while (nr_pages);
L
Linus Torvalds 已提交
1388 1389
	return i;
}
N
Nick Piggin 已提交
1390

1391 1392 1393 1394 1395
/**
 * get_user_pages() - pin user pages in memory
 * @tsk:	task_struct of target task
 * @mm:		mm_struct of target mm
 * @start:	starting user address
1396
 * @nr_pages:	number of pages from start to pin
1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407
 * @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
1408
 * requested. If nr_pages is 0 or negative, returns 0. If no pages
1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440
 * 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
1441
int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
1442
		unsigned long start, int nr_pages, int write, int force,
N
Nick Piggin 已提交
1443 1444
		struct page **pages, struct vm_area_struct **vmas)
{
H
Hugh Dickins 已提交
1445
	int flags = FOLL_TOUCH;
N
Nick Piggin 已提交
1446

H
Hugh Dickins 已提交
1447 1448
	if (pages)
		flags |= FOLL_GET;
N
Nick Piggin 已提交
1449
	if (write)
H
Hugh Dickins 已提交
1450
		flags |= FOLL_WRITE;
N
Nick Piggin 已提交
1451
	if (force)
H
Hugh Dickins 已提交
1452
		flags |= FOLL_FORCE;
N
Nick Piggin 已提交
1453

1454
	return __get_user_pages(tsk, mm, start, nr_pages, flags, pages, vmas);
N
Nick Piggin 已提交
1455
}
L
Linus Torvalds 已提交
1456 1457
EXPORT_SYMBOL(get_user_pages);

H
Hugh Dickins 已提交
1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478
/**
 * 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,
H
Hugh Dickins 已提交
1479
			FOLL_FORCE | FOLL_DUMP | FOLL_GET, &page, &vma) < 1)
H
Hugh Dickins 已提交
1480 1481 1482 1483 1484 1485
		return NULL;
	flush_cache_page(vma, addr, page_to_pfn(page));
	return page;
}
#endif /* CONFIG_ELF_CORE */

H
Harvey Harrison 已提交
1486 1487
pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
			spinlock_t **ptl)
1488 1489 1490 1491
{
	pgd_t * pgd = pgd_offset(mm, addr);
	pud_t * pud = pud_alloc(mm, pgd, addr);
	if (pud) {
1492
		pmd_t * pmd = pmd_alloc(mm, pud, addr);
1493 1494 1495 1496 1497 1498
		if (pmd)
			return pte_alloc_map_lock(mm, pmd, addr, ptl);
	}
	return NULL;
}

1499 1500 1501 1502 1503 1504 1505
/*
 * 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
1506 1507
static int insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page, pgprot_t prot)
1508
{
N
Nick Piggin 已提交
1509
	struct mm_struct *mm = vma->vm_mm;
1510
	int retval;
1511
	pte_t *pte;
1512 1513
	spinlock_t *ptl;

1514
	retval = -EINVAL;
1515
	if (PageAnon(page))
1516
		goto out;
1517 1518
	retval = -ENOMEM;
	flush_dcache_page(page);
1519
	pte = get_locked_pte(mm, addr, &ptl);
1520
	if (!pte)
1521
		goto out;
1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532
	retval = -EBUSY;
	if (!pte_none(*pte))
		goto out_unlock;

	/* Ok, finally just insert the thing.. */
	get_page(page);
	inc_mm_counter(mm, file_rss);
	page_add_file_rmap(page);
	set_pte_at(mm, addr, pte, mk_pte(page, prot));

	retval = 0;
1533 1534
	pte_unmap_unlock(pte, ptl);
	return retval;
1535 1536 1537 1538 1539 1540
out_unlock:
	pte_unmap_unlock(pte, ptl);
out:
	return retval;
}

1541 1542 1543 1544 1545 1546
/**
 * 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
 *
1547 1548 1549 1550 1551 1552
 * 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
1553
 * (see split_page()).
1554 1555 1556 1557 1558 1559 1560 1561 1562
 *
 * 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
1563 1564
int vm_insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page)
1565 1566 1567 1568 1569
{
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
	if (!page_count(page))
		return -EINVAL;
1570
	vma->vm_flags |= VM_INSERTPAGE;
N
Nick Piggin 已提交
1571
	return insert_page(vma, addr, page, vma->vm_page_prot);
1572
}
1573
EXPORT_SYMBOL(vm_insert_page);
1574

N
Nick Piggin 已提交
1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602
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);
	update_mmu_cache(vma, addr, entry); /* XXX: why not for insert_page? */

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

N
Nick Piggin 已提交
1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613
/**
 * 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
1614 1615 1616 1617 1618
 *
 * 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
1619 1620
 */
int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
N
Nick Piggin 已提交
1621
			unsigned long pfn)
N
Nick Piggin 已提交
1622
{
1623
	int ret;
1624
	pgprot_t pgprot = vma->vm_page_prot;
N
Nick Piggin 已提交
1625 1626 1627 1628 1629 1630
	/*
	 * 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
1631 1632 1633 1634 1635
	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
1636

N
Nick Piggin 已提交
1637 1638
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
1639
	if (track_pfn_vma_new(vma, &pgprot, pfn, PAGE_SIZE))
1640 1641
		return -EINVAL;

1642
	ret = insert_pfn(vma, addr, pfn, pgprot);
1643 1644 1645 1646 1647

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

	return ret;
N
Nick Piggin 已提交
1648 1649
}
EXPORT_SYMBOL(vm_insert_pfn);
N
Nick Piggin 已提交
1650

N
Nick Piggin 已提交
1651 1652 1653 1654
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
1655

N
Nick Piggin 已提交
1656 1657
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
N
Nick Piggin 已提交
1658

N
Nick Piggin 已提交
1659 1660 1661 1662
	/*
	 * 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
1663 1664
	 * 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
1665 1666 1667 1668 1669 1670 1671 1672
	 */
	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
1673
}
N
Nick Piggin 已提交
1674
EXPORT_SYMBOL(vm_insert_mixed);
N
Nick Piggin 已提交
1675

L
Linus Torvalds 已提交
1676 1677 1678 1679 1680 1681 1682 1683 1684 1685
/*
 * 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
1686
	spinlock_t *ptl;
L
Linus Torvalds 已提交
1687

H
Hugh Dickins 已提交
1688
	pte = pte_alloc_map_lock(mm, pmd, addr, &ptl);
L
Linus Torvalds 已提交
1689 1690
	if (!pte)
		return -ENOMEM;
1691
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
1692 1693
	do {
		BUG_ON(!pte_none(*pte));
N
Nick Piggin 已提交
1694
		set_pte_at(mm, addr, pte, pte_mkspecial(pfn_pte(pfn, prot)));
L
Linus Torvalds 已提交
1695 1696
		pfn++;
	} while (pte++, addr += PAGE_SIZE, addr != end);
1697
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
1698
	pte_unmap_unlock(pte - 1, ptl);
L
Linus Torvalds 已提交
1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741
	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;
	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;
}

1742 1743 1744 1745 1746 1747 1748 1749 1750 1751
/**
 * 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
1752 1753 1754 1755 1756
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;
1757
	unsigned long end = addr + PAGE_ALIGN(size);
L
Linus Torvalds 已提交
1758 1759 1760 1761 1762 1763 1764 1765
	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
1766 1767 1768 1769 1770
	 *   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.
1771 1772 1773
	 *   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
1774 1775 1776 1777
	 *
	 * 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
1778
	 */
1779
	if (addr == vma->vm_start && end == vma->vm_end) {
L
Linus Torvalds 已提交
1780
		vma->vm_pgoff = pfn;
1781
		vma->vm_flags |= VM_PFN_AT_MMAP;
1782
	} else if (is_cow_mapping(vma->vm_flags))
1783
		return -EINVAL;
L
Linus Torvalds 已提交
1784

1785
	vma->vm_flags |= VM_IO | VM_RESERVED | VM_PFNMAP;
L
Linus Torvalds 已提交
1786

1787
	err = track_pfn_vma_new(vma, &prot, pfn, PAGE_ALIGN(size));
1788 1789 1790 1791 1792 1793
	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);
1794
		vma->vm_flags &= ~VM_PFN_AT_MMAP;
1795
		return -EINVAL;
1796
	}
1797

L
Linus Torvalds 已提交
1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808
	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);
1809 1810 1811 1812

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

L
Linus Torvalds 已提交
1813 1814 1815 1816
	return err;
}
EXPORT_SYMBOL(remap_pfn_range);

1817 1818 1819 1820 1821 1822
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;
1823
	pgtable_t token;
1824
	spinlock_t *uninitialized_var(ptl);
1825 1826 1827 1828 1829 1830 1831 1832 1833

	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));

1834 1835
	arch_enter_lazy_mmu_mode();

1836
	token = pmd_pgtable(*pmd);
1837 1838

	do {
1839
		err = fn(pte++, token, addr, data);
1840 1841
		if (err)
			break;
1842
	} while (addr += PAGE_SIZE, addr != end);
1843

1844 1845
	arch_leave_lazy_mmu_mode();

1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858
	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
1859 1860
	BUG_ON(pud_huge(*pud));

1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901
	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;
A
Andrea Arcangeli 已提交
1902
	unsigned long start = addr, end = addr + size;
1903 1904 1905
	int err;

	BUG_ON(addr >= end);
A
Andrea Arcangeli 已提交
1906
	mmu_notifier_invalidate_range_start(mm, start, end);
1907 1908 1909 1910 1911 1912 1913
	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);
A
Andrea Arcangeli 已提交
1914
	mmu_notifier_invalidate_range_end(mm, start, end);
1915 1916 1917 1918
	return err;
}
EXPORT_SYMBOL_GPL(apply_to_page_range);

1919 1920 1921 1922 1923 1924 1925 1926 1927
/*
 * 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
1928
static inline int pte_unmap_same(struct mm_struct *mm, pmd_t *pmd,
1929 1930 1931 1932 1933
				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
1934 1935
		spinlock_t *ptl = pte_lockptr(mm, pmd);
		spin_lock(ptl);
1936
		same = pte_same(*page_table, orig_pte);
H
Hugh Dickins 已提交
1937
		spin_unlock(ptl);
1938 1939 1940 1941 1942 1943
	}
#endif
	pte_unmap(page_table);
	return same;
}

L
Linus Torvalds 已提交
1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956
/*
 * Do pte_mkwrite, but only if the vma says VM_WRITE.  We do this when
 * servicing faults for write access.  In the normal case, do always want
 * pte_mkwrite.  But get_user_pages can cause write faults for mappings
 * that do not have writing enabled, when used by access_process_vm.
 */
static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma)
{
	if (likely(vma->vm_flags & VM_WRITE))
		pte = pte_mkwrite(pte);
	return pte;
}

1957
static inline void cow_user_page(struct page *dst, struct page *src, unsigned long va, struct vm_area_struct *vma)
1958 1959 1960 1961 1962 1963 1964 1965 1966
{
	/*
	 * 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
1967 1968 1969 1970 1971 1972 1973 1974 1975
		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))
1976 1977
			memset(kaddr, 0, PAGE_SIZE);
		kunmap_atomic(kaddr, KM_USER0);
1978
		flush_dcache_page(dst);
N
Nick Piggin 已提交
1979 1980
	} else
		copy_user_highpage(dst, src, va, vma);
1981 1982
}

L
Linus Torvalds 已提交
1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
/*
 * 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.
 *
1997 1998 1999
 * 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
2000
 */
2001 2002
static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2003
		spinlock_t *ptl, pte_t orig_pte)
L
Linus Torvalds 已提交
2004
{
2005
	struct page *old_page, *new_page;
L
Linus Torvalds 已提交
2006
	pte_t entry;
N
Nick Piggin 已提交
2007
	int reuse = 0, ret = 0;
2008
	int page_mkwrite = 0;
2009
	struct page *dirty_page = NULL;
L
Linus Torvalds 已提交
2010

2011
	old_page = vm_normal_page(vma, address, orig_pte);
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
	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;
2023
		goto gotten;
2024
	}
L
Linus Torvalds 已提交
2025

2026
	/*
P
Peter Zijlstra 已提交
2027 2028
	 * Take out anonymous pages first, anonymous shared vmas are
	 * not dirty accountable.
2029
	 */
H
Hugh Dickins 已提交
2030
	if (PageAnon(old_page) && !PageKsm(old_page)) {
2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042
		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);
				page_cache_release(old_page);
				goto unlock;
			}
			page_cache_release(old_page);
P
Peter Zijlstra 已提交
2043
		}
2044
		reuse = reuse_swap_page(old_page);
2045
		unlock_page(old_page);
P
Peter Zijlstra 已提交
2046
	} else if (unlikely((vma->vm_flags & (VM_WRITE|VM_SHARED)) ==
2047
					(VM_WRITE|VM_SHARED))) {
P
Peter Zijlstra 已提交
2048 2049 2050 2051 2052
		/*
		 * Only catch write-faults on shared writable pages,
		 * read-only shared pages can get COWed by
		 * get_user_pages(.write=1, .force=1).
		 */
2053
		if (vma->vm_ops && vma->vm_ops->page_mkwrite) {
2054 2055 2056 2057 2058 2059 2060 2061 2062
			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;

2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073
			/*
			 * 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);

2074 2075 2076 2077
			tmp = vma->vm_ops->page_mkwrite(vma, &vmf);
			if (unlikely(tmp &
					(VM_FAULT_ERROR | VM_FAULT_NOPAGE))) {
				ret = tmp;
2078
				goto unwritable_page;
2079
			}
N
Nick Piggin 已提交
2080 2081 2082 2083 2084 2085 2086 2087 2088
			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));
2089 2090 2091 2092 2093 2094 2095 2096 2097

			/*
			 * 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
2098 2099 2100
			if (!pte_same(*page_table, orig_pte)) {
				unlock_page(old_page);
				page_cache_release(old_page);
2101
				goto unlock;
N
Nick Piggin 已提交
2102
			}
2103 2104

			page_mkwrite = 1;
L
Linus Torvalds 已提交
2105
		}
2106 2107
		dirty_page = old_page;
		get_page(dirty_page);
2108 2109 2110 2111
		reuse = 1;
	}

	if (reuse) {
2112
reuse:
2113 2114 2115
		flush_cache_page(vma, address, pte_pfn(orig_pte));
		entry = pte_mkyoung(orig_pte);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
2116
		if (ptep_set_access_flags(vma, address, page_table, entry,1))
2117
			update_mmu_cache(vma, address, entry);
2118 2119
		ret |= VM_FAULT_WRITE;
		goto unlock;
L
Linus Torvalds 已提交
2120 2121 2122 2123 2124
	}

	/*
	 * Ok, we need to copy. Oh, well..
	 */
N
Nick Piggin 已提交
2125
	page_cache_get(old_page);
H
Hugh Dickins 已提交
2126
gotten:
2127
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
2128 2129

	if (unlikely(anon_vma_prepare(vma)))
2130
		goto oom;
H
Hugh Dickins 已提交
2131

H
Hugh Dickins 已提交
2132
	if (is_zero_pfn(pte_pfn(orig_pte))) {
H
Hugh Dickins 已提交
2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143
		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);

N
Nick Piggin 已提交
2144 2145 2146 2147
	/*
	 * Don't let another task, with possibly unlocked vma,
	 * keep the mlocked page.
	 */
2148
	if ((vma->vm_flags & VM_LOCKED) && old_page) {
N
Nick Piggin 已提交
2149 2150 2151 2152
		lock_page(old_page);	/* for LRU manipulation */
		clear_page_mlock(old_page);
		unlock_page(old_page);
	}
2153

K
KAMEZAWA Hiroyuki 已提交
2154
	if (mem_cgroup_newpage_charge(new_page, mm, GFP_KERNEL))
2155 2156
		goto oom_free_new;

L
Linus Torvalds 已提交
2157 2158 2159
	/*
	 * Re-check the pte - we dropped the lock
	 */
2160
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
2161
	if (likely(pte_same(*page_table, orig_pte))) {
H
Hugh Dickins 已提交
2162 2163 2164 2165 2166 2167
		if (old_page) {
			if (!PageAnon(old_page)) {
				dec_mm_counter(mm, file_rss);
				inc_mm_counter(mm, anon_rss);
			}
		} else
2168
			inc_mm_counter(mm, anon_rss);
2169
		flush_cache_page(vma, address, pte_pfn(orig_pte));
2170 2171
		entry = mk_pte(new_page, vma->vm_page_prot);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
2172 2173 2174 2175 2176 2177
		/*
		 * 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.
		 */
2178
		ptep_clear_flush(vma, address, page_table);
N
Nick Piggin 已提交
2179
		page_add_new_anon_rmap(new_page, vma, address);
2180 2181 2182 2183 2184 2185
		/*
		 * 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);
2186
		update_mmu_cache(vma, address, entry);
N
Nick Piggin 已提交
2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209
		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.
			 */
2210
			page_remove_rmap(old_page);
N
Nick Piggin 已提交
2211 2212
		}

L
Linus Torvalds 已提交
2213 2214
		/* Free the old page.. */
		new_page = old_page;
N
Nick Piggin 已提交
2215
		ret |= VM_FAULT_WRITE;
2216 2217 2218
	} else
		mem_cgroup_uncharge_page(new_page);

H
Hugh Dickins 已提交
2219 2220 2221 2222
	if (new_page)
		page_cache_release(new_page);
	if (old_page)
		page_cache_release(old_page);
2223
unlock:
2224
	pte_unmap_unlock(page_table, ptl);
2225
	if (dirty_page) {
2226 2227 2228 2229 2230 2231 2232 2233
		/*
		 * 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.
		 */
N
Nick Piggin 已提交
2234 2235 2236 2237
		if (!page_mkwrite) {
			wait_on_page_locked(dirty_page);
			set_page_dirty_balance(dirty_page, page_mkwrite);
		}
2238
		put_page(dirty_page);
N
Nick Piggin 已提交
2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256
		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);
2257
	}
N
Nick Piggin 已提交
2258
	return ret;
2259
oom_free_new:
2260
	page_cache_release(new_page);
2261
oom:
N
Nick Piggin 已提交
2262 2263 2264 2265 2266
	if (old_page) {
		if (page_mkwrite) {
			unlock_page(old_page);
			page_cache_release(old_page);
		}
H
Hugh Dickins 已提交
2267
		page_cache_release(old_page);
N
Nick Piggin 已提交
2268
	}
L
Linus Torvalds 已提交
2269
	return VM_FAULT_OOM;
2270 2271 2272

unwritable_page:
	page_cache_release(old_page);
2273
	return ret;
L
Linus Torvalds 已提交
2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300
}

/*
 * 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
2301
 * large vma, note the restart_addr from unmap_vmas when it breaks out:
L
Linus Torvalds 已提交
2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327
 * 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;

2328 2329
	/*
	 * files that support invalidating or truncating portions of the
N
Nick Piggin 已提交
2330
	 * file from under mmaped areas must have their ->fault function
N
Nick Piggin 已提交
2331 2332
	 * return a locked page (and set VM_FAULT_LOCKED in the return).
	 * This provides synchronisation against concurrent unmapping here.
2333 2334
	 */

L
Linus Torvalds 已提交
2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345
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;
		}
	}

2346 2347
	restart_addr = zap_page_range(vma, start_addr,
					end_addr - start_addr, details);
N
Nick Piggin 已提交
2348
	need_break = need_resched() || spin_needbreak(details->i_mmap_lock);
L
Linus Torvalds 已提交
2349

2350
	if (restart_addr >= end_addr) {
L
Linus Torvalds 已提交
2351 2352 2353 2354 2355 2356
		/* 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 */
2357
		vma->vm_truncate_count = restart_addr;
L
Linus Torvalds 已提交
2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423
		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;
	}
}

/**
2424
 * 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
2425
 * @mapping: the address space containing mmaps to be unmapped.
L
Linus Torvalds 已提交
2426 2427
 * @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
2428
 * boundary.  Note that this is different from truncate_pagecache(), which
L
Linus Torvalds 已提交
2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461
 * 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);

2462
	/* Protect against endless unmapping loops */
L
Linus Torvalds 已提交
2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478
	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);

2479 2480 2481 2482 2483 2484 2485 2486 2487
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
2488
	if (!inode->i_op->truncate_range)
2489 2490
		return -ENOSYS;

2491
	mutex_lock(&inode->i_mutex);
2492 2493 2494
	down_write(&inode->i_alloc_sem);
	unmap_mapping_range(mapping, offset, (end - offset), 1);
	truncate_inode_pages_range(mapping, offset, end);
2495
	unmap_mapping_range(mapping, offset, (end - offset), 1);
2496 2497
	inode->i_op->truncate_range(inode, offset, end);
	up_write(&inode->i_alloc_sem);
2498
	mutex_unlock(&inode->i_mutex);
2499 2500 2501 2502

	return 0;
}

L
Linus Torvalds 已提交
2503
/*
2504 2505 2506
 * 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
2507
 */
2508 2509
static int do_swap_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2510
		unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
2511
{
2512
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2513
	struct page *page;
2514
	swp_entry_t entry;
L
Linus Torvalds 已提交
2515
	pte_t pte;
2516
	struct mem_cgroup *ptr = NULL;
N
Nick Piggin 已提交
2517
	int ret = 0;
L
Linus Torvalds 已提交
2518

H
Hugh Dickins 已提交
2519
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
2520
		goto out;
2521 2522

	entry = pte_to_swp_entry(orig_pte);
2523 2524 2525 2526 2527 2528 2529
	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
2530
			ret = VM_FAULT_SIGBUS;
2531
		}
2532 2533
		goto out;
	}
2534
	delayacct_set_flag(DELAYACCT_PF_SWAPIN);
L
Linus Torvalds 已提交
2535 2536
	page = lookup_swap_cache(entry);
	if (!page) {
H
Hugh Dickins 已提交
2537
		grab_swap_token(mm); /* Contend for token _before_ read-in */
2538 2539
		page = swapin_readahead(entry,
					GFP_HIGHUSER_MOVABLE, vma, address);
L
Linus Torvalds 已提交
2540 2541
		if (!page) {
			/*
2542 2543
			 * Back out if somebody else faulted in this pte
			 * while we released the pte lock.
L
Linus Torvalds 已提交
2544
			 */
2545
			page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
2546 2547
			if (likely(pte_same(*page_table, orig_pte)))
				ret = VM_FAULT_OOM;
2548
			delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2549
			goto unlock;
L
Linus Torvalds 已提交
2550 2551 2552 2553
		}

		/* Had to read the page from swap area: Major fault */
		ret = VM_FAULT_MAJOR;
2554
		count_vm_event(PGMAJFAULT);
2555 2556 2557
	} else if (PageHWPoison(page)) {
		ret = VM_FAULT_HWPOISON;
		delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2558
		goto out_release;
L
Linus Torvalds 已提交
2559 2560
	}

2561 2562 2563
	lock_page(page);
	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);

H
Hugh Dickins 已提交
2564 2565 2566 2567 2568 2569
	page = ksm_might_need_to_copy(page, vma, address);
	if (!page) {
		ret = VM_FAULT_OOM;
		goto out;
	}

K
KAMEZAWA Hiroyuki 已提交
2570
	if (mem_cgroup_try_charge_swapin(mm, page, GFP_KERNEL, &ptr)) {
2571
		ret = VM_FAULT_OOM;
2572
		goto out_page;
2573 2574
	}

L
Linus Torvalds 已提交
2575
	/*
2576
	 * Back out if somebody else already faulted in this pte.
L
Linus Torvalds 已提交
2577
	 */
2578
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
H
Hugh Dickins 已提交
2579
	if (unlikely(!pte_same(*page_table, orig_pte)))
2580 2581 2582 2583 2584
		goto out_nomap;

	if (unlikely(!PageUptodate(page))) {
		ret = VM_FAULT_SIGBUS;
		goto out_nomap;
L
Linus Torvalds 已提交
2585 2586
	}

2587 2588 2589 2590 2591 2592 2593 2594
	/*
	 * 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.
2595 2596 2597 2598
	 * 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().
2599
	 */
L
Linus Torvalds 已提交
2600

2601
	inc_mm_counter(mm, anon_rss);
L
Linus Torvalds 已提交
2602
	pte = mk_pte(page, vma->vm_page_prot);
2603
	if ((flags & FAULT_FLAG_WRITE) && reuse_swap_page(page)) {
L
Linus Torvalds 已提交
2604
		pte = maybe_mkwrite(pte_mkdirty(pte), vma);
2605
		flags &= ~FAULT_FLAG_WRITE;
L
Linus Torvalds 已提交
2606 2607 2608 2609
	}
	flush_icache_page(vma, page);
	set_pte_at(mm, address, page_table, pte);
	page_add_anon_rmap(page, vma, address);
2610 2611
	/* It's better to call commit-charge after rmap is established */
	mem_cgroup_commit_charge_swapin(page, ptr);
L
Linus Torvalds 已提交
2612

2613
	swap_free(entry);
N
Nick Piggin 已提交
2614
	if (vm_swap_full() || (vma->vm_flags & VM_LOCKED) || PageMlocked(page))
2615
		try_to_free_swap(page);
2616 2617
	unlock_page(page);

2618
	if (flags & FAULT_FLAG_WRITE) {
2619 2620 2621
		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
2622 2623 2624 2625 2626
		goto out;
	}

	/* No need to invalidate - it was non-present before */
	update_mmu_cache(vma, address, pte);
2627
unlock:
2628
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
2629 2630
out:
	return ret;
2631
out_nomap:
2632
	mem_cgroup_cancel_charge_swapin(ptr);
2633
	pte_unmap_unlock(page_table, ptl);
2634
out_page:
2635
	unlock_page(page);
2636
out_release:
2637
	page_cache_release(page);
2638
	return ret;
L
Linus Torvalds 已提交
2639 2640 2641
}

/*
2642 2643 2644
 * 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
2645
 */
2646 2647
static int do_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2648
		unsigned int flags)
L
Linus Torvalds 已提交
2649
{
2650 2651
	struct page *page;
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2652 2653
	pte_t entry;

H
Hugh Dickins 已提交
2654 2655 2656
	if (!(flags & FAULT_FLAG_WRITE)) {
		entry = pte_mkspecial(pfn_pte(my_zero_pfn(address),
						vma->vm_page_prot));
H
Hugh Dickins 已提交
2657 2658 2659 2660 2661 2662 2663
		ptl = pte_lockptr(mm, pmd);
		spin_lock(ptl);
		if (!pte_none(*page_table))
			goto unlock;
		goto setpte;
	}

N
Nick Piggin 已提交
2664 2665
	/* Allocate our own private page. */
	pte_unmap(page_table);
2666

N
Nick Piggin 已提交
2667 2668 2669 2670 2671
	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
2672
	__SetPageUptodate(page);
2673

K
KAMEZAWA Hiroyuki 已提交
2674
	if (mem_cgroup_newpage_charge(page, mm, GFP_KERNEL))
2675 2676
		goto oom_free_page;

N
Nick Piggin 已提交
2677
	entry = mk_pte(page, vma->vm_page_prot);
H
Hugh Dickins 已提交
2678 2679
	if (vma->vm_flags & VM_WRITE)
		entry = pte_mkwrite(pte_mkdirty(entry));
L
Linus Torvalds 已提交
2680

N
Nick Piggin 已提交
2681
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
2682
	if (!pte_none(*page_table))
N
Nick Piggin 已提交
2683
		goto release;
H
Hugh Dickins 已提交
2684

N
Nick Piggin 已提交
2685 2686
	inc_mm_counter(mm, anon_rss);
	page_add_new_anon_rmap(page, vma, address);
H
Hugh Dickins 已提交
2687
setpte:
2688
	set_pte_at(mm, address, page_table, entry);
L
Linus Torvalds 已提交
2689 2690

	/* No need to invalidate - it was non-present before */
2691 2692
	update_mmu_cache(vma, address, entry);
unlock:
2693
	pte_unmap_unlock(page_table, ptl);
N
Nick Piggin 已提交
2694
	return 0;
2695
release:
2696
	mem_cgroup_uncharge_page(page);
2697 2698
	page_cache_release(page);
	goto unlock;
2699
oom_free_page:
2700
	page_cache_release(page);
2701
oom:
L
Linus Torvalds 已提交
2702 2703 2704 2705
	return VM_FAULT_OOM;
}

/*
2706
 * __do_fault() tries to create a new page mapping. It aggressively
L
Linus Torvalds 已提交
2707
 * tries to share with existing pages, but makes a separate copy if
2708 2709
 * the FAULT_FLAG_WRITE is set in the flags parameter in order to avoid
 * the next page fault.
L
Linus Torvalds 已提交
2710 2711 2712 2713
 *
 * As this is called only for pages that do not currently exist, we
 * do not need to flush old virtual caches or the TLB.
 *
2714
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
2715
 * but allow concurrent faults), and pte neither mapped nor locked.
2716
 * We return with mmap_sem still held, but pte unmapped and unlocked.
L
Linus Torvalds 已提交
2717
 */
2718
static int __do_fault(struct mm_struct *mm, struct vm_area_struct *vma,
2719
		unsigned long address, pmd_t *pmd,
2720
		pgoff_t pgoff, unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
2721
{
2722
	pte_t *page_table;
2723
	spinlock_t *ptl;
N
Nick Piggin 已提交
2724
	struct page *page;
L
Linus Torvalds 已提交
2725 2726
	pte_t entry;
	int anon = 0;
2727
	int charged = 0;
2728
	struct page *dirty_page = NULL;
N
Nick Piggin 已提交
2729 2730
	struct vm_fault vmf;
	int ret;
2731
	int page_mkwrite = 0;
2732

N
Nick Piggin 已提交
2733 2734 2735 2736
	vmf.virtual_address = (void __user *)(address & PAGE_MASK);
	vmf.pgoff = pgoff;
	vmf.flags = flags;
	vmf.page = NULL;
L
Linus Torvalds 已提交
2737

N
Nick Piggin 已提交
2738 2739 2740
	ret = vma->vm_ops->fault(vma, &vmf);
	if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE)))
		return ret;
L
Linus Torvalds 已提交
2741

2742 2743 2744 2745 2746 2747
	if (unlikely(PageHWPoison(vmf.page))) {
		if (ret & VM_FAULT_LOCKED)
			unlock_page(vmf.page);
		return VM_FAULT_HWPOISON;
	}

2748
	/*
N
Nick Piggin 已提交
2749
	 * For consistency in subsequent calls, make the faulted page always
2750 2751
	 * locked.
	 */
N
Nick Piggin 已提交
2752
	if (unlikely(!(ret & VM_FAULT_LOCKED)))
N
Nick Piggin 已提交
2753
		lock_page(vmf.page);
2754
	else
N
Nick Piggin 已提交
2755
		VM_BUG_ON(!PageLocked(vmf.page));
2756

L
Linus Torvalds 已提交
2757 2758 2759
	/*
	 * Should we do an early C-O-W break?
	 */
N
Nick Piggin 已提交
2760
	page = vmf.page;
2761
	if (flags & FAULT_FLAG_WRITE) {
2762
		if (!(vma->vm_flags & VM_SHARED)) {
2763
			anon = 1;
2764
			if (unlikely(anon_vma_prepare(vma))) {
N
Nick Piggin 已提交
2765
				ret = VM_FAULT_OOM;
2766
				goto out;
2767
			}
N
Nick Piggin 已提交
2768 2769
			page = alloc_page_vma(GFP_HIGHUSER_MOVABLE,
						vma, address);
2770
			if (!page) {
N
Nick Piggin 已提交
2771
				ret = VM_FAULT_OOM;
2772
				goto out;
2773
			}
K
KAMEZAWA Hiroyuki 已提交
2774
			if (mem_cgroup_newpage_charge(page, mm, GFP_KERNEL)) {
2775 2776 2777 2778 2779
				ret = VM_FAULT_OOM;
				page_cache_release(page);
				goto out;
			}
			charged = 1;
N
Nick Piggin 已提交
2780 2781 2782 2783 2784 2785
			/*
			 * Don't let another task, with possibly unlocked vma,
			 * keep the mlocked page.
			 */
			if (vma->vm_flags & VM_LOCKED)
				clear_page_mlock(vmf.page);
N
Nick Piggin 已提交
2786
			copy_user_highpage(page, vmf.page, address, vma);
N
Nick Piggin 已提交
2787
			__SetPageUptodate(page);
2788
		} else {
2789 2790
			/*
			 * If the page will be shareable, see if the backing
2791
			 * address space wants to know that the page is about
2792 2793
			 * to become writable
			 */
2794
			if (vma->vm_ops->page_mkwrite) {
2795 2796
				int tmp;

2797
				unlock_page(page);
N
Nick Piggin 已提交
2798
				vmf.flags = FAULT_FLAG_WRITE|FAULT_FLAG_MKWRITE;
2799 2800 2801 2802
				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
2803
					goto unwritable_page;
N
Nick Piggin 已提交
2804
				}
N
Nick Piggin 已提交
2805 2806 2807 2808 2809 2810 2811 2812 2813
				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));
2814
				page_mkwrite = 1;
2815 2816
			}
		}
2817

L
Linus Torvalds 已提交
2818 2819
	}

2820
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
2821 2822 2823 2824 2825 2826

	/*
	 * This silly early PAGE_DIRTY setting removes a race
	 * due to the bad i386 page protection. But it's valid
	 * for other architectures too.
	 *
2827
	 * Note that if FAULT_FLAG_WRITE is set, we either now have
L
Linus Torvalds 已提交
2828 2829 2830 2831 2832
	 * 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... */
2833
	if (likely(pte_same(*page_table, orig_pte))) {
2834 2835
		flush_icache_page(vma, page);
		entry = mk_pte(page, vma->vm_page_prot);
2836
		if (flags & FAULT_FLAG_WRITE)
L
Linus Torvalds 已提交
2837 2838
			entry = maybe_mkwrite(pte_mkdirty(entry), vma);
		if (anon) {
2839 2840
			inc_mm_counter(mm, anon_rss);
			page_add_new_anon_rmap(page, vma, address);
2841
		} else {
2842
			inc_mm_counter(mm, file_rss);
2843
			page_add_file_rmap(page);
2844
			if (flags & FAULT_FLAG_WRITE) {
2845
				dirty_page = page;
2846 2847
				get_page(dirty_page);
			}
2848
		}
2849
		set_pte_at(mm, address, page_table, entry);
2850 2851 2852

		/* no need to invalidate: a not-present page won't be cached */
		update_mmu_cache(vma, address, entry);
L
Linus Torvalds 已提交
2853
	} else {
2854 2855
		if (charged)
			mem_cgroup_uncharge_page(page);
2856 2857 2858
		if (anon)
			page_cache_release(page);
		else
2859
			anon = 1; /* no anon but release faulted_page */
L
Linus Torvalds 已提交
2860 2861
	}

2862
	pte_unmap_unlock(page_table, ptl);
2863 2864

out:
N
Nick Piggin 已提交
2865 2866
	if (dirty_page) {
		struct address_space *mapping = page->mapping;
2867

N
Nick Piggin 已提交
2868 2869 2870
		if (set_page_dirty(dirty_page))
			page_mkwrite = 1;
		unlock_page(dirty_page);
2871
		put_page(dirty_page);
N
Nick Piggin 已提交
2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886
		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);
2887
	}
2888

N
Nick Piggin 已提交
2889
	return ret;
N
Nick Piggin 已提交
2890 2891 2892 2893

unwritable_page:
	page_cache_release(page);
	return ret;
2894
}
2895

2896 2897
static int do_linear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2898
		unsigned int flags, pte_t orig_pte)
2899 2900
{
	pgoff_t pgoff = (((address & PAGE_MASK)
2901
			- vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
2902

2903 2904
	pte_unmap(page_table);
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
2905 2906
}

L
Linus Torvalds 已提交
2907 2908 2909 2910
/*
 * Fault of a previously existing named mapping. Repopulate the pte
 * from the encoded file_pte if possible. This enables swappable
 * nonlinear vmas.
2911 2912 2913 2914
 *
 * 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
2915
 */
N
Nick Piggin 已提交
2916
static int do_nonlinear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
2917
		unsigned long address, pte_t *page_table, pmd_t *pmd,
2918
		unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
2919
{
2920
	pgoff_t pgoff;
L
Linus Torvalds 已提交
2921

2922 2923
	flags |= FAULT_FLAG_NONLINEAR;

H
Hugh Dickins 已提交
2924
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
N
Nick Piggin 已提交
2925
		return 0;
L
Linus Torvalds 已提交
2926

2927
	if (unlikely(!(vma->vm_flags & VM_NONLINEAR))) {
2928 2929 2930
		/*
		 * Page table corrupted: show pte and kill process.
		 */
2931
		print_bad_pte(vma, address, orig_pte, NULL);
H
Hugh Dickins 已提交
2932
		return VM_FAULT_SIGBUS;
2933 2934 2935
	}

	pgoff = pte_to_pgoff(orig_pte);
2936
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
L
Linus Torvalds 已提交
2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947
}

/*
 * 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
2948 2949 2950
 * 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
2951 2952
 */
static inline int handle_pte_fault(struct mm_struct *mm,
2953
		struct vm_area_struct *vma, unsigned long address,
2954
		pte_t *pte, pmd_t *pmd, unsigned int flags)
L
Linus Torvalds 已提交
2955 2956
{
	pte_t entry;
2957
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2958

2959
	entry = *pte;
L
Linus Torvalds 已提交
2960
	if (!pte_present(entry)) {
2961
		if (pte_none(entry)) {
J
Jes Sorensen 已提交
2962
			if (vma->vm_ops) {
N
Nick Piggin 已提交
2963
				if (likely(vma->vm_ops->fault))
2964
					return do_linear_fault(mm, vma, address,
2965
						pte, pmd, flags, entry);
J
Jes Sorensen 已提交
2966 2967
			}
			return do_anonymous_page(mm, vma, address,
2968
						 pte, pmd, flags);
2969
		}
L
Linus Torvalds 已提交
2970
		if (pte_file(entry))
N
Nick Piggin 已提交
2971
			return do_nonlinear_fault(mm, vma, address,
2972
					pte, pmd, flags, entry);
2973
		return do_swap_page(mm, vma, address,
2974
					pte, pmd, flags, entry);
L
Linus Torvalds 已提交
2975 2976
	}

H
Hugh Dickins 已提交
2977
	ptl = pte_lockptr(mm, pmd);
2978 2979 2980
	spin_lock(ptl);
	if (unlikely(!pte_same(*pte, entry)))
		goto unlock;
2981
	if (flags & FAULT_FLAG_WRITE) {
L
Linus Torvalds 已提交
2982
		if (!pte_write(entry))
2983 2984
			return do_wp_page(mm, vma, address,
					pte, pmd, ptl, entry);
L
Linus Torvalds 已提交
2985 2986 2987
		entry = pte_mkdirty(entry);
	}
	entry = pte_mkyoung(entry);
2988
	if (ptep_set_access_flags(vma, address, pte, entry, flags & FAULT_FLAG_WRITE)) {
2989 2990 2991 2992 2993 2994 2995 2996
		update_mmu_cache(vma, address, entry);
	} 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.
		 */
2997
		if (flags & FAULT_FLAG_WRITE)
2998 2999
			flush_tlb_page(vma, address);
	}
3000 3001
unlock:
	pte_unmap_unlock(pte, ptl);
N
Nick Piggin 已提交
3002
	return 0;
L
Linus Torvalds 已提交
3003 3004 3005 3006 3007
}

/*
 * By the time we get here, we already hold the mm semaphore
 */
N
Nick Piggin 已提交
3008
int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
3009
		unsigned long address, unsigned int flags)
L
Linus Torvalds 已提交
3010 3011 3012 3013 3014 3015 3016 3017
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

	__set_current_state(TASK_RUNNING);

3018
	count_vm_event(PGFAULT);
L
Linus Torvalds 已提交
3019

3020
	if (unlikely(is_vm_hugetlb_page(vma)))
3021
		return hugetlb_fault(mm, vma, address, flags);
L
Linus Torvalds 已提交
3022 3023 3024 3025

	pgd = pgd_offset(mm, address);
	pud = pud_alloc(mm, pgd, address);
	if (!pud)
H
Hugh Dickins 已提交
3026
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
3027 3028
	pmd = pmd_alloc(mm, pud, address);
	if (!pmd)
H
Hugh Dickins 已提交
3029
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
3030 3031
	pte = pte_alloc_map(mm, pmd, address);
	if (!pte)
H
Hugh Dickins 已提交
3032
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
3033

3034
	return handle_pte_fault(mm, vma, address, pte, pmd, flags);
L
Linus Torvalds 已提交
3035 3036 3037 3038 3039
}

#ifndef __PAGETABLE_PUD_FOLDED
/*
 * Allocate page upper directory.
H
Hugh Dickins 已提交
3040
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
3041
 */
3042
int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
L
Linus Torvalds 已提交
3043
{
H
Hugh Dickins 已提交
3044 3045
	pud_t *new = pud_alloc_one(mm, address);
	if (!new)
3046
		return -ENOMEM;
L
Linus Torvalds 已提交
3047

3048 3049
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
3050
	spin_lock(&mm->page_table_lock);
3051
	if (pgd_present(*pgd))		/* Another has populated it */
3052
		pud_free(mm, new);
3053 3054
	else
		pgd_populate(mm, pgd, new);
H
Hugh Dickins 已提交
3055
	spin_unlock(&mm->page_table_lock);
3056
	return 0;
L
Linus Torvalds 已提交
3057 3058 3059 3060 3061 3062
}
#endif /* __PAGETABLE_PUD_FOLDED */

#ifndef __PAGETABLE_PMD_FOLDED
/*
 * Allocate page middle directory.
H
Hugh Dickins 已提交
3063
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
3064
 */
3065
int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
L
Linus Torvalds 已提交
3066
{
H
Hugh Dickins 已提交
3067 3068
	pmd_t *new = pmd_alloc_one(mm, address);
	if (!new)
3069
		return -ENOMEM;
L
Linus Torvalds 已提交
3070

3071 3072
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
3073
	spin_lock(&mm->page_table_lock);
L
Linus Torvalds 已提交
3074
#ifndef __ARCH_HAS_4LEVEL_HACK
3075
	if (pud_present(*pud))		/* Another has populated it */
3076
		pmd_free(mm, new);
3077 3078
	else
		pud_populate(mm, pud, new);
L
Linus Torvalds 已提交
3079
#else
3080
	if (pgd_present(*pud))		/* Another has populated it */
3081
		pmd_free(mm, new);
3082 3083
	else
		pgd_populate(mm, pud, new);
L
Linus Torvalds 已提交
3084
#endif /* __ARCH_HAS_4LEVEL_HACK */
H
Hugh Dickins 已提交
3085
	spin_unlock(&mm->page_table_lock);
3086
	return 0;
3087
}
L
Linus Torvalds 已提交
3088 3089 3090 3091 3092 3093 3094 3095 3096
#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
3097
		return -ENOMEM;
L
Linus Torvalds 已提交
3098
	write = (vma->vm_flags & VM_WRITE) != 0;
3099 3100
	BUG_ON(addr >= end);
	BUG_ON(end > vma->vm_end);
3101
	len = DIV_ROUND_UP(end, PAGE_SIZE) - addr/PAGE_SIZE;
L
Linus Torvalds 已提交
3102 3103
	ret = get_user_pages(current, current->mm, addr,
			len, write, 0, NULL, NULL);
3104
	if (ret < 0)
L
Linus Torvalds 已提交
3105
		return ret;
3106
	return ret == len ? 0 : -EFAULT;
L
Linus Torvalds 已提交
3107 3108 3109 3110 3111
}

#if !defined(__HAVE_ARCH_GATE_AREA)

#if defined(AT_SYSINFO_EHDR)
3112
static struct vm_area_struct gate_vma;
L
Linus Torvalds 已提交
3113 3114 3115 3116 3117 3118

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
3119 3120
	gate_vma.vm_flags = VM_READ | VM_MAYREAD | VM_EXEC | VM_MAYEXEC;
	gate_vma.vm_page_prot = __P101;
3121 3122 3123 3124 3125 3126 3127
	/*
	 * 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
3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151
	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 */
3152

J
Johannes Weiner 已提交
3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189
static int follow_pte(struct mm_struct *mm, unsigned long address,
		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);
	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;
}

J
Johannes Weiner 已提交
3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218
/**
 * 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);

3219
#ifdef CONFIG_HAVE_IOREMAP_PROT
3220 3221 3222
int follow_phys(struct vm_area_struct *vma,
		unsigned long address, unsigned int flags,
		unsigned long *prot, resource_size_t *phys)
3223
{
3224
	int ret = -EINVAL;
3225 3226 3227
	pte_t *ptep, pte;
	spinlock_t *ptl;

3228 3229
	if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
		goto out;
3230

3231
	if (follow_pte(vma->vm_mm, address, &ptep, &ptl))
3232
		goto out;
3233
	pte = *ptep;
3234

3235 3236 3237 3238
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;

	*prot = pgprot_val(pte_pgprot(pte));
3239
	*phys = (resource_size_t)pte_pfn(pte) << PAGE_SHIFT;
3240

3241
	ret = 0;
3242 3243 3244
unlock:
	pte_unmap_unlock(ptep, ptl);
out:
3245
	return ret;
3246 3247 3248 3249 3250 3251 3252
}

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
3253
	void __iomem *maddr;
3254 3255
	int offset = addr & (PAGE_SIZE-1);

3256
	if (follow_phys(vma, addr, write, &prot, &phys_addr))
3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269
		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

3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285
/*
 * 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
3286
	/* ignore errors, just check how much was successfully transferred */
3287 3288 3289
	while (len) {
		int bytes, ret, offset;
		void *maddr;
3290
		struct page *page = NULL;
3291 3292 3293

		ret = get_user_pages(tsk, mm, addr, 1,
				write, 1, &page, &vma);
3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309
		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;
3310
		} else {
3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326
			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);
3327 3328 3329 3330 3331 3332 3333 3334 3335 3336
		}
		len -= bytes;
		buf += bytes;
		addr += bytes;
	}
	up_read(&mm->mmap_sem);
	mmput(mm);

	return buf - old_buf;
}
3337 3338 3339 3340 3341 3342 3343 3344 3345

/*
 * 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;

3346 3347 3348 3349 3350 3351 3352
	/*
	 * Do not print if we are in atomic
	 * contexts (in exception stacks, etc.):
	 */
	if (preempt_count())
		return;

3353 3354 3355 3356 3357 3358 3359 3360
	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;

3361
			p = d_path(&f->f_path, buf, PAGE_SIZE);
3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374
			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);
}
3375 3376 3377 3378

#ifdef CONFIG_PROVE_LOCKING
void might_fault(void)
{
3379 3380 3381 3382 3383 3384 3385 3386 3387
	/*
	 * 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;

3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398
	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