memory.c 82.6 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 48 49
/*
 *  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>
#include <linux/rmap.h>
#include <linux/module.h>
50
#include <linux/delayacct.h>
L
Linus Torvalds 已提交
51
#include <linux/init.h>
P
Peter Zijlstra 已提交
52
#include <linux/writeback.h>
53
#include <linux/memcontrol.h>
A
Andrea Arcangeli 已提交
54
#include <linux/mmu_notifier.h>
L
Linus Torvalds 已提交
55 56 57 58 59 60 61 62 63 64

#include <asm/pgalloc.h>
#include <asm/uaccess.h>
#include <asm/tlb.h>
#include <asm/tlbflush.h>
#include <asm/pgtable.h>

#include <linux/swapops.h>
#include <linux/elf.h>

65 66
#include "internal.h"

A
Andy Whitcroft 已提交
67
#ifndef CONFIG_NEED_MULTIPLE_NODES
L
Linus Torvalds 已提交
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88
/* 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);

89 90 91 92 93 94 95 96 97 98 99 100
/*
 * 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
101 102 103 104

static int __init disable_randmaps(char *s)
{
	randomize_va_space = 0;
105
	return 1;
106 107 108 109
}
__setup("norandmaps", disable_randmaps);


L
Linus Torvalds 已提交
110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137
/*
 * 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.
 */
138
static void free_pte_range(struct mmu_gather *tlb, pmd_t *pmd)
L
Linus Torvalds 已提交
139
{
140
	pgtable_t token = pmd_pgtable(*pmd);
141
	pmd_clear(pmd);
142
	pte_free_tlb(tlb, token);
143
	tlb->mm->nr_ptes--;
L
Linus Torvalds 已提交
144 145
}

146 147 148
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
149 150 151
{
	pmd_t *pmd;
	unsigned long next;
152
	unsigned long start;
L
Linus Torvalds 已提交
153

154
	start = addr;
L
Linus Torvalds 已提交
155 156 157 158 159
	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
160
		free_pte_range(tlb, pmd);
L
Linus Torvalds 已提交
161 162
	} while (pmd++, addr = next, addr != end);

163 164 165 166 167 168 169
	start &= PUD_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PUD_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
170
	}
171 172 173 174 175 176
	if (end - 1 > ceiling - 1)
		return;

	pmd = pmd_offset(pud, start);
	pud_clear(pud);
	pmd_free_tlb(tlb, pmd);
L
Linus Torvalds 已提交
177 178
}

179 180 181
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
182 183 184
{
	pud_t *pud;
	unsigned long next;
185
	unsigned long start;
L
Linus Torvalds 已提交
186

187
	start = addr;
L
Linus Torvalds 已提交
188 189 190 191 192
	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
193
		free_pmd_range(tlb, pud, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
194 195
	} while (pud++, addr = next, addr != end);

196 197 198 199 200 201 202
	start &= PGDIR_MASK;
	if (start < floor)
		return;
	if (ceiling) {
		ceiling &= PGDIR_MASK;
		if (!ceiling)
			return;
L
Linus Torvalds 已提交
203
	}
204 205 206 207 208 209
	if (end - 1 > ceiling - 1)
		return;

	pud = pud_offset(pgd, start);
	pgd_clear(pgd);
	pud_free_tlb(tlb, pud);
L
Linus Torvalds 已提交
210 211 212
}

/*
213 214
 * This function frees user-level page tables of a process.
 *
L
Linus Torvalds 已提交
215 216
 * Must be called with pagetable lock held.
 */
217
void free_pgd_range(struct mmu_gather *tlb,
218 219
			unsigned long addr, unsigned long end,
			unsigned long floor, unsigned long ceiling)
L
Linus Torvalds 已提交
220 221 222
{
	pgd_t *pgd;
	unsigned long next;
223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249
	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
250

251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267
	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;
268
	pgd = pgd_offset(tlb->mm, addr);
L
Linus Torvalds 已提交
269 270 271 272
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
273
		free_pud_range(tlb, pgd, addr, next, floor, ceiling);
L
Linus Torvalds 已提交
274
	} while (pgd++, addr = next, addr != end);
275 276
}

277
void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *vma,
278
		unsigned long floor, unsigned long ceiling)
279 280 281 282 283
{
	while (vma) {
		struct vm_area_struct *next = vma->vm_next;
		unsigned long addr = vma->vm_start;

284 285 286 287 288 289
		/*
		 * Hide vma from rmap and vmtruncate before freeing pgtables
		 */
		anon_vma_unlink(vma);
		unlink_file_vma(vma);

290
		if (is_vm_hugetlb_page(vma)) {
291
			hugetlb_free_pgd_range(tlb, addr, vma->vm_end,
292
				floor, next? next->vm_start: ceiling);
293 294 295 296 297
		} else {
			/*
			 * Optimization: gather nearby vmas into one call down
			 */
			while (next && next->vm_start <= vma->vm_end + PMD_SIZE
298
			       && !is_vm_hugetlb_page(next)) {
299 300
				vma = next;
				next = vma->vm_next;
301 302
				anon_vma_unlink(vma);
				unlink_file_vma(vma);
303 304 305 306
			}
			free_pgd_range(tlb, addr, vma->vm_end,
				floor, next? next->vm_start: ceiling);
		}
307 308
		vma = next;
	}
L
Linus Torvalds 已提交
309 310
}

311
int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
312
{
313
	pgtable_t new = pte_alloc_one(mm, address);
314 315 316
	if (!new)
		return -ENOMEM;

317 318 319 320 321 322 323 324 325 326 327 328 329 330 331
	/*
	 * 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
332
	spin_lock(&mm->page_table_lock);
333
	if (!pmd_present(*pmd)) {	/* Has another populated it ? */
L
Linus Torvalds 已提交
334 335
		mm->nr_ptes++;
		pmd_populate(mm, pmd, new);
336
		new = NULL;
L
Linus Torvalds 已提交
337
	}
H
Hugh Dickins 已提交
338
	spin_unlock(&mm->page_table_lock);
339 340
	if (new)
		pte_free(mm, new);
341
	return 0;
L
Linus Torvalds 已提交
342 343
}

344
int __pte_alloc_kernel(pmd_t *pmd, unsigned long address)
L
Linus Torvalds 已提交
345
{
346 347 348 349
	pte_t *new = pte_alloc_one_kernel(&init_mm, address);
	if (!new)
		return -ENOMEM;

350 351
	smp_wmb(); /* See comment in __pte_alloc */

352
	spin_lock(&init_mm.page_table_lock);
353
	if (!pmd_present(*pmd)) {	/* Has another populated it ? */
354
		pmd_populate_kernel(&init_mm, pmd, new);
355 356
		new = NULL;
	}
357
	spin_unlock(&init_mm.page_table_lock);
358 359
	if (new)
		pte_free_kernel(&init_mm, new);
360
	return 0;
L
Linus Torvalds 已提交
361 362
}

363 364 365 366 367 368 369 370
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
371
/*
372 373 374
 * 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
375 376 377
 *
 * The calling function must still handle the error.
 */
378 379
static void print_bad_pte(struct vm_area_struct *vma, pte_t pte,
			  unsigned long vaddr)
N
Nick Piggin 已提交
380 381 382 383 384 385 386 387 388
{
	printk(KERN_ERR "Bad pte = %08llx, process = %s, "
			"vm_flags = %lx, vaddr = %lx\n",
		(long long)pte_val(pte),
		(vma->vm_mm == current->mm ? current->comm : "???"),
		vma->vm_flags, vaddr);
	dump_stack();
}

389 390 391 392 393
static inline int is_cow_mapping(unsigned int flags)
{
	return (flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
}

H
Hugh Dickins 已提交
394
/*
N
Nick Piggin 已提交
395
 * vm_normal_page -- This function gets the "struct page" associated with a pte.
396
 *
N
Nick Piggin 已提交
397 398 399
 * "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
400
 *
N
Nick Piggin 已提交
401 402 403 404 405 406 407 408
 * 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.
409
 *
J
Jared Hulbert 已提交
410 411
 * 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
412 413
 * set, and the vm_pgoff will point to the first PFN mapped: thus every special
 * mapping will always honor the rule
414 415 416
 *
 *	pfn_of_page == vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT)
 *
N
Nick Piggin 已提交
417 418 419 420 421 422
 * 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
423 424
 *
 *
N
Nick Piggin 已提交
425
 * In order to support COW of arbitrary special mappings, we have VM_MIXEDMAP.
J
Jared Hulbert 已提交
426 427 428 429 430 431 432 433 434
 *
 * 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
435
 */
N
Nick Piggin 已提交
436 437 438 439 440 441 442
#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
443
{
N
Nick Piggin 已提交
444 445 446 447 448 449 450 451 452 453 454 455 456 457
	unsigned long pfn;

	if (HAVE_PTE_SPECIAL) {
		if (likely(!pte_special(pte))) {
			VM_BUG_ON(!pfn_valid(pte_pfn(pte)));
			return pte_page(pte);
		}
		VM_BUG_ON(!(vma->vm_flags & (VM_PFNMAP | VM_MIXEDMAP)));
		return NULL;
	}

	/* !HAVE_PTE_SPECIAL case follows: */

	pfn = pte_pfn(pte);
458

J
Jared Hulbert 已提交
459 460 461 462 463 464
	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
465 466
			unsigned long off;
			off = (addr - vma->vm_start) >> PAGE_SHIFT;
J
Jared Hulbert 已提交
467 468 469 470 471
			if (pfn == vma->vm_pgoff + off)
				return NULL;
			if (!is_cow_mapping(vma->vm_flags))
				return NULL;
		}
472 473
	}

N
Nick Piggin 已提交
474
	VM_BUG_ON(!pfn_valid(pfn));
475 476

	/*
N
Nick Piggin 已提交
477
	 * NOTE! We still have PageReserved() pages in the page tables.
478
	 *
N
Nick Piggin 已提交
479
	 * eg. VDSO mappings can cause them to exist.
480
	 */
J
Jared Hulbert 已提交
481
out:
482
	return pfn_to_page(pfn);
H
Hugh Dickins 已提交
483 484
}

L
Linus Torvalds 已提交
485 486 487 488 489 490
/*
 * 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
491
static inline void
L
Linus Torvalds 已提交
492
copy_one_pte(struct mm_struct *dst_mm, struct mm_struct *src_mm,
N
Nick Piggin 已提交
493
		pte_t *dst_pte, pte_t *src_pte, struct vm_area_struct *vma,
H
Hugh Dickins 已提交
494
		unsigned long addr, int *rss)
L
Linus Torvalds 已提交
495
{
N
Nick Piggin 已提交
496
	unsigned long vm_flags = vma->vm_flags;
L
Linus Torvalds 已提交
497 498 499 500 501 502
	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)) {
503 504 505
			swp_entry_t entry = pte_to_swp_entry(pte);

			swap_duplicate(entry);
L
Linus Torvalds 已提交
506 507 508
			/* make sure dst_mm is on swapoff's mmlist. */
			if (unlikely(list_empty(&dst_mm->mmlist))) {
				spin_lock(&mmlist_lock);
509 510 511
				if (list_empty(&dst_mm->mmlist))
					list_add(&dst_mm->mmlist,
						 &src_mm->mmlist);
L
Linus Torvalds 已提交
512 513
				spin_unlock(&mmlist_lock);
			}
514 515 516 517 518 519 520 521 522 523
			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
524
		}
525
		goto out_set_pte;
L
Linus Torvalds 已提交
526 527 528 529 530 531
	}

	/*
	 * If it's a COW mapping, write protect it both
	 * in the parent and the child
	 */
532
	if (is_cow_mapping(vm_flags)) {
L
Linus Torvalds 已提交
533
		ptep_set_wrprotect(src_mm, addr, src_pte);
534
		pte = pte_wrprotect(pte);
L
Linus Torvalds 已提交
535 536 537 538 539 540 541 542 543
	}

	/*
	 * 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);
544 545 546 547

	page = vm_normal_page(vma, addr, pte);
	if (page) {
		get_page(page);
N
Nick Piggin 已提交
548
		page_dup_rmap(page, vma, addr);
549 550
		rss[!!PageAnon(page)]++;
	}
551 552 553

out_set_pte:
	set_pte_at(dst_mm, addr, dst_pte, pte);
L
Linus Torvalds 已提交
554 555 556 557 558 559 560
}

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)
{
	pte_t *src_pte, *dst_pte;
H
Hugh Dickins 已提交
561
	spinlock_t *src_ptl, *dst_ptl;
562
	int progress = 0;
H
Hugh Dickins 已提交
563
	int rss[2];
L
Linus Torvalds 已提交
564 565

again:
566
	rss[1] = rss[0] = 0;
H
Hugh Dickins 已提交
567
	dst_pte = pte_alloc_map_lock(dst_mm, dst_pmd, addr, &dst_ptl);
L
Linus Torvalds 已提交
568 569 570
	if (!dst_pte)
		return -ENOMEM;
	src_pte = pte_offset_map_nested(src_pmd, addr);
H
Hugh Dickins 已提交
571
	src_ptl = pte_lockptr(src_mm, src_pmd);
I
Ingo Molnar 已提交
572
	spin_lock_nested(src_ptl, SINGLE_DEPTH_NESTING);
573
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
574 575 576 577 578 579

	do {
		/*
		 * We are holding two locks at this point - either of them
		 * could generate latencies in another task on another CPU.
		 */
580 581 582
		if (progress >= 32) {
			progress = 0;
			if (need_resched() ||
N
Nick Piggin 已提交
583
			    spin_needbreak(src_ptl) || spin_needbreak(dst_ptl))
584 585
				break;
		}
L
Linus Torvalds 已提交
586 587 588 589
		if (pte_none(*src_pte)) {
			progress++;
			continue;
		}
H
Hugh Dickins 已提交
590
		copy_one_pte(dst_mm, src_mm, dst_pte, src_pte, vma, addr, rss);
L
Linus Torvalds 已提交
591 592 593
		progress += 8;
	} while (dst_pte++, src_pte++, addr += PAGE_SIZE, addr != end);

594
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
595
	spin_unlock(src_ptl);
L
Linus Torvalds 已提交
596
	pte_unmap_nested(src_pte - 1);
597
	add_mm_rss(dst_mm, rss[0], rss[1]);
H
Hugh Dickins 已提交
598 599
	pte_unmap_unlock(dst_pte - 1, dst_ptl);
	cond_resched();
L
Linus Torvalds 已提交
600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655
	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
656
	int ret;
L
Linus Torvalds 已提交
657

658 659 660 661 662 663
	/*
	 * 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.
	 */
664
	if (!(vma->vm_flags & (VM_HUGETLB|VM_NONLINEAR|VM_PFNMAP|VM_INSERTPAGE))) {
665 666 667 668
		if (!vma->anon_vma)
			return 0;
	}

L
Linus Torvalds 已提交
669 670 671
	if (is_vm_hugetlb_page(vma))
		return copy_hugetlb_page_range(dst_mm, src_mm, vma);

672
	if (unlikely(is_pfn_mapping(vma))) {
673 674 675 676 677 678 679 680 681
		/*
		 * 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
682 683 684 685 686 687 688 689 690 691
	/*
	 * 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
692 693 694 695 696 697
	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
698 699 700 701 702
		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
703
	} while (dst_pgd++, src_pgd++, addr = next, addr != end);
A
Andrea Arcangeli 已提交
704 705 706 707 708

	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
709 710
}

711
static unsigned long zap_pte_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
712
				struct vm_area_struct *vma, pmd_t *pmd,
L
Linus Torvalds 已提交
713
				unsigned long addr, unsigned long end,
714
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
715
{
N
Nick Piggin 已提交
716
	struct mm_struct *mm = tlb->mm;
L
Linus Torvalds 已提交
717
	pte_t *pte;
718
	spinlock_t *ptl;
719 720
	int file_rss = 0;
	int anon_rss = 0;
L
Linus Torvalds 已提交
721

722
	pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
723
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
724 725
	do {
		pte_t ptent = *pte;
726 727
		if (pte_none(ptent)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
728
			continue;
729
		}
730 731 732

		(*zap_work) -= PAGE_SIZE;

L
Linus Torvalds 已提交
733
		if (pte_present(ptent)) {
H
Hugh Dickins 已提交
734
			struct page *page;
735

736
			page = vm_normal_page(vma, addr, ptent);
L
Linus Torvalds 已提交
737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754
			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
755
			ptent = ptep_get_and_clear_full(mm, addr, pte,
756
							tlb->fullmm);
L
Linus Torvalds 已提交
757 758 759 760 761 762
			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
763
				set_pte_at(mm, addr, pte,
L
Linus Torvalds 已提交
764 765
					   pgoff_to_pte(page->index));
			if (PageAnon(page))
H
Hugh Dickins 已提交
766
				anon_rss--;
767 768 769 770
			else {
				if (pte_dirty(ptent))
					set_page_dirty(page);
				if (pte_young(ptent))
771
					SetPageReferenced(page);
H
Hugh Dickins 已提交
772
				file_rss--;
773
			}
N
Nick Piggin 已提交
774
			page_remove_rmap(page, vma);
L
Linus Torvalds 已提交
775 776 777 778 779 780 781 782 783 784 785
			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;
		if (!pte_file(ptent))
			free_swap_and_cache(pte_to_swp_entry(ptent));
786
		pte_clear_not_present_full(mm, addr, pte, tlb->fullmm);
787
	} while (pte++, addr += PAGE_SIZE, (addr != end && *zap_work > 0));
788

H
Hugh Dickins 已提交
789
	add_mm_rss(mm, file_rss, anon_rss);
790
	arch_leave_lazy_mmu_mode();
791
	pte_unmap_unlock(pte - 1, ptl);
792 793

	return addr;
L
Linus Torvalds 已提交
794 795
}

796
static inline unsigned long zap_pmd_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
797
				struct vm_area_struct *vma, pud_t *pud,
L
Linus Torvalds 已提交
798
				unsigned long addr, unsigned long end,
799
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
800 801 802 803 804 805 806
{
	pmd_t *pmd;
	unsigned long next;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
807 808
		if (pmd_none_or_clear_bad(pmd)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
809
			continue;
810 811 812 813 814 815
		}
		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
816 817
}

818
static inline unsigned long zap_pud_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
819
				struct vm_area_struct *vma, pgd_t *pgd,
L
Linus Torvalds 已提交
820
				unsigned long addr, unsigned long end,
821
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
822 823 824 825 826 827 828
{
	pud_t *pud;
	unsigned long next;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
829 830
		if (pud_none_or_clear_bad(pud)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
831
			continue;
832 833 834 835 836 837
		}
		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
838 839
}

840 841
static unsigned long unmap_page_range(struct mmu_gather *tlb,
				struct vm_area_struct *vma,
L
Linus Torvalds 已提交
842
				unsigned long addr, unsigned long end,
843
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
844 845 846 847 848 849 850 851 852 853 854 855
{
	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);
856 857
		if (pgd_none_or_clear_bad(pgd)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
858
			continue;
859 860 861 862
		}
		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
863
	tlb_end_vma(tlb, vma);
864 865

	return addr;
L
Linus Torvalds 已提交
866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883
}

#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
 *
884
 * Returns the end address of the unmapping (restart addr if interrupted).
L
Linus Torvalds 已提交
885
 *
886
 * Unmap all pages in the vma list.
L
Linus Torvalds 已提交
887
 *
888 889
 * 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
890 891 892 893 894 895 896 897 898 899 900
 * 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.
 */
901
unsigned long unmap_vmas(struct mmu_gather **tlbp,
L
Linus Torvalds 已提交
902 903 904 905
		struct vm_area_struct *vma, unsigned long start_addr,
		unsigned long end_addr, unsigned long *nr_accounted,
		struct zap_details *details)
{
906
	long zap_work = ZAP_BLOCK_SIZE;
L
Linus Torvalds 已提交
907 908
	unsigned long tlb_start = 0;	/* For tlb_finish_mmu */
	int tlb_start_valid = 0;
909
	unsigned long start = start_addr;
L
Linus Torvalds 已提交
910
	spinlock_t *i_mmap_lock = details? details->i_mmap_lock: NULL;
911
	int fullmm = (*tlbp)->fullmm;
A
Andrea Arcangeli 已提交
912
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
913

A
Andrea Arcangeli 已提交
914
	mmu_notifier_invalidate_range_start(mm, start_addr, end_addr);
L
Linus Torvalds 已提交
915 916 917 918 919 920 921 922 923 924 925 926 927
	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;

928
		if (unlikely(is_pfn_mapping(vma)))
929 930
			untrack_pfn_vma(vma, 0, 0);

L
Linus Torvalds 已提交
931 932 933 934 935 936
		while (start != end) {
			if (!tlb_start_valid) {
				tlb_start = start;
				tlb_start_valid = 1;
			}

937
			if (unlikely(is_vm_hugetlb_page(vma))) {
938 939 940 941 942 943 944 945 946 947 948 949 950 951
				/*
				 * 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) /
952
					pages_per_huge_page(hstate_vma(vma));
953 954
				}

955 956 957 958 959 960 961 962
				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
963 964 965 966 967
			}

			tlb_finish_mmu(*tlbp, tlb_start, start);

			if (need_resched() ||
N
Nick Piggin 已提交
968
				(i_mmap_lock && spin_needbreak(i_mmap_lock))) {
L
Linus Torvalds 已提交
969
				if (i_mmap_lock) {
970
					*tlbp = NULL;
L
Linus Torvalds 已提交
971 972 973 974 975
					goto out;
				}
				cond_resched();
			}

976
			*tlbp = tlb_gather_mmu(vma->vm_mm, fullmm);
L
Linus Torvalds 已提交
977
			tlb_start_valid = 0;
978
			zap_work = ZAP_BLOCK_SIZE;
L
Linus Torvalds 已提交
979 980 981
		}
	}
out:
A
Andrea Arcangeli 已提交
982
	mmu_notifier_invalidate_range_end(mm, start_addr, end_addr);
983
	return start;	/* which is now the end (or restart) address */
L
Linus Torvalds 已提交
984 985 986 987 988 989 990 991 992
}

/**
 * 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
 */
993
unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
L
Linus Torvalds 已提交
994 995 996 997 998 999 1000 1001 1002
		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);
1003
	update_hiwater_rss(mm);
1004 1005 1006
	end = unmap_vmas(&tlb, vma, address, end, &nr_accounted, details);
	if (tlb)
		tlb_finish_mmu(tlb, address, end);
1007
	return end;
L
Linus Torvalds 已提交
1008 1009
}

1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032
/**
 * 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
1033 1034 1035
/*
 * Do a quick page-table lookup for a single page.
 */
1036
struct page *follow_page(struct vm_area_struct *vma, unsigned long address,
1037
			unsigned int flags)
L
Linus Torvalds 已提交
1038 1039 1040 1041 1042
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *ptep, pte;
1043
	spinlock_t *ptl;
L
Linus Torvalds 已提交
1044
	struct page *page;
1045
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
1046

1047 1048 1049 1050 1051
	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
1052

1053
	page = NULL;
L
Linus Torvalds 已提交
1054 1055
	pgd = pgd_offset(mm, address);
	if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
1056
		goto no_page_table;
L
Linus Torvalds 已提交
1057 1058

	pud = pud_offset(pgd, address);
A
Andi Kleen 已提交
1059
	if (pud_none(*pud))
1060
		goto no_page_table;
A
Andi Kleen 已提交
1061 1062 1063 1064 1065 1066 1067 1068
	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
1069
	pmd = pmd_offset(pud, address);
1070
	if (pmd_none(*pmd))
1071 1072 1073 1074
		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
1075
		goto out;
1076
	}
1077 1078 1079
	if (unlikely(pmd_bad(*pmd)))
		goto no_page_table;

1080
	ptep = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
1081 1082

	pte = *ptep;
1083
	if (!pte_present(pte))
1084
		goto no_page;
1085 1086
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;
1087 1088
	page = vm_normal_page(vma, address, pte);
	if (unlikely(!page))
1089
		goto bad_page;
L
Linus Torvalds 已提交
1090

1091 1092 1093 1094 1095 1096 1097 1098 1099 1100
	if (flags & FOLL_GET)
		get_page(page);
	if (flags & FOLL_TOUCH) {
		if ((flags & FOLL_WRITE) &&
		    !pte_dirty(pte) && !PageDirty(page))
			set_page_dirty(page);
		mark_page_accessed(page);
	}
unlock:
	pte_unmap_unlock(ptep, ptl);
L
Linus Torvalds 已提交
1101
out:
1102
	return page;
L
Linus Torvalds 已提交
1103

1104 1105 1106 1107 1108 1109 1110 1111 1112
bad_page:
	pte_unmap_unlock(ptep, ptl);
	return ERR_PTR(-EFAULT);

no_page:
	pte_unmap_unlock(ptep, ptl);
	if (!pte_none(pte))
		return page;
	/* Fall through to ZERO_PAGE handling */
1113 1114 1115 1116 1117 1118
no_page_table:
	/*
	 * When core dumping an enormous anonymous area that nobody
	 * has touched so far, we don't want to allocate page tables.
	 */
	if (flags & FOLL_ANON) {
N
Nick Piggin 已提交
1119
		page = ZERO_PAGE(0);
1120 1121 1122 1123 1124
		if (flags & FOLL_GET)
			get_page(page);
		BUG_ON(flags & FOLL_WRITE);
	}
	return page;
L
Linus Torvalds 已提交
1125 1126
}

1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139
/* Can we do the FOLL_ANON optimization? */
static inline int use_zero_page(struct vm_area_struct *vma)
{
	/*
	 * We don't want to optimize FOLL_ANON for make_pages_present()
	 * when it tries to page in a VM_LOCKED region. As to VM_SHARED,
	 * we want to get the page from the page tables to make sure
	 * that we serialize and update with any other user of that
	 * mapping.
	 */
	if (vma->vm_flags & (VM_LOCKED | VM_SHARED))
		return 0;
	/*
N
Nick Piggin 已提交
1140
	 * And if we have a fault routine, it's not an anonymous region.
1141
	 */
N
Nick Piggin 已提交
1142
	return !vma->vm_ops || !vma->vm_ops->fault;
1143 1144
}

N
Nick Piggin 已提交
1145 1146 1147 1148


int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
		     unsigned long start, int len, int flags,
L
Linus Torvalds 已提交
1149 1150 1151
		struct page **pages, struct vm_area_struct **vmas)
{
	int i;
N
Nick Piggin 已提交
1152 1153 1154 1155
	unsigned int vm_flags = 0;
	int write = !!(flags & GUP_FLAGS_WRITE);
	int force = !!(flags & GUP_FLAGS_FORCE);
	int ignore = !!(flags & GUP_FLAGS_IGNORE_VMA_PERMISSIONS);
L
Linus Torvalds 已提交
1156

1157 1158
	if (len <= 0)
		return 0;
L
Linus Torvalds 已提交
1159 1160 1161 1162
	/* 
	 * Require read or write permissions.
	 * If 'force' is set, we only require the "MAY" flags.
	 */
1163 1164
	vm_flags  = write ? (VM_WRITE | VM_MAYWRITE) : (VM_READ | VM_MAYREAD);
	vm_flags &= force ? (VM_MAYREAD | VM_MAYWRITE) : (VM_READ | VM_WRITE);
L
Linus Torvalds 已提交
1165 1166 1167
	i = 0;

	do {
1168 1169
		struct vm_area_struct *vma;
		unsigned int foll_flags;
L
Linus Torvalds 已提交
1170 1171 1172 1173 1174 1175 1176 1177 1178

		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
1179 1180 1181

			/* user gate pages are read-only */
			if (!ignore && write)
L
Linus Torvalds 已提交
1182 1183 1184 1185 1186 1187 1188 1189 1190
				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);
1191 1192
			if (pmd_none(*pmd))
				return i ? : -EFAULT;
L
Linus Torvalds 已提交
1193
			pte = pte_offset_map(pmd, pg);
1194 1195 1196 1197
			if (pte_none(*pte)) {
				pte_unmap(pte);
				return i ? : -EFAULT;
			}
L
Linus Torvalds 已提交
1198
			if (pages) {
1199
				struct page *page = vm_normal_page(gate_vma, start, *pte);
1200 1201 1202
				pages[i] = page;
				if (page)
					get_page(page);
L
Linus Torvalds 已提交
1203 1204 1205 1206 1207 1208 1209 1210 1211 1212
			}
			pte_unmap(pte);
			if (vmas)
				vmas[i] = gate_vma;
			i++;
			start += PAGE_SIZE;
			len--;
			continue;
		}

N
Nick Piggin 已提交
1213 1214 1215
		if (!vma ||
		    (vma->vm_flags & (VM_IO | VM_PFNMAP)) ||
		    (!ignore && !(vm_flags & vma->vm_flags)))
L
Linus Torvalds 已提交
1216 1217 1218 1219
			return i ? : -EFAULT;

		if (is_vm_hugetlb_page(vma)) {
			i = follow_hugetlb_page(mm, vma, pages, vmas,
1220
						&start, &len, i, write);
L
Linus Torvalds 已提交
1221 1222
			continue;
		}
1223 1224 1225 1226

		foll_flags = FOLL_TOUCH;
		if (pages)
			foll_flags |= FOLL_GET;
1227
		if (!write && use_zero_page(vma))
1228 1229
			foll_flags |= FOLL_ANON;

L
Linus Torvalds 已提交
1230
		do {
1231
			struct page *page;
L
Linus Torvalds 已提交
1232

1233 1234 1235 1236 1237 1238
			/*
			 * If tsk is ooming, cut off its access to large memory
			 * allocations. It has a pending SIGKILL, but it can't
			 * be processed until returning to user space.
			 */
			if (unlikely(test_tsk_thread_flag(tsk, TIF_MEMDIE)))
1239
				return i ? i : -ENOMEM;
1240

1241 1242
			if (write)
				foll_flags |= FOLL_WRITE;
1243

1244
			cond_resched();
1245
			while (!(page = follow_page(vma, start, foll_flags))) {
1246
				int ret;
N
Nick Piggin 已提交
1247
				ret = handle_mm_fault(mm, vma, start,
1248
						foll_flags & FOLL_WRITE);
N
Nick Piggin 已提交
1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260
				if (ret & VM_FAULT_ERROR) {
					if (ret & VM_FAULT_OOM)
						return i ? i : -ENOMEM;
					else if (ret & VM_FAULT_SIGBUS)
						return i ? i : -EFAULT;
					BUG();
				}
				if (ret & VM_FAULT_MAJOR)
					tsk->maj_flt++;
				else
					tsk->min_flt++;

1261
				/*
N
Nick Piggin 已提交
1262 1263 1264 1265 1266
				 * 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
				 * page lookups as if they were reads.
1267 1268
				 */
				if (ret & VM_FAULT_WRITE)
1269
					foll_flags &= ~FOLL_WRITE;
N
Nick Piggin 已提交
1270

1271
				cond_resched();
L
Linus Torvalds 已提交
1272
			}
1273 1274
			if (IS_ERR(page))
				return i ? i : PTR_ERR(page);
L
Linus Torvalds 已提交
1275
			if (pages) {
1276
				pages[i] = page;
1277

1278
				flush_anon_page(vma, page, start);
1279
				flush_dcache_page(page);
L
Linus Torvalds 已提交
1280 1281 1282 1283 1284 1285
			}
			if (vmas)
				vmas[i] = vma;
			i++;
			start += PAGE_SIZE;
			len--;
1286 1287
		} while (len && start < vma->vm_end);
	} while (len);
L
Linus Torvalds 已提交
1288 1289
	return i;
}
N
Nick Piggin 已提交
1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306

int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
		unsigned long start, int len, int write, int force,
		struct page **pages, struct vm_area_struct **vmas)
{
	int flags = 0;

	if (write)
		flags |= GUP_FLAGS_WRITE;
	if (force)
		flags |= GUP_FLAGS_FORCE;

	return __get_user_pages(tsk, mm,
				start, len, flags,
				pages, vmas);
}

L
Linus Torvalds 已提交
1307 1308
EXPORT_SYMBOL(get_user_pages);

H
Harvey Harrison 已提交
1309 1310
pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
			spinlock_t **ptl)
1311 1312 1313 1314
{
	pgd_t * pgd = pgd_offset(mm, addr);
	pud_t * pud = pud_alloc(mm, pgd, addr);
	if (pud) {
1315
		pmd_t * pmd = pmd_alloc(mm, pud, addr);
1316 1317 1318 1319 1320 1321
		if (pmd)
			return pte_alloc_map_lock(mm, pmd, addr, ptl);
	}
	return NULL;
}

1322 1323 1324 1325 1326 1327 1328
/*
 * 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
1329 1330
static int insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page, pgprot_t prot)
1331
{
N
Nick Piggin 已提交
1332
	struct mm_struct *mm = vma->vm_mm;
1333
	int retval;
1334
	pte_t *pte;
1335 1336
	spinlock_t *ptl;

1337
	retval = -EINVAL;
1338
	if (PageAnon(page))
1339
		goto out;
1340 1341
	retval = -ENOMEM;
	flush_dcache_page(page);
1342
	pte = get_locked_pte(mm, addr, &ptl);
1343
	if (!pte)
1344
		goto out;
1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355
	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;
1356 1357
	pte_unmap_unlock(pte, ptl);
	return retval;
1358 1359 1360 1361 1362 1363
out_unlock:
	pte_unmap_unlock(pte, ptl);
out:
	return retval;
}

1364 1365 1366 1367 1368 1369
/**
 * 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
 *
1370 1371 1372 1373 1374 1375
 * 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
1376
 * (see split_page()).
1377 1378 1379 1380 1381 1382 1383 1384 1385
 *
 * 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
1386 1387
int vm_insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page)
1388 1389 1390 1391 1392
{
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
	if (!page_count(page))
		return -EINVAL;
1393
	vma->vm_flags |= VM_INSERTPAGE;
N
Nick Piggin 已提交
1394
	return insert_page(vma, addr, page, vma->vm_page_prot);
1395
}
1396
EXPORT_SYMBOL(vm_insert_page);
1397

N
Nick Piggin 已提交
1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425
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
1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436
/**
 * 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
1437 1438 1439 1440 1441
 *
 * 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
1442 1443
 */
int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
N
Nick Piggin 已提交
1444
			unsigned long pfn)
N
Nick Piggin 已提交
1445
{
1446
	int ret;
N
Nick Piggin 已提交
1447 1448 1449 1450 1451 1452
	/*
	 * 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
1453 1454 1455 1456 1457
	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
1458

N
Nick Piggin 已提交
1459 1460
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
1461 1462 1463 1464 1465 1466 1467 1468 1469
	if (track_pfn_vma_new(vma, vma->vm_page_prot, pfn, PAGE_SIZE))
		return -EINVAL;

	ret = insert_pfn(vma, addr, pfn, vma->vm_page_prot);

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

	return ret;
N
Nick Piggin 已提交
1470 1471
}
EXPORT_SYMBOL(vm_insert_pfn);
N
Nick Piggin 已提交
1472

N
Nick Piggin 已提交
1473 1474 1475 1476
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
1477

N
Nick Piggin 已提交
1478 1479
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
N
Nick Piggin 已提交
1480

N
Nick Piggin 已提交
1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493
	/*
	 * 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
	 * than insert_pfn).
	 */
	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
1494
}
N
Nick Piggin 已提交
1495
EXPORT_SYMBOL(vm_insert_mixed);
N
Nick Piggin 已提交
1496

L
Linus Torvalds 已提交
1497 1498 1499 1500 1501 1502 1503 1504 1505 1506
/*
 * 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
1507
	spinlock_t *ptl;
L
Linus Torvalds 已提交
1508

H
Hugh Dickins 已提交
1509
	pte = pte_alloc_map_lock(mm, pmd, addr, &ptl);
L
Linus Torvalds 已提交
1510 1511
	if (!pte)
		return -ENOMEM;
1512
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
1513 1514
	do {
		BUG_ON(!pte_none(*pte));
N
Nick Piggin 已提交
1515
		set_pte_at(mm, addr, pte, pte_mkspecial(pfn_pte(pfn, prot)));
L
Linus Torvalds 已提交
1516 1517
		pfn++;
	} while (pte++, addr += PAGE_SIZE, addr != end);
1518
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
1519
	pte_unmap_unlock(pte - 1, ptl);
L
Linus Torvalds 已提交
1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562
	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;
}

1563 1564 1565 1566 1567 1568 1569 1570 1571 1572
/**
 * 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
1573 1574 1575 1576 1577
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;
1578
	unsigned long end = addr + PAGE_ALIGN(size);
L
Linus Torvalds 已提交
1579 1580 1581 1582 1583 1584 1585 1586
	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
1587 1588 1589 1590 1591
	 *   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.
1592 1593 1594
	 *   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
1595 1596 1597 1598
	 *
	 * 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
1599
	 */
1600
	if (addr == vma->vm_start && end == vma->vm_end)
L
Linus Torvalds 已提交
1601
		vma->vm_pgoff = pfn;
1602 1603
	else if (is_cow_mapping(vma->vm_flags))
		return -EINVAL;
L
Linus Torvalds 已提交
1604

1605
	vma->vm_flags |= VM_IO | VM_RESERVED | VM_PFNMAP;
L
Linus Torvalds 已提交
1606

1607 1608 1609 1610
	err = track_pfn_vma_new(vma, prot, pfn, PAGE_ALIGN(size));
	if (err)
		return -EINVAL;

L
Linus Torvalds 已提交
1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621
	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);
1622 1623 1624 1625

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

L
Linus Torvalds 已提交
1626 1627 1628 1629
	return err;
}
EXPORT_SYMBOL(remap_pfn_range);

1630 1631 1632 1633 1634 1635
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;
1636
	pgtable_t token;
1637
	spinlock_t *uninitialized_var(ptl);
1638 1639 1640 1641 1642 1643 1644 1645 1646

	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));

1647
	token = pmd_pgtable(*pmd);
1648 1649

	do {
1650
		err = fn(pte, token, addr, data);
1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667
		if (err)
			break;
	} while (pte++, addr += PAGE_SIZE, addr != end);

	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
1668 1669
	BUG_ON(pud_huge(*pud));

1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710
	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
1711
	unsigned long start = addr, end = addr + size;
1712 1713 1714
	int err;

	BUG_ON(addr >= end);
A
Andrea Arcangeli 已提交
1715
	mmu_notifier_invalidate_range_start(mm, start, end);
1716 1717 1718 1719 1720 1721 1722
	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
1723
	mmu_notifier_invalidate_range_end(mm, start, end);
1724 1725 1726 1727
	return err;
}
EXPORT_SYMBOL_GPL(apply_to_page_range);

1728 1729 1730 1731 1732 1733 1734 1735 1736
/*
 * 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
1737
static inline int pte_unmap_same(struct mm_struct *mm, pmd_t *pmd,
1738 1739 1740 1741 1742
				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
1743 1744
		spinlock_t *ptl = pte_lockptr(mm, pmd);
		spin_lock(ptl);
1745
		same = pte_same(*page_table, orig_pte);
H
Hugh Dickins 已提交
1746
		spin_unlock(ptl);
1747 1748 1749 1750 1751 1752
	}
#endif
	pte_unmap(page_table);
	return same;
}

L
Linus Torvalds 已提交
1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765
/*
 * 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;
}

1766
static inline void cow_user_page(struct page *dst, struct page *src, unsigned long va, struct vm_area_struct *vma)
1767 1768 1769 1770 1771 1772 1773 1774 1775
{
	/*
	 * 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
1776 1777 1778 1779 1780 1781 1782 1783 1784
		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))
1785 1786
			memset(kaddr, 0, PAGE_SIZE);
		kunmap_atomic(kaddr, KM_USER0);
1787
		flush_dcache_page(dst);
N
Nick Piggin 已提交
1788 1789
	} else
		copy_user_highpage(dst, src, va, vma);
1790 1791
}

L
Linus Torvalds 已提交
1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805
/*
 * 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.
 *
1806 1807 1808
 * 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
1809
 */
1810 1811
static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
1812
		spinlock_t *ptl, pte_t orig_pte)
L
Linus Torvalds 已提交
1813
{
1814
	struct page *old_page, *new_page;
L
Linus Torvalds 已提交
1815
	pte_t entry;
N
Nick Piggin 已提交
1816
	int reuse = 0, ret = 0;
1817
	int page_mkwrite = 0;
1818
	struct page *dirty_page = NULL;
L
Linus Torvalds 已提交
1819

1820
	old_page = vm_normal_page(vma, address, orig_pte);
1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831
	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;
1832
		goto gotten;
1833
	}
L
Linus Torvalds 已提交
1834

1835
	/*
P
Peter Zijlstra 已提交
1836 1837
	 * Take out anonymous pages first, anonymous shared vmas are
	 * not dirty accountable.
1838
	 */
P
Peter Zijlstra 已提交
1839
	if (PageAnon(old_page)) {
N
Nick Piggin 已提交
1840
		if (trylock_page(old_page)) {
P
Peter Zijlstra 已提交
1841 1842 1843 1844
			reuse = can_share_swap_page(old_page);
			unlock_page(old_page);
		}
	} else if (unlikely((vma->vm_flags & (VM_WRITE|VM_SHARED)) ==
1845
					(VM_WRITE|VM_SHARED))) {
P
Peter Zijlstra 已提交
1846 1847 1848 1849 1850
		/*
		 * Only catch write-faults on shared writable pages,
		 * read-only shared pages can get COWed by
		 * get_user_pages(.write=1, .force=1).
		 */
1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873
		if (vma->vm_ops && vma->vm_ops->page_mkwrite) {
			/*
			 * 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);

			if (vma->vm_ops->page_mkwrite(vma, old_page) < 0)
				goto unwritable_page;

			/*
			 * 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);
1874
			page_cache_release(old_page);
1875 1876
			if (!pte_same(*page_table, orig_pte))
				goto unlock;
1877 1878

			page_mkwrite = 1;
L
Linus Torvalds 已提交
1879
		}
1880 1881
		dirty_page = old_page;
		get_page(dirty_page);
1882 1883 1884 1885
		reuse = 1;
	}

	if (reuse) {
1886
reuse:
1887 1888 1889
		flush_cache_page(vma, address, pte_pfn(orig_pte));
		entry = pte_mkyoung(orig_pte);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
1890
		if (ptep_set_access_flags(vma, address, page_table, entry,1))
1891
			update_mmu_cache(vma, address, entry);
1892 1893
		ret |= VM_FAULT_WRITE;
		goto unlock;
L
Linus Torvalds 已提交
1894 1895 1896 1897 1898
	}

	/*
	 * Ok, we need to copy. Oh, well..
	 */
N
Nick Piggin 已提交
1899
	page_cache_get(old_page);
H
Hugh Dickins 已提交
1900
gotten:
1901
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
1902 1903

	if (unlikely(anon_vma_prepare(vma)))
1904
		goto oom;
N
Nick Piggin 已提交
1905 1906 1907 1908
	VM_BUG_ON(old_page == ZERO_PAGE(0));
	new_page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address);
	if (!new_page)
		goto oom;
N
Nick Piggin 已提交
1909 1910 1911 1912 1913 1914 1915 1916 1917
	/*
	 * Don't let another task, with possibly unlocked vma,
	 * keep the mlocked page.
	 */
	if (vma->vm_flags & VM_LOCKED) {
		lock_page(old_page);	/* for LRU manipulation */
		clear_page_mlock(old_page);
		unlock_page(old_page);
	}
N
Nick Piggin 已提交
1918
	cow_user_page(new_page, old_page, address, vma);
N
Nick Piggin 已提交
1919
	__SetPageUptodate(new_page);
1920

1921
	if (mem_cgroup_charge(new_page, mm, GFP_KERNEL))
1922 1923
		goto oom_free_new;

L
Linus Torvalds 已提交
1924 1925 1926
	/*
	 * Re-check the pte - we dropped the lock
	 */
1927
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
1928
	if (likely(pte_same(*page_table, orig_pte))) {
H
Hugh Dickins 已提交
1929 1930 1931 1932 1933 1934
		if (old_page) {
			if (!PageAnon(old_page)) {
				dec_mm_counter(mm, file_rss);
				inc_mm_counter(mm, anon_rss);
			}
		} else
1935
			inc_mm_counter(mm, anon_rss);
1936
		flush_cache_page(vma, address, pte_pfn(orig_pte));
1937 1938
		entry = mk_pte(new_page, vma->vm_page_prot);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
1939 1940 1941 1942 1943 1944
		/*
		 * 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.
		 */
A
Andrea Arcangeli 已提交
1945
		ptep_clear_flush_notify(vma, address, page_table);
R
Rik van Riel 已提交
1946
		SetPageSwapBacked(new_page);
1947
		lru_cache_add_active_or_unevictable(new_page, vma);
N
Nick Piggin 已提交
1948
		page_add_new_anon_rmap(new_page, vma, address);
L
Linus Torvalds 已提交
1949

1950 1951 1952
//TODO:  is this safe?  do_anonymous_page() does it this way.
		set_pte_at(mm, address, page_table, entry);
		update_mmu_cache(vma, address, entry);
N
Nick Piggin 已提交
1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978
		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.
			 */
			page_remove_rmap(old_page, vma);
		}

L
Linus Torvalds 已提交
1979 1980
		/* Free the old page.. */
		new_page = old_page;
N
Nick Piggin 已提交
1981
		ret |= VM_FAULT_WRITE;
1982 1983 1984
	} else
		mem_cgroup_uncharge_page(new_page);

H
Hugh Dickins 已提交
1985 1986 1987 1988
	if (new_page)
		page_cache_release(new_page);
	if (old_page)
		page_cache_release(old_page);
1989
unlock:
1990
	pte_unmap_unlock(page_table, ptl);
1991
	if (dirty_page) {
1992 1993 1994
		if (vma->vm_file)
			file_update_time(vma->vm_file);

1995 1996 1997 1998 1999 2000 2001 2002 2003
		/*
		 * 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.
		 */
		wait_on_page_locked(dirty_page);
2004
		set_page_dirty_balance(dirty_page, page_mkwrite);
2005 2006
		put_page(dirty_page);
	}
N
Nick Piggin 已提交
2007
	return ret;
2008
oom_free_new:
2009
	page_cache_release(new_page);
2010
oom:
H
Hugh Dickins 已提交
2011 2012
	if (old_page)
		page_cache_release(old_page);
L
Linus Torvalds 已提交
2013
	return VM_FAULT_OOM;
2014 2015 2016 2017

unwritable_page:
	page_cache_release(old_page);
	return VM_FAULT_SIGBUS;
L
Linus Torvalds 已提交
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044
}

/*
 * 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
2045
 * large vma, note the restart_addr from unmap_vmas when it breaks out:
L
Linus Torvalds 已提交
2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071
 * 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;

2072 2073
	/*
	 * files that support invalidating or truncating portions of the
N
Nick Piggin 已提交
2074
	 * file from under mmaped areas must have their ->fault function
N
Nick Piggin 已提交
2075 2076
	 * return a locked page (and set VM_FAULT_LOCKED in the return).
	 * This provides synchronisation against concurrent unmapping here.
2077 2078
	 */

L
Linus Torvalds 已提交
2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089
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;
		}
	}

2090 2091
	restart_addr = zap_page_range(vma, start_addr,
					end_addr - start_addr, details);
N
Nick Piggin 已提交
2092
	need_break = need_resched() || spin_needbreak(details->i_mmap_lock);
L
Linus Torvalds 已提交
2093

2094
	if (restart_addr >= end_addr) {
L
Linus Torvalds 已提交
2095 2096 2097 2098 2099 2100
		/* 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 */
2101
		vma->vm_truncate_count = restart_addr;
L
Linus Torvalds 已提交
2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167
		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;
	}
}

/**
2168
 * 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
2169
 * @mapping: the address space containing mmaps to be unmapped.
L
Linus Torvalds 已提交
2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205
 * @holebegin: byte in first page to unmap, relative to the start of
 * the underlying file.  This will be rounded down to a PAGE_SIZE
 * boundary.  Note that this is different from vmtruncate(), which
 * 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);

2206
	/* Protect against endless unmapping loops */
L
Linus Torvalds 已提交
2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222
	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);

2223 2224 2225 2226
/**
 * vmtruncate - unmap mappings "freed" by truncate() syscall
 * @inode: inode of the file used
 * @offset: file offset to start truncating
L
Linus Torvalds 已提交
2227 2228 2229 2230 2231 2232 2233
 *
 * NOTE! We have to be ready to update the memory sharing
 * between the file and the memory map for a potential last
 * incomplete page.  Ugly, but necessary.
 */
int vmtruncate(struct inode * inode, loff_t offset)
{
C
Christoph Hellwig 已提交
2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244
	if (inode->i_size < offset) {
		unsigned long limit;

		limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur;
		if (limit != RLIM_INFINITY && offset > limit)
			goto out_sig;
		if (offset > inode->i_sb->s_maxbytes)
			goto out_big;
		i_size_write(inode, offset);
	} else {
		struct address_space *mapping = inode->i_mapping;
L
Linus Torvalds 已提交
2245

C
Christoph Hellwig 已提交
2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267
		/*
		 * truncation of in-use swapfiles is disallowed - it would
		 * cause subsequent swapout to scribble on the now-freed
		 * blocks.
		 */
		if (IS_SWAPFILE(inode))
			return -ETXTBSY;
		i_size_write(inode, offset);

		/*
		 * unmap_mapping_range is called twice, first simply for
		 * efficiency so that truncate_inode_pages does fewer
		 * single-page unmaps.  However after this first call, and
		 * before truncate_inode_pages finishes, it is possible for
		 * private pages to be COWed, which remain after
		 * truncate_inode_pages finishes, hence the second
		 * unmap_mapping_range call must be made for correctness.
		 */
		unmap_mapping_range(mapping, offset + PAGE_SIZE - 1, 0, 1);
		truncate_inode_pages(mapping, offset);
		unmap_mapping_range(mapping, offset + PAGE_SIZE - 1, 0, 1);
	}
2268

L
Linus Torvalds 已提交
2269 2270 2271
	if (inode->i_op && inode->i_op->truncate)
		inode->i_op->truncate(inode);
	return 0;
C
Christoph Hellwig 已提交
2272

L
Linus Torvalds 已提交
2273 2274 2275 2276 2277 2278 2279
out_sig:
	send_sig(SIGXFSZ, current, 0);
out_big:
	return -EFBIG;
}
EXPORT_SYMBOL(vmtruncate);

2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291
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.
	 */
	if (!inode->i_op || !inode->i_op->truncate_range)
		return -ENOSYS;

2292
	mutex_lock(&inode->i_mutex);
2293 2294 2295
	down_write(&inode->i_alloc_sem);
	unmap_mapping_range(mapping, offset, (end - offset), 1);
	truncate_inode_pages_range(mapping, offset, end);
2296
	unmap_mapping_range(mapping, offset, (end - offset), 1);
2297 2298
	inode->i_op->truncate_range(inode, offset, end);
	up_write(&inode->i_alloc_sem);
2299
	mutex_unlock(&inode->i_mutex);
2300 2301 2302 2303

	return 0;
}

L
Linus Torvalds 已提交
2304
/*
2305 2306 2307
 * 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
2308
 */
2309 2310 2311
static int do_swap_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
		int write_access, pte_t orig_pte)
L
Linus Torvalds 已提交
2312
{
2313
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2314
	struct page *page;
2315
	swp_entry_t entry;
L
Linus Torvalds 已提交
2316
	pte_t pte;
N
Nick Piggin 已提交
2317
	int ret = 0;
L
Linus Torvalds 已提交
2318

H
Hugh Dickins 已提交
2319
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
2320
		goto out;
2321 2322

	entry = pte_to_swp_entry(orig_pte);
2323 2324 2325 2326
	if (is_migration_entry(entry)) {
		migration_entry_wait(mm, pmd, address);
		goto out;
	}
2327
	delayacct_set_flag(DELAYACCT_PF_SWAPIN);
L
Linus Torvalds 已提交
2328 2329
	page = lookup_swap_cache(entry);
	if (!page) {
2330
		grab_swap_token(); /* Contend for token _before_ read-in */
2331 2332
		page = swapin_readahead(entry,
					GFP_HIGHUSER_MOVABLE, vma, address);
L
Linus Torvalds 已提交
2333 2334
		if (!page) {
			/*
2335 2336
			 * Back out if somebody else faulted in this pte
			 * while we released the pte lock.
L
Linus Torvalds 已提交
2337
			 */
2338
			page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
2339 2340
			if (likely(pte_same(*page_table, orig_pte)))
				ret = VM_FAULT_OOM;
2341
			delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2342
			goto unlock;
L
Linus Torvalds 已提交
2343 2344 2345 2346
		}

		/* Had to read the page from swap area: Major fault */
		ret = VM_FAULT_MAJOR;
2347
		count_vm_event(PGMAJFAULT);
L
Linus Torvalds 已提交
2348 2349
	}

2350 2351 2352 2353 2354
	mark_page_accessed(page);

	lock_page(page);
	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);

2355
	if (mem_cgroup_charge(page, mm, GFP_KERNEL)) {
2356
		ret = VM_FAULT_OOM;
2357
		unlock_page(page);
2358 2359 2360
		goto out;
	}

L
Linus Torvalds 已提交
2361
	/*
2362
	 * Back out if somebody else already faulted in this pte.
L
Linus Torvalds 已提交
2363
	 */
2364
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
H
Hugh Dickins 已提交
2365
	if (unlikely(!pte_same(*page_table, orig_pte)))
2366 2367 2368 2369 2370
		goto out_nomap;

	if (unlikely(!PageUptodate(page))) {
		ret = VM_FAULT_SIGBUS;
		goto out_nomap;
L
Linus Torvalds 已提交
2371 2372 2373 2374
	}

	/* The page isn't present yet, go ahead with the fault. */

2375
	inc_mm_counter(mm, anon_rss);
L
Linus Torvalds 已提交
2376 2377 2378 2379 2380 2381 2382 2383 2384 2385
	pte = mk_pte(page, vma->vm_page_prot);
	if (write_access && can_share_swap_page(page)) {
		pte = maybe_mkwrite(pte_mkdirty(pte), vma);
		write_access = 0;
	}

	flush_icache_page(vma, page);
	set_pte_at(mm, address, page_table, pte);
	page_add_anon_rmap(page, vma, address);

2386
	swap_free(entry);
N
Nick Piggin 已提交
2387
	if (vm_swap_full() || (vma->vm_flags & VM_LOCKED) || PageMlocked(page))
2388 2389 2390
		remove_exclusive_swap_page(page);
	unlock_page(page);

L
Linus Torvalds 已提交
2391
	if (write_access) {
2392 2393 2394
		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
2395 2396 2397 2398 2399
		goto out;
	}

	/* No need to invalidate - it was non-present before */
	update_mmu_cache(vma, address, pte);
2400
unlock:
2401
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
2402 2403
out:
	return ret;
2404
out_nomap:
2405
	mem_cgroup_uncharge_page(page);
2406
	pte_unmap_unlock(page_table, ptl);
2407 2408
	unlock_page(page);
	page_cache_release(page);
2409
	return ret;
L
Linus Torvalds 已提交
2410 2411 2412
}

/*
2413 2414 2415
 * 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
2416
 */
2417 2418 2419
static int do_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
		int write_access)
L
Linus Torvalds 已提交
2420
{
2421 2422
	struct page *page;
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2423 2424
	pte_t entry;

N
Nick Piggin 已提交
2425 2426
	/* Allocate our own private page. */
	pte_unmap(page_table);
2427

N
Nick Piggin 已提交
2428 2429 2430 2431 2432
	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
2433
	__SetPageUptodate(page);
2434

2435
	if (mem_cgroup_charge(page, mm, GFP_KERNEL))
2436 2437
		goto oom_free_page;

N
Nick Piggin 已提交
2438 2439
	entry = mk_pte(page, vma->vm_page_prot);
	entry = maybe_mkwrite(pte_mkdirty(entry), vma);
L
Linus Torvalds 已提交
2440

N
Nick Piggin 已提交
2441 2442 2443 2444
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
	if (!pte_none(*page_table))
		goto release;
	inc_mm_counter(mm, anon_rss);
R
Rik van Riel 已提交
2445
	SetPageSwapBacked(page);
2446
	lru_cache_add_active_or_unevictable(page, vma);
N
Nick Piggin 已提交
2447
	page_add_new_anon_rmap(page, vma, address);
2448
	set_pte_at(mm, address, page_table, entry);
L
Linus Torvalds 已提交
2449 2450

	/* No need to invalidate - it was non-present before */
2451 2452
	update_mmu_cache(vma, address, entry);
unlock:
2453
	pte_unmap_unlock(page_table, ptl);
N
Nick Piggin 已提交
2454
	return 0;
2455
release:
2456
	mem_cgroup_uncharge_page(page);
2457 2458
	page_cache_release(page);
	goto unlock;
2459
oom_free_page:
2460
	page_cache_release(page);
2461
oom:
L
Linus Torvalds 已提交
2462 2463 2464 2465
	return VM_FAULT_OOM;
}

/*
2466
 * __do_fault() tries to create a new page mapping. It aggressively
L
Linus Torvalds 已提交
2467
 * tries to share with existing pages, but makes a separate copy if
2468 2469
 * the FAULT_FLAG_WRITE is set in the flags parameter in order to avoid
 * the next page fault.
L
Linus Torvalds 已提交
2470 2471 2472 2473
 *
 * As this is called only for pages that do not currently exist, we
 * do not need to flush old virtual caches or the TLB.
 *
2474
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
2475
 * but allow concurrent faults), and pte neither mapped nor locked.
2476
 * We return with mmap_sem still held, but pte unmapped and unlocked.
L
Linus Torvalds 已提交
2477
 */
2478
static int __do_fault(struct mm_struct *mm, struct vm_area_struct *vma,
2479
		unsigned long address, pmd_t *pmd,
2480
		pgoff_t pgoff, unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
2481
{
2482
	pte_t *page_table;
2483
	spinlock_t *ptl;
N
Nick Piggin 已提交
2484
	struct page *page;
L
Linus Torvalds 已提交
2485 2486
	pte_t entry;
	int anon = 0;
2487
	int charged = 0;
2488
	struct page *dirty_page = NULL;
N
Nick Piggin 已提交
2489 2490
	struct vm_fault vmf;
	int ret;
2491
	int page_mkwrite = 0;
2492

N
Nick Piggin 已提交
2493 2494 2495 2496
	vmf.virtual_address = (void __user *)(address & PAGE_MASK);
	vmf.pgoff = pgoff;
	vmf.flags = flags;
	vmf.page = NULL;
L
Linus Torvalds 已提交
2497

N
Nick Piggin 已提交
2498 2499 2500
	ret = vma->vm_ops->fault(vma, &vmf);
	if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE)))
		return ret;
L
Linus Torvalds 已提交
2501

2502
	/*
N
Nick Piggin 已提交
2503
	 * For consistency in subsequent calls, make the faulted page always
2504 2505
	 * locked.
	 */
N
Nick Piggin 已提交
2506
	if (unlikely(!(ret & VM_FAULT_LOCKED)))
N
Nick Piggin 已提交
2507
		lock_page(vmf.page);
2508
	else
N
Nick Piggin 已提交
2509
		VM_BUG_ON(!PageLocked(vmf.page));
2510

L
Linus Torvalds 已提交
2511 2512 2513
	/*
	 * Should we do an early C-O-W break?
	 */
N
Nick Piggin 已提交
2514
	page = vmf.page;
2515
	if (flags & FAULT_FLAG_WRITE) {
2516
		if (!(vma->vm_flags & VM_SHARED)) {
2517
			anon = 1;
2518
			if (unlikely(anon_vma_prepare(vma))) {
N
Nick Piggin 已提交
2519
				ret = VM_FAULT_OOM;
2520
				goto out;
2521
			}
N
Nick Piggin 已提交
2522 2523
			page = alloc_page_vma(GFP_HIGHUSER_MOVABLE,
						vma, address);
2524
			if (!page) {
N
Nick Piggin 已提交
2525
				ret = VM_FAULT_OOM;
2526
				goto out;
2527
			}
2528 2529 2530 2531 2532 2533
			if (mem_cgroup_charge(page, mm, GFP_KERNEL)) {
				ret = VM_FAULT_OOM;
				page_cache_release(page);
				goto out;
			}
			charged = 1;
N
Nick Piggin 已提交
2534 2535 2536 2537 2538 2539
			/*
			 * 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
2540
			copy_user_highpage(page, vmf.page, address, vma);
N
Nick Piggin 已提交
2541
			__SetPageUptodate(page);
2542
		} else {
2543 2544
			/*
			 * If the page will be shareable, see if the backing
2545
			 * address space wants to know that the page is about
2546 2547
			 * to become writable
			 */
2548 2549 2550
			if (vma->vm_ops->page_mkwrite) {
				unlock_page(page);
				if (vma->vm_ops->page_mkwrite(vma, page) < 0) {
N
Nick Piggin 已提交
2551 2552
					ret = VM_FAULT_SIGBUS;
					anon = 1; /* no anon but release vmf.page */
2553 2554 2555
					goto out_unlocked;
				}
				lock_page(page);
N
Nick Piggin 已提交
2556 2557 2558 2559 2560 2561 2562 2563
				/*
				 * XXX: this is not quite right (racy vs
				 * invalidate) to unlock and relock the page
				 * like this, however a better fix requires
				 * reworking page_mkwrite locking API, which
				 * is better done later.
				 */
				if (!page->mapping) {
N
Nick Piggin 已提交
2564
					ret = 0;
N
Nick Piggin 已提交
2565 2566 2567
					anon = 1; /* no anon but release vmf.page */
					goto out;
				}
2568
				page_mkwrite = 1;
2569 2570
			}
		}
2571

L
Linus Torvalds 已提交
2572 2573
	}

2574
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586

	/*
	 * This silly early PAGE_DIRTY setting removes a race
	 * due to the bad i386 page protection. But it's valid
	 * for other architectures too.
	 *
	 * Note that if write_access is true, we either now have
	 * 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... */
2587
	if (likely(pte_same(*page_table, orig_pte))) {
2588 2589
		flush_icache_page(vma, page);
		entry = mk_pte(page, vma->vm_page_prot);
2590
		if (flags & FAULT_FLAG_WRITE)
L
Linus Torvalds 已提交
2591 2592
			entry = maybe_mkwrite(pte_mkdirty(entry), vma);
		if (anon) {
2593
			inc_mm_counter(mm, anon_rss);
R
Rik van Riel 已提交
2594
			SetPageSwapBacked(page);
2595 2596
			lru_cache_add_active_or_unevictable(page, vma);
			page_add_new_anon_rmap(page, vma, address);
2597
		} else {
2598
			inc_mm_counter(mm, file_rss);
2599
			page_add_file_rmap(page);
2600
			if (flags & FAULT_FLAG_WRITE) {
2601
				dirty_page = page;
2602 2603
				get_page(dirty_page);
			}
2604
		}
2605 2606
//TODO:  is this safe?  do_anonymous_page() does it this way.
		set_pte_at(mm, address, page_table, entry);
2607 2608 2609

		/* no need to invalidate: a not-present page won't be cached */
		update_mmu_cache(vma, address, entry);
L
Linus Torvalds 已提交
2610
	} else {
2611 2612
		if (charged)
			mem_cgroup_uncharge_page(page);
2613 2614 2615
		if (anon)
			page_cache_release(page);
		else
2616
			anon = 1; /* no anon but release faulted_page */
L
Linus Torvalds 已提交
2617 2618
	}

2619
	pte_unmap_unlock(page_table, ptl);
2620 2621

out:
N
Nick Piggin 已提交
2622
	unlock_page(vmf.page);
2623
out_unlocked:
2624
	if (anon)
N
Nick Piggin 已提交
2625
		page_cache_release(vmf.page);
2626
	else if (dirty_page) {
2627 2628 2629
		if (vma->vm_file)
			file_update_time(vma->vm_file);

2630
		set_page_dirty_balance(dirty_page, page_mkwrite);
2631 2632
		put_page(dirty_page);
	}
2633

N
Nick Piggin 已提交
2634
	return ret;
2635
}
2636

2637 2638 2639 2640 2641
static int do_linear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
		int write_access, pte_t orig_pte)
{
	pgoff_t pgoff = (((address & PAGE_MASK)
2642
			- vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
2643 2644
	unsigned int flags = (write_access ? FAULT_FLAG_WRITE : 0);

2645 2646
	pte_unmap(page_table);
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
2647 2648
}

L
Linus Torvalds 已提交
2649 2650 2651 2652
/*
 * Fault of a previously existing named mapping. Repopulate the pte
 * from the encoded file_pte if possible. This enables swappable
 * nonlinear vmas.
2653 2654 2655 2656
 *
 * 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
2657
 */
N
Nick Piggin 已提交
2658
static int do_nonlinear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
2659 2660
		unsigned long address, pte_t *page_table, pmd_t *pmd,
		int write_access, pte_t orig_pte)
L
Linus Torvalds 已提交
2661
{
N
Nick Piggin 已提交
2662 2663
	unsigned int flags = FAULT_FLAG_NONLINEAR |
				(write_access ? FAULT_FLAG_WRITE : 0);
2664
	pgoff_t pgoff;
L
Linus Torvalds 已提交
2665

H
Hugh Dickins 已提交
2666
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
N
Nick Piggin 已提交
2667
		return 0;
L
Linus Torvalds 已提交
2668

N
Nick Piggin 已提交
2669 2670
	if (unlikely(!(vma->vm_flags & VM_NONLINEAR) ||
			!(vma->vm_flags & VM_CAN_NONLINEAR))) {
2671 2672 2673
		/*
		 * Page table corrupted: show pte and kill process.
		 */
N
Nick Piggin 已提交
2674
		print_bad_pte(vma, orig_pte, address);
2675 2676 2677 2678
		return VM_FAULT_OOM;
	}

	pgoff = pte_to_pgoff(orig_pte);
2679
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
L
Linus Torvalds 已提交
2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690
}

/*
 * 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
2691 2692 2693
 * 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
2694 2695
 */
static inline int handle_pte_fault(struct mm_struct *mm,
2696 2697
		struct vm_area_struct *vma, unsigned long address,
		pte_t *pte, pmd_t *pmd, int write_access)
L
Linus Torvalds 已提交
2698 2699
{
	pte_t entry;
2700
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2701

2702
	entry = *pte;
L
Linus Torvalds 已提交
2703
	if (!pte_present(entry)) {
2704
		if (pte_none(entry)) {
J
Jes Sorensen 已提交
2705
			if (vma->vm_ops) {
N
Nick Piggin 已提交
2706
				if (likely(vma->vm_ops->fault))
2707 2708
					return do_linear_fault(mm, vma, address,
						pte, pmd, write_access, entry);
J
Jes Sorensen 已提交
2709 2710 2711
			}
			return do_anonymous_page(mm, vma, address,
						 pte, pmd, write_access);
2712
		}
L
Linus Torvalds 已提交
2713
		if (pte_file(entry))
N
Nick Piggin 已提交
2714
			return do_nonlinear_fault(mm, vma, address,
2715 2716 2717
					pte, pmd, write_access, entry);
		return do_swap_page(mm, vma, address,
					pte, pmd, write_access, entry);
L
Linus Torvalds 已提交
2718 2719
	}

H
Hugh Dickins 已提交
2720
	ptl = pte_lockptr(mm, pmd);
2721 2722 2723
	spin_lock(ptl);
	if (unlikely(!pte_same(*pte, entry)))
		goto unlock;
L
Linus Torvalds 已提交
2724 2725
	if (write_access) {
		if (!pte_write(entry))
2726 2727
			return do_wp_page(mm, vma, address,
					pte, pmd, ptl, entry);
L
Linus Torvalds 已提交
2728 2729 2730
		entry = pte_mkdirty(entry);
	}
	entry = pte_mkyoung(entry);
2731
	if (ptep_set_access_flags(vma, address, pte, entry, write_access)) {
2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742
		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.
		 */
		if (write_access)
			flush_tlb_page(vma, address);
	}
2743 2744
unlock:
	pte_unmap_unlock(pte, ptl);
N
Nick Piggin 已提交
2745
	return 0;
L
Linus Torvalds 已提交
2746 2747 2748 2749 2750
}

/*
 * By the time we get here, we already hold the mm semaphore
 */
N
Nick Piggin 已提交
2751
int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
L
Linus Torvalds 已提交
2752 2753 2754 2755 2756 2757 2758 2759 2760
		unsigned long address, int write_access)
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

	__set_current_state(TASK_RUNNING);

2761
	count_vm_event(PGFAULT);
L
Linus Torvalds 已提交
2762

2763 2764
	if (unlikely(is_vm_hugetlb_page(vma)))
		return hugetlb_fault(mm, vma, address, write_access);
L
Linus Torvalds 已提交
2765 2766 2767 2768

	pgd = pgd_offset(mm, address);
	pud = pud_alloc(mm, pgd, address);
	if (!pud)
H
Hugh Dickins 已提交
2769
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
2770 2771
	pmd = pmd_alloc(mm, pud, address);
	if (!pmd)
H
Hugh Dickins 已提交
2772
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
2773 2774
	pte = pte_alloc_map(mm, pmd, address);
	if (!pte)
H
Hugh Dickins 已提交
2775
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
2776

H
Hugh Dickins 已提交
2777
	return handle_pte_fault(mm, vma, address, pte, pmd, write_access);
L
Linus Torvalds 已提交
2778 2779 2780 2781 2782
}

#ifndef __PAGETABLE_PUD_FOLDED
/*
 * Allocate page upper directory.
H
Hugh Dickins 已提交
2783
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
2784
 */
2785
int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
L
Linus Torvalds 已提交
2786
{
H
Hugh Dickins 已提交
2787 2788
	pud_t *new = pud_alloc_one(mm, address);
	if (!new)
2789
		return -ENOMEM;
L
Linus Torvalds 已提交
2790

2791 2792
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
2793
	spin_lock(&mm->page_table_lock);
2794
	if (pgd_present(*pgd))		/* Another has populated it */
2795
		pud_free(mm, new);
2796 2797
	else
		pgd_populate(mm, pgd, new);
H
Hugh Dickins 已提交
2798
	spin_unlock(&mm->page_table_lock);
2799
	return 0;
L
Linus Torvalds 已提交
2800 2801 2802 2803 2804 2805
}
#endif /* __PAGETABLE_PUD_FOLDED */

#ifndef __PAGETABLE_PMD_FOLDED
/*
 * Allocate page middle directory.
H
Hugh Dickins 已提交
2806
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
2807
 */
2808
int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
L
Linus Torvalds 已提交
2809
{
H
Hugh Dickins 已提交
2810 2811
	pmd_t *new = pmd_alloc_one(mm, address);
	if (!new)
2812
		return -ENOMEM;
L
Linus Torvalds 已提交
2813

2814 2815
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
2816
	spin_lock(&mm->page_table_lock);
L
Linus Torvalds 已提交
2817
#ifndef __ARCH_HAS_4LEVEL_HACK
2818
	if (pud_present(*pud))		/* Another has populated it */
2819
		pmd_free(mm, new);
2820 2821
	else
		pud_populate(mm, pud, new);
L
Linus Torvalds 已提交
2822
#else
2823
	if (pgd_present(*pud))		/* Another has populated it */
2824
		pmd_free(mm, new);
2825 2826
	else
		pgd_populate(mm, pud, new);
L
Linus Torvalds 已提交
2827
#endif /* __ARCH_HAS_4LEVEL_HACK */
H
Hugh Dickins 已提交
2828
	spin_unlock(&mm->page_table_lock);
2829
	return 0;
2830
}
L
Linus Torvalds 已提交
2831 2832 2833 2834 2835 2836 2837 2838 2839
#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
2840
		return -ENOMEM;
L
Linus Torvalds 已提交
2841
	write = (vma->vm_flags & VM_WRITE) != 0;
2842 2843
	BUG_ON(addr >= end);
	BUG_ON(end > vma->vm_end);
2844
	len = DIV_ROUND_UP(end, PAGE_SIZE) - addr/PAGE_SIZE;
L
Linus Torvalds 已提交
2845 2846
	ret = get_user_pages(current, current->mm, addr,
			len, write, 0, NULL, NULL);
2847
	if (ret < 0)
L
Linus Torvalds 已提交
2848
		return ret;
2849
	return ret == len ? 0 : -EFAULT;
L
Linus Torvalds 已提交
2850 2851 2852 2853 2854
}

#if !defined(__HAVE_ARCH_GATE_AREA)

#if defined(AT_SYSINFO_EHDR)
2855
static struct vm_area_struct gate_vma;
L
Linus Torvalds 已提交
2856 2857 2858 2859 2860 2861

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
2862 2863
	gate_vma.vm_flags = VM_READ | VM_MAYREAD | VM_EXEC | VM_MAYEXEC;
	gate_vma.vm_page_prot = __P101;
2864 2865 2866 2867 2868 2869 2870
	/*
	 * 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
2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894
	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 */
2895

2896
#ifdef CONFIG_HAVE_IOREMAP_PROT
2897 2898 2899
int follow_phys(struct vm_area_struct *vma,
		unsigned long address, unsigned int flags,
		unsigned long *prot, resource_size_t *phys)
2900 2901 2902 2903 2904 2905 2906 2907
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *ptep, pte;
	spinlock_t *ptl;
	resource_size_t phys_addr = 0;
	struct mm_struct *mm = vma->vm_mm;
2908
	int ret = -EINVAL;
2909

2910 2911
	if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
		goto out;
2912 2913 2914

	pgd = pgd_offset(mm, address);
	if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
2915
		goto out;
2916 2917 2918

	pud = pud_offset(pgd, address);
	if (pud_none(*pud) || unlikely(pud_bad(*pud)))
2919
		goto out;
2920 2921 2922

	pmd = pmd_offset(pud, address);
	if (pmd_none(*pmd) || unlikely(pmd_bad(*pmd)))
2923
		goto out;
2924 2925 2926

	/* We cannot handle huge page PFN maps. Luckily they don't exist. */
	if (pmd_huge(*pmd))
2927
		goto out;
2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941

	ptep = pte_offset_map_lock(mm, pmd, address, &ptl);
	if (!ptep)
		goto out;

	pte = *ptep;
	if (!pte_present(pte))
		goto unlock;
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;
	phys_addr = pte_pfn(pte);
	phys_addr <<= PAGE_SHIFT; /* Shift here to avoid overflow on PAE */

	*prot = pgprot_val(pte_pgprot(pte));
2942 2943
	*phys = phys_addr;
	ret = 0;
2944 2945 2946 2947

unlock:
	pte_unmap_unlock(ptep, ptl);
out:
2948
	return ret;
2949 2950 2951 2952 2953 2954 2955 2956 2957 2958
}

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;
	void *maddr;
	int offset = addr & (PAGE_SIZE-1);

2959
	if (follow_phys(vma, addr, write, &prot, &phys_addr))
2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972
		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

2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988
/*
 * 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
2989
	/* ignore errors, just check how much was successfully transferred */
2990 2991 2992
	while (len) {
		int bytes, ret, offset;
		void *maddr;
2993
		struct page *page = NULL;
2994 2995 2996

		ret = get_user_pages(tsk, mm, addr, 1,
				write, 1, &page, &vma);
2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012
		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;
3013
		} else {
3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029
			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);
3030 3031 3032 3033 3034 3035 3036 3037 3038 3039
		}
		len -= bytes;
		buf += bytes;
		addr += bytes;
	}
	up_read(&mm->mmap_sem);
	mmput(mm);

	return buf - old_buf;
}
3040 3041 3042 3043 3044 3045 3046 3047 3048

/*
 * 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;

3049 3050 3051 3052 3053 3054 3055
	/*
	 * Do not print if we are in atomic
	 * contexts (in exception stacks, etc.):
	 */
	if (preempt_count())
		return;

3056 3057 3058 3059 3060 3061 3062 3063
	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;

3064
			p = d_path(&f->f_path, buf, PAGE_SIZE);
3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077
			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);
}
3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092

#ifdef CONFIG_PROVE_LOCKING
void might_fault(void)
{
	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