memory.c 86.1 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>
55 56 57
#include <linux/kallsyms.h>
#include <linux/swapops.h>
#include <linux/elf.h>
L
Linus Torvalds 已提交
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>

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, unsigned long addr,
			  pte_t pte, struct page *page)
N
Nick Piggin 已提交
380
{
381 382 383 384 385
	pgd_t *pgd = pgd_offset(vma->vm_mm, addr);
	pud_t *pud = pud_offset(pgd, addr);
	pmd_t *pmd = pmd_offset(pud, addr);
	struct address_space *mapping;
	pgoff_t index;
386 387 388 389 390 391 392 393 394 395 396 397 398 399
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

	/*
	 * Allow a burst of 60 reports, then keep quiet for that minute;
	 * or allow a steady drip of one report per second.
	 */
	if (nr_shown == 60) {
		if (time_before(jiffies, resume)) {
			nr_unshown++;
			return;
		}
		if (nr_unshown) {
400 401
			printk(KERN_ALERT
				"BUG: Bad page map: %lu messages suppressed\n",
402 403 404 405 406 407 408
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;
409 410 411 412

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

413 414
	printk(KERN_ALERT
		"BUG: Bad page map in process %s  pte:%08llx pmd:%08llx\n",
415 416 417
		current->comm,
		(long long)pte_val(pte), (long long)pmd_val(*pmd));
	if (page) {
418
		printk(KERN_ALERT
419 420 421 422
		"page:%p flags:%p count:%d mapcount:%d mapping:%p index:%lx\n",
		page, (void *)page->flags, page_count(page),
		page_mapcount(page), page->mapping, page->index);
	}
423
	printk(KERN_ALERT
424 425 426 427 428 429
		"addr:%p vm_flags:%08lx anon_vma:%p mapping:%p index:%lx\n",
		(void *)addr, vma->vm_flags, vma->anon_vma, mapping, index);
	/*
	 * Choose text because data symbols depend on CONFIG_KALLSYMS_ALL=y
	 */
	if (vma->vm_ops)
430
		print_symbol(KERN_ALERT "vma->vm_ops->fault: %s\n",
431 432
				(unsigned long)vma->vm_ops->fault);
	if (vma->vm_file && vma->vm_file->f_op)
433
		print_symbol(KERN_ALERT "vma->vm_file->f_op->mmap: %s\n",
434
				(unsigned long)vma->vm_file->f_op->mmap);
N
Nick Piggin 已提交
435
	dump_stack();
436
	add_taint(TAINT_BAD_PAGE);
N
Nick Piggin 已提交
437 438
}

439 440 441 442 443
static inline int is_cow_mapping(unsigned int flags)
{
	return (flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
}

H
Hugh Dickins 已提交
444
/*
N
Nick Piggin 已提交
445
 * vm_normal_page -- This function gets the "struct page" associated with a pte.
446
 *
N
Nick Piggin 已提交
447 448 449
 * "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
450
 *
N
Nick Piggin 已提交
451 452 453 454 455 456 457 458
 * 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.
459
 *
J
Jared Hulbert 已提交
460 461
 * 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
462 463
 * set, and the vm_pgoff will point to the first PFN mapped: thus every special
 * mapping will always honor the rule
464 465 466
 *
 *	pfn_of_page == vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT)
 *
N
Nick Piggin 已提交
467 468 469 470 471 472
 * 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
473 474
 *
 *
N
Nick Piggin 已提交
475
 * In order to support COW of arbitrary special mappings, we have VM_MIXEDMAP.
J
Jared Hulbert 已提交
476 477 478 479 480 481 482 483 484
 *
 * 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
485
 */
N
Nick Piggin 已提交
486 487 488 489 490 491 492
#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
493
{
494
	unsigned long pfn = pte_pfn(pte);
N
Nick Piggin 已提交
495 496

	if (HAVE_PTE_SPECIAL) {
497 498 499 500
		if (likely(!pte_special(pte)))
			goto check_pfn;
		if (!(vma->vm_flags & (VM_PFNMAP | VM_MIXEDMAP)))
			print_bad_pte(vma, addr, pte, NULL);
N
Nick Piggin 已提交
501 502 503 504 505
		return NULL;
	}

	/* !HAVE_PTE_SPECIAL case follows: */

J
Jared Hulbert 已提交
506 507 508 509 510 511
	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
512 513
			unsigned long off;
			off = (addr - vma->vm_start) >> PAGE_SHIFT;
J
Jared Hulbert 已提交
514 515 516 517 518
			if (pfn == vma->vm_pgoff + off)
				return NULL;
			if (!is_cow_mapping(vma->vm_flags))
				return NULL;
		}
519 520
	}

521 522 523 524 525
check_pfn:
	if (unlikely(pfn > highest_memmap_pfn)) {
		print_bad_pte(vma, addr, pte, NULL);
		return NULL;
	}
526 527

	/*
N
Nick Piggin 已提交
528 529
	 * NOTE! We still have PageReserved() pages in the page tables.
	 * eg. VDSO mappings can cause them to exist.
530
	 */
J
Jared Hulbert 已提交
531
out:
532
	return pfn_to_page(pfn);
H
Hugh Dickins 已提交
533 534
}

L
Linus Torvalds 已提交
535 536 537 538 539 540
/*
 * 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
541
static inline void
L
Linus Torvalds 已提交
542
copy_one_pte(struct mm_struct *dst_mm, struct mm_struct *src_mm,
N
Nick Piggin 已提交
543
		pte_t *dst_pte, pte_t *src_pte, struct vm_area_struct *vma,
H
Hugh Dickins 已提交
544
		unsigned long addr, int *rss)
L
Linus Torvalds 已提交
545
{
N
Nick Piggin 已提交
546
	unsigned long vm_flags = vma->vm_flags;
L
Linus Torvalds 已提交
547 548 549 550 551 552
	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)) {
553 554 555
			swp_entry_t entry = pte_to_swp_entry(pte);

			swap_duplicate(entry);
L
Linus Torvalds 已提交
556 557 558
			/* make sure dst_mm is on swapoff's mmlist. */
			if (unlikely(list_empty(&dst_mm->mmlist))) {
				spin_lock(&mmlist_lock);
559 560 561
				if (list_empty(&dst_mm->mmlist))
					list_add(&dst_mm->mmlist,
						 &src_mm->mmlist);
L
Linus Torvalds 已提交
562 563
				spin_unlock(&mmlist_lock);
			}
564 565 566 567 568 569 570 571 572 573
			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
574
		}
575
		goto out_set_pte;
L
Linus Torvalds 已提交
576 577 578 579 580 581
	}

	/*
	 * If it's a COW mapping, write protect it both
	 * in the parent and the child
	 */
582
	if (is_cow_mapping(vm_flags)) {
L
Linus Torvalds 已提交
583
		ptep_set_wrprotect(src_mm, addr, src_pte);
584
		pte = pte_wrprotect(pte);
L
Linus Torvalds 已提交
585 586 587 588 589 590 591 592 593
	}

	/*
	 * 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);
594 595 596 597

	page = vm_normal_page(vma, addr, pte);
	if (page) {
		get_page(page);
N
Nick Piggin 已提交
598
		page_dup_rmap(page, vma, addr);
599 600
		rss[!!PageAnon(page)]++;
	}
601 602 603

out_set_pte:
	set_pte_at(dst_mm, addr, dst_pte, pte);
L
Linus Torvalds 已提交
604 605 606 607 608 609 610
}

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
611
	spinlock_t *src_ptl, *dst_ptl;
612
	int progress = 0;
H
Hugh Dickins 已提交
613
	int rss[2];
L
Linus Torvalds 已提交
614 615

again:
616
	rss[1] = rss[0] = 0;
H
Hugh Dickins 已提交
617
	dst_pte = pte_alloc_map_lock(dst_mm, dst_pmd, addr, &dst_ptl);
L
Linus Torvalds 已提交
618 619 620
	if (!dst_pte)
		return -ENOMEM;
	src_pte = pte_offset_map_nested(src_pmd, addr);
H
Hugh Dickins 已提交
621
	src_ptl = pte_lockptr(src_mm, src_pmd);
I
Ingo Molnar 已提交
622
	spin_lock_nested(src_ptl, SINGLE_DEPTH_NESTING);
623
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
624 625 626 627 628 629

	do {
		/*
		 * We are holding two locks at this point - either of them
		 * could generate latencies in another task on another CPU.
		 */
630 631 632
		if (progress >= 32) {
			progress = 0;
			if (need_resched() ||
N
Nick Piggin 已提交
633
			    spin_needbreak(src_ptl) || spin_needbreak(dst_ptl))
634 635
				break;
		}
L
Linus Torvalds 已提交
636 637 638 639
		if (pte_none(*src_pte)) {
			progress++;
			continue;
		}
H
Hugh Dickins 已提交
640
		copy_one_pte(dst_mm, src_mm, dst_pte, src_pte, vma, addr, rss);
L
Linus Torvalds 已提交
641 642 643
		progress += 8;
	} while (dst_pte++, src_pte++, addr += PAGE_SIZE, addr != end);

644
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
645
	spin_unlock(src_ptl);
L
Linus Torvalds 已提交
646
	pte_unmap_nested(src_pte - 1);
647
	add_mm_rss(dst_mm, rss[0], rss[1]);
H
Hugh Dickins 已提交
648 649
	pte_unmap_unlock(dst_pte - 1, dst_ptl);
	cond_resched();
L
Linus Torvalds 已提交
650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705
	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
706
	int ret;
L
Linus Torvalds 已提交
707

708 709 710 711 712 713
	/*
	 * 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.
	 */
714
	if (!(vma->vm_flags & (VM_HUGETLB|VM_NONLINEAR|VM_PFNMAP|VM_INSERTPAGE))) {
715 716 717 718
		if (!vma->anon_vma)
			return 0;
	}

L
Linus Torvalds 已提交
719 720 721
	if (is_vm_hugetlb_page(vma))
		return copy_hugetlb_page_range(dst_mm, src_mm, vma);

722
	if (unlikely(is_pfn_mapping(vma))) {
723 724 725 726 727 728 729 730 731
		/*
		 * 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
732 733 734 735 736 737 738 739 740 741
	/*
	 * 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
742 743 744 745 746 747
	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
748 749 750 751 752
		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
753
	} while (dst_pgd++, src_pgd++, addr = next, addr != end);
A
Andrea Arcangeli 已提交
754 755 756 757 758

	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
759 760
}

761
static unsigned long zap_pte_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
762
				struct vm_area_struct *vma, pmd_t *pmd,
L
Linus Torvalds 已提交
763
				unsigned long addr, unsigned long end,
764
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
765
{
N
Nick Piggin 已提交
766
	struct mm_struct *mm = tlb->mm;
L
Linus Torvalds 已提交
767
	pte_t *pte;
768
	spinlock_t *ptl;
769 770
	int file_rss = 0;
	int anon_rss = 0;
L
Linus Torvalds 已提交
771

772
	pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
773
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
774 775
	do {
		pte_t ptent = *pte;
776 777
		if (pte_none(ptent)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
778
			continue;
779
		}
780 781 782

		(*zap_work) -= PAGE_SIZE;

L
Linus Torvalds 已提交
783
		if (pte_present(ptent)) {
H
Hugh Dickins 已提交
784
			struct page *page;
785

786
			page = vm_normal_page(vma, addr, ptent);
L
Linus Torvalds 已提交
787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804
			if (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
805
			ptent = ptep_get_and_clear_full(mm, addr, pte,
806
							tlb->fullmm);
L
Linus Torvalds 已提交
807 808 809 810 811 812
			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
813
				set_pte_at(mm, addr, pte,
L
Linus Torvalds 已提交
814 815
					   pgoff_to_pte(page->index));
			if (PageAnon(page))
H
Hugh Dickins 已提交
816
				anon_rss--;
817 818 819
			else {
				if (pte_dirty(ptent))
					set_page_dirty(page);
820 821
				if (pte_young(ptent) &&
				    likely(!VM_SequentialReadHint(vma)))
822
					mark_page_accessed(page);
H
Hugh Dickins 已提交
823
				file_rss--;
824
			}
825
			page_remove_rmap(page);
826 827
			if (unlikely(page_mapcount(page) < 0))
				print_bad_pte(vma, addr, ptent, page);
L
Linus Torvalds 已提交
828 829 830 831 832 833 834 835 836
			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;
837 838 839 840 841 842
		if (pte_file(ptent)) {
			if (unlikely(!(vma->vm_flags & VM_NONLINEAR)))
				print_bad_pte(vma, addr, ptent, NULL);
		} else if
		  (unlikely(!free_swap_and_cache(pte_to_swp_entry(ptent))))
			print_bad_pte(vma, addr, ptent, NULL);
843
		pte_clear_not_present_full(mm, addr, pte, tlb->fullmm);
844
	} while (pte++, addr += PAGE_SIZE, (addr != end && *zap_work > 0));
845

H
Hugh Dickins 已提交
846
	add_mm_rss(mm, file_rss, anon_rss);
847
	arch_leave_lazy_mmu_mode();
848
	pte_unmap_unlock(pte - 1, ptl);
849 850

	return addr;
L
Linus Torvalds 已提交
851 852
}

853
static inline unsigned long zap_pmd_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
854
				struct vm_area_struct *vma, pud_t *pud,
L
Linus Torvalds 已提交
855
				unsigned long addr, unsigned long end,
856
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
857 858 859 860 861 862 863
{
	pmd_t *pmd;
	unsigned long next;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
864 865
		if (pmd_none_or_clear_bad(pmd)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
866
			continue;
867 868 869 870 871 872
		}
		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
873 874
}

875
static inline unsigned long zap_pud_range(struct mmu_gather *tlb,
N
Nick Piggin 已提交
876
				struct vm_area_struct *vma, pgd_t *pgd,
L
Linus Torvalds 已提交
877
				unsigned long addr, unsigned long end,
878
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
879 880 881 882 883 884 885
{
	pud_t *pud;
	unsigned long next;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
886 887
		if (pud_none_or_clear_bad(pud)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
888
			continue;
889 890 891 892 893 894
		}
		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
895 896
}

897 898
static unsigned long unmap_page_range(struct mmu_gather *tlb,
				struct vm_area_struct *vma,
L
Linus Torvalds 已提交
899
				unsigned long addr, unsigned long end,
900
				long *zap_work, struct zap_details *details)
L
Linus Torvalds 已提交
901 902 903 904 905 906 907 908 909 910 911 912
{
	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);
913 914
		if (pgd_none_or_clear_bad(pgd)) {
			(*zap_work)--;
L
Linus Torvalds 已提交
915
			continue;
916 917 918 919
		}
		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
920
	tlb_end_vma(tlb, vma);
921 922

	return addr;
L
Linus Torvalds 已提交
923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940
}

#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
 *
941
 * Returns the end address of the unmapping (restart addr if interrupted).
L
Linus Torvalds 已提交
942
 *
943
 * Unmap all pages in the vma list.
L
Linus Torvalds 已提交
944
 *
945 946
 * 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
947 948 949 950 951 952 953 954 955 956 957
 * 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.
 */
958
unsigned long unmap_vmas(struct mmu_gather **tlbp,
L
Linus Torvalds 已提交
959 960 961 962
		struct vm_area_struct *vma, unsigned long start_addr,
		unsigned long end_addr, unsigned long *nr_accounted,
		struct zap_details *details)
{
963
	long zap_work = ZAP_BLOCK_SIZE;
L
Linus Torvalds 已提交
964 965
	unsigned long tlb_start = 0;	/* For tlb_finish_mmu */
	int tlb_start_valid = 0;
966
	unsigned long start = start_addr;
L
Linus Torvalds 已提交
967
	spinlock_t *i_mmap_lock = details? details->i_mmap_lock: NULL;
968
	int fullmm = (*tlbp)->fullmm;
A
Andrea Arcangeli 已提交
969
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
970

A
Andrea Arcangeli 已提交
971
	mmu_notifier_invalidate_range_start(mm, start_addr, end_addr);
L
Linus Torvalds 已提交
972 973 974 975 976 977 978 979 980 981 982 983 984
	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;

985
		if (unlikely(is_pfn_mapping(vma)))
986 987
			untrack_pfn_vma(vma, 0, 0);

L
Linus Torvalds 已提交
988 989 990 991 992 993
		while (start != end) {
			if (!tlb_start_valid) {
				tlb_start = start;
				tlb_start_valid = 1;
			}

994
			if (unlikely(is_vm_hugetlb_page(vma))) {
995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008
				/*
				 * 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) /
1009
					pages_per_huge_page(hstate_vma(vma));
1010 1011
				}

1012 1013 1014 1015 1016 1017 1018 1019
				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
1020 1021 1022 1023 1024
			}

			tlb_finish_mmu(*tlbp, tlb_start, start);

			if (need_resched() ||
N
Nick Piggin 已提交
1025
				(i_mmap_lock && spin_needbreak(i_mmap_lock))) {
L
Linus Torvalds 已提交
1026
				if (i_mmap_lock) {
1027
					*tlbp = NULL;
L
Linus Torvalds 已提交
1028 1029 1030 1031 1032
					goto out;
				}
				cond_resched();
			}

1033
			*tlbp = tlb_gather_mmu(vma->vm_mm, fullmm);
L
Linus Torvalds 已提交
1034
			tlb_start_valid = 0;
1035
			zap_work = ZAP_BLOCK_SIZE;
L
Linus Torvalds 已提交
1036 1037 1038
		}
	}
out:
A
Andrea Arcangeli 已提交
1039
	mmu_notifier_invalidate_range_end(mm, start_addr, end_addr);
1040
	return start;	/* which is now the end (or restart) address */
L
Linus Torvalds 已提交
1041 1042 1043 1044 1045 1046 1047 1048 1049
}

/**
 * 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
 */
1050
unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
L
Linus Torvalds 已提交
1051 1052 1053 1054 1055 1056 1057 1058 1059
		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);
1060
	update_hiwater_rss(mm);
1061 1062 1063
	end = unmap_vmas(&tlb, vma, address, end, &nr_accounted, details);
	if (tlb)
		tlb_finish_mmu(tlb, address, end);
1064
	return end;
L
Linus Torvalds 已提交
1065 1066
}

1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089
/**
 * 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
1090 1091 1092
/*
 * Do a quick page-table lookup for a single page.
 */
1093
struct page *follow_page(struct vm_area_struct *vma, unsigned long address,
1094
			unsigned int flags)
L
Linus Torvalds 已提交
1095 1096 1097 1098 1099
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *ptep, pte;
1100
	spinlock_t *ptl;
L
Linus Torvalds 已提交
1101
	struct page *page;
1102
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
1103

1104 1105 1106 1107 1108
	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
1109

1110
	page = NULL;
L
Linus Torvalds 已提交
1111 1112
	pgd = pgd_offset(mm, address);
	if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
1113
		goto no_page_table;
L
Linus Torvalds 已提交
1114 1115

	pud = pud_offset(pgd, address);
A
Andi Kleen 已提交
1116
	if (pud_none(*pud))
1117
		goto no_page_table;
A
Andi Kleen 已提交
1118 1119 1120 1121 1122 1123 1124 1125
	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
1126
	pmd = pmd_offset(pud, address);
1127
	if (pmd_none(*pmd))
1128 1129 1130 1131
		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
1132
		goto out;
1133
	}
1134 1135 1136
	if (unlikely(pmd_bad(*pmd)))
		goto no_page_table;

1137
	ptep = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
1138 1139

	pte = *ptep;
1140
	if (!pte_present(pte))
1141
		goto no_page;
1142 1143
	if ((flags & FOLL_WRITE) && !pte_write(pte))
		goto unlock;
1144 1145
	page = vm_normal_page(vma, address, pte);
	if (unlikely(!page))
1146
		goto bad_page;
L
Linus Torvalds 已提交
1147

1148 1149 1150 1151 1152 1153 1154 1155 1156 1157
	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
1158
out:
1159
	return page;
L
Linus Torvalds 已提交
1160

1161 1162 1163 1164 1165 1166 1167 1168 1169
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 */
1170 1171 1172 1173 1174 1175
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
1176
		page = ZERO_PAGE(0);
1177 1178 1179 1180 1181
		if (flags & FOLL_GET)
			get_page(page);
		BUG_ON(flags & FOLL_WRITE);
	}
	return page;
L
Linus Torvalds 已提交
1182 1183
}

1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196
/* 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
1197
	 * And if we have a fault routine, it's not an anonymous region.
1198
	 */
N
Nick Piggin 已提交
1199
	return !vma->vm_ops || !vma->vm_ops->fault;
1200 1201
}

N
Nick Piggin 已提交
1202 1203 1204 1205


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
1206 1207 1208
		struct page **pages, struct vm_area_struct **vmas)
{
	int i;
N
Nick Piggin 已提交
1209 1210 1211 1212
	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);
1213
	int ignore_sigkill = !!(flags & GUP_FLAGS_IGNORE_SIGKILL);
L
Linus Torvalds 已提交
1214

1215 1216
	if (len <= 0)
		return 0;
L
Linus Torvalds 已提交
1217 1218 1219 1220
	/* 
	 * Require read or write permissions.
	 * If 'force' is set, we only require the "MAY" flags.
	 */
1221 1222
	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
1223 1224 1225
	i = 0;

	do {
1226 1227
		struct vm_area_struct *vma;
		unsigned int foll_flags;
L
Linus Torvalds 已提交
1228 1229 1230 1231 1232 1233 1234 1235 1236

		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
1237 1238 1239

			/* user gate pages are read-only */
			if (!ignore && write)
L
Linus Torvalds 已提交
1240 1241 1242 1243 1244 1245 1246 1247 1248
				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);
1249 1250
			if (pmd_none(*pmd))
				return i ? : -EFAULT;
L
Linus Torvalds 已提交
1251
			pte = pte_offset_map(pmd, pg);
1252 1253 1254 1255
			if (pte_none(*pte)) {
				pte_unmap(pte);
				return i ? : -EFAULT;
			}
L
Linus Torvalds 已提交
1256
			if (pages) {
1257
				struct page *page = vm_normal_page(gate_vma, start, *pte);
1258 1259 1260
				pages[i] = page;
				if (page)
					get_page(page);
L
Linus Torvalds 已提交
1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
			}
			pte_unmap(pte);
			if (vmas)
				vmas[i] = gate_vma;
			i++;
			start += PAGE_SIZE;
			len--;
			continue;
		}

N
Nick Piggin 已提交
1271 1272 1273
		if (!vma ||
		    (vma->vm_flags & (VM_IO | VM_PFNMAP)) ||
		    (!ignore && !(vm_flags & vma->vm_flags)))
L
Linus Torvalds 已提交
1274 1275 1276 1277
			return i ? : -EFAULT;

		if (is_vm_hugetlb_page(vma)) {
			i = follow_hugetlb_page(mm, vma, pages, vmas,
1278
						&start, &len, i, write);
L
Linus Torvalds 已提交
1279 1280
			continue;
		}
1281 1282 1283 1284

		foll_flags = FOLL_TOUCH;
		if (pages)
			foll_flags |= FOLL_GET;
1285
		if (!write && use_zero_page(vma))
1286 1287
			foll_flags |= FOLL_ANON;

L
Linus Torvalds 已提交
1288
		do {
1289
			struct page *page;
L
Linus Torvalds 已提交
1290

1291
			/*
1292 1293 1294 1295 1296
			 * If we have a pending SIGKILL, don't keep faulting
			 * pages and potentially allocating memory, unless
			 * current is handling munlock--e.g., on exit. In
			 * that case, we are not allocating memory.  Rather,
			 * we're only unlocking already resident/mapped pages.
1297
			 */
1298 1299 1300
			if (unlikely(!ignore_sigkill &&
					fatal_signal_pending(current)))
				return i ? i : -ERESTARTSYS;
1301

1302 1303
			if (write)
				foll_flags |= FOLL_WRITE;
1304

1305
			cond_resched();
1306
			while (!(page = follow_page(vma, start, foll_flags))) {
1307
				int ret;
N
Nick Piggin 已提交
1308
				ret = handle_mm_fault(mm, vma, start,
1309
						foll_flags & FOLL_WRITE);
N
Nick Piggin 已提交
1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321
				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++;

1322
				/*
N
Nick Piggin 已提交
1323 1324 1325 1326
				 * 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
1327 1328 1329 1330 1331 1332
				 * page lookups as if they were reads. But only
				 * do so when looping for pte_write is futile:
				 * in some cases userspace may also be wanting
				 * to write to the gotten user page, which a
				 * read fault here might prevent (a readonly
				 * page might get reCOWed by userspace write).
1333
				 */
1334 1335
				if ((ret & VM_FAULT_WRITE) &&
				    !(vma->vm_flags & VM_WRITE))
1336
					foll_flags &= ~FOLL_WRITE;
N
Nick Piggin 已提交
1337

1338
				cond_resched();
L
Linus Torvalds 已提交
1339
			}
1340 1341
			if (IS_ERR(page))
				return i ? i : PTR_ERR(page);
L
Linus Torvalds 已提交
1342
			if (pages) {
1343
				pages[i] = page;
1344

1345
				flush_anon_page(vma, page, start);
1346
				flush_dcache_page(page);
L
Linus Torvalds 已提交
1347 1348 1349 1350 1351 1352
			}
			if (vmas)
				vmas[i] = vma;
			i++;
			start += PAGE_SIZE;
			len--;
1353 1354
		} while (len && start < vma->vm_end);
	} while (len);
L
Linus Torvalds 已提交
1355 1356
	return i;
}
N
Nick Piggin 已提交
1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373

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
1374 1375
EXPORT_SYMBOL(get_user_pages);

H
Harvey Harrison 已提交
1376 1377
pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
			spinlock_t **ptl)
1378 1379 1380 1381
{
	pgd_t * pgd = pgd_offset(mm, addr);
	pud_t * pud = pud_alloc(mm, pgd, addr);
	if (pud) {
1382
		pmd_t * pmd = pmd_alloc(mm, pud, addr);
1383 1384 1385 1386 1387 1388
		if (pmd)
			return pte_alloc_map_lock(mm, pmd, addr, ptl);
	}
	return NULL;
}

1389 1390 1391 1392 1393 1394 1395
/*
 * 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
1396 1397
static int insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page, pgprot_t prot)
1398
{
N
Nick Piggin 已提交
1399
	struct mm_struct *mm = vma->vm_mm;
1400
	int retval;
1401
	pte_t *pte;
1402 1403
	spinlock_t *ptl;

1404
	retval = -EINVAL;
1405
	if (PageAnon(page))
1406
		goto out;
1407 1408
	retval = -ENOMEM;
	flush_dcache_page(page);
1409
	pte = get_locked_pte(mm, addr, &ptl);
1410
	if (!pte)
1411
		goto out;
1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422
	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;
1423 1424
	pte_unmap_unlock(pte, ptl);
	return retval;
1425 1426 1427 1428 1429 1430
out_unlock:
	pte_unmap_unlock(pte, ptl);
out:
	return retval;
}

1431 1432 1433 1434 1435 1436
/**
 * 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
 *
1437 1438 1439 1440 1441 1442
 * 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
1443
 * (see split_page()).
1444 1445 1446 1447 1448 1449 1450 1451 1452
 *
 * 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
1453 1454
int vm_insert_page(struct vm_area_struct *vma, unsigned long addr,
			struct page *page)
1455 1456 1457 1458 1459
{
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
	if (!page_count(page))
		return -EINVAL;
1460
	vma->vm_flags |= VM_INSERTPAGE;
N
Nick Piggin 已提交
1461
	return insert_page(vma, addr, page, vma->vm_page_prot);
1462
}
1463
EXPORT_SYMBOL(vm_insert_page);
1464

N
Nick Piggin 已提交
1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492
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
1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503
/**
 * 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
1504 1505 1506 1507 1508
 *
 * 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
1509 1510
 */
int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
N
Nick Piggin 已提交
1511
			unsigned long pfn)
N
Nick Piggin 已提交
1512
{
1513
	int ret;
N
Nick Piggin 已提交
1514 1515 1516 1517 1518 1519
	/*
	 * 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
1520 1521 1522 1523 1524
	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
1525

N
Nick Piggin 已提交
1526 1527
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
1528 1529 1530 1531 1532 1533 1534 1535 1536
	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
1537 1538
}
EXPORT_SYMBOL(vm_insert_pfn);
N
Nick Piggin 已提交
1539

N
Nick Piggin 已提交
1540 1541 1542 1543
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
1544

N
Nick Piggin 已提交
1545 1546
	if (addr < vma->vm_start || addr >= vma->vm_end)
		return -EFAULT;
N
Nick Piggin 已提交
1547

N
Nick Piggin 已提交
1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560
	/*
	 * 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
1561
}
N
Nick Piggin 已提交
1562
EXPORT_SYMBOL(vm_insert_mixed);
N
Nick Piggin 已提交
1563

L
Linus Torvalds 已提交
1564 1565 1566 1567 1568 1569 1570 1571 1572 1573
/*
 * 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
1574
	spinlock_t *ptl;
L
Linus Torvalds 已提交
1575

H
Hugh Dickins 已提交
1576
	pte = pte_alloc_map_lock(mm, pmd, addr, &ptl);
L
Linus Torvalds 已提交
1577 1578
	if (!pte)
		return -ENOMEM;
1579
	arch_enter_lazy_mmu_mode();
L
Linus Torvalds 已提交
1580 1581
	do {
		BUG_ON(!pte_none(*pte));
N
Nick Piggin 已提交
1582
		set_pte_at(mm, addr, pte, pte_mkspecial(pfn_pte(pfn, prot)));
L
Linus Torvalds 已提交
1583 1584
		pfn++;
	} while (pte++, addr += PAGE_SIZE, addr != end);
1585
	arch_leave_lazy_mmu_mode();
H
Hugh Dickins 已提交
1586
	pte_unmap_unlock(pte - 1, ptl);
L
Linus Torvalds 已提交
1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629
	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;
}

1630 1631 1632 1633 1634 1635 1636 1637 1638 1639
/**
 * 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
1640 1641 1642 1643 1644
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;
1645
	unsigned long end = addr + PAGE_ALIGN(size);
L
Linus Torvalds 已提交
1646 1647 1648 1649 1650 1651 1652 1653
	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
1654 1655 1656 1657 1658
	 *   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.
1659 1660 1661
	 *   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
1662 1663 1664 1665
	 *
	 * 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
1666
	 */
1667
	if (addr == vma->vm_start && end == vma->vm_end)
L
Linus Torvalds 已提交
1668
		vma->vm_pgoff = pfn;
1669 1670
	else if (is_cow_mapping(vma->vm_flags))
		return -EINVAL;
L
Linus Torvalds 已提交
1671

1672
	vma->vm_flags |= VM_IO | VM_RESERVED | VM_PFNMAP;
L
Linus Torvalds 已提交
1673

1674
	err = track_pfn_vma_new(vma, prot, pfn, PAGE_ALIGN(size));
1675 1676 1677 1678 1679 1680
	if (err) {
		/*
		 * To indicate that track_pfn related cleanup is not
		 * needed from higher level routine calling unmap_vmas
		 */
		vma->vm_flags &= ~(VM_IO | VM_RESERVED | VM_PFNMAP);
1681
		return -EINVAL;
1682
	}
1683

L
Linus Torvalds 已提交
1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694
	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);
1695 1696 1697 1698

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

L
Linus Torvalds 已提交
1699 1700 1701 1702
	return err;
}
EXPORT_SYMBOL(remap_pfn_range);

1703 1704 1705 1706 1707 1708
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;
1709
	pgtable_t token;
1710
	spinlock_t *uninitialized_var(ptl);
1711 1712 1713 1714 1715 1716 1717 1718 1719

	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));

1720 1721
	arch_enter_lazy_mmu_mode();

1722
	token = pmd_pgtable(*pmd);
1723 1724

	do {
1725
		err = fn(pte, token, addr, data);
1726 1727 1728 1729
		if (err)
			break;
	} while (pte++, addr += PAGE_SIZE, addr != end);

1730 1731
	arch_leave_lazy_mmu_mode();

1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744
	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
1745 1746
	BUG_ON(pud_huge(*pud));

1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787
	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
1788
	unsigned long start = addr, end = addr + size;
1789 1790 1791
	int err;

	BUG_ON(addr >= end);
A
Andrea Arcangeli 已提交
1792
	mmu_notifier_invalidate_range_start(mm, start, end);
1793 1794 1795 1796 1797 1798 1799
	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
1800
	mmu_notifier_invalidate_range_end(mm, start, end);
1801 1802 1803 1804
	return err;
}
EXPORT_SYMBOL_GPL(apply_to_page_range);

1805 1806 1807 1808 1809 1810 1811 1812 1813
/*
 * 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
1814
static inline int pte_unmap_same(struct mm_struct *mm, pmd_t *pmd,
1815 1816 1817 1818 1819
				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
1820 1821
		spinlock_t *ptl = pte_lockptr(mm, pmd);
		spin_lock(ptl);
1822
		same = pte_same(*page_table, orig_pte);
H
Hugh Dickins 已提交
1823
		spin_unlock(ptl);
1824 1825 1826 1827 1828 1829
	}
#endif
	pte_unmap(page_table);
	return same;
}

L
Linus Torvalds 已提交
1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842
/*
 * 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;
}

1843
static inline void cow_user_page(struct page *dst, struct page *src, unsigned long va, struct vm_area_struct *vma)
1844 1845 1846 1847 1848 1849 1850 1851 1852
{
	/*
	 * 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
1853 1854 1855 1856 1857 1858 1859 1860 1861
		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))
1862 1863
			memset(kaddr, 0, PAGE_SIZE);
		kunmap_atomic(kaddr, KM_USER0);
1864
		flush_dcache_page(dst);
N
Nick Piggin 已提交
1865 1866
	} else
		copy_user_highpage(dst, src, va, vma);
1867 1868
}

L
Linus Torvalds 已提交
1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882
/*
 * 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.
 *
1883 1884 1885
 * 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
1886
 */
1887 1888
static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long address, pte_t *page_table, pmd_t *pmd,
1889
		spinlock_t *ptl, pte_t orig_pte)
L
Linus Torvalds 已提交
1890
{
1891
	struct page *old_page, *new_page;
L
Linus Torvalds 已提交
1892
	pte_t entry;
N
Nick Piggin 已提交
1893
	int reuse = 0, ret = 0;
1894
	int page_mkwrite = 0;
1895
	struct page *dirty_page = NULL;
L
Linus Torvalds 已提交
1896

1897
	old_page = vm_normal_page(vma, address, orig_pte);
1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908
	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;
1909
		goto gotten;
1910
	}
L
Linus Torvalds 已提交
1911

1912
	/*
P
Peter Zijlstra 已提交
1913 1914
	 * Take out anonymous pages first, anonymous shared vmas are
	 * not dirty accountable.
1915
	 */
P
Peter Zijlstra 已提交
1916
	if (PageAnon(old_page)) {
1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928
		if (!trylock_page(old_page)) {
			page_cache_get(old_page);
			pte_unmap_unlock(page_table, ptl);
			lock_page(old_page);
			page_table = pte_offset_map_lock(mm, pmd, address,
							 &ptl);
			if (!pte_same(*page_table, orig_pte)) {
				unlock_page(old_page);
				page_cache_release(old_page);
				goto unlock;
			}
			page_cache_release(old_page);
P
Peter Zijlstra 已提交
1929
		}
1930
		reuse = reuse_swap_page(old_page);
1931
		unlock_page(old_page);
P
Peter Zijlstra 已提交
1932
	} else if (unlikely((vma->vm_flags & (VM_WRITE|VM_SHARED)) ==
1933
					(VM_WRITE|VM_SHARED))) {
P
Peter Zijlstra 已提交
1934 1935 1936 1937 1938
		/*
		 * Only catch write-faults on shared writable pages,
		 * read-only shared pages can get COWed by
		 * get_user_pages(.write=1, .force=1).
		 */
1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961
		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);
1962
			page_cache_release(old_page);
1963 1964
			if (!pte_same(*page_table, orig_pte))
				goto unlock;
1965 1966

			page_mkwrite = 1;
L
Linus Torvalds 已提交
1967
		}
1968 1969
		dirty_page = old_page;
		get_page(dirty_page);
1970 1971 1972 1973
		reuse = 1;
	}

	if (reuse) {
1974
reuse:
1975 1976 1977
		flush_cache_page(vma, address, pte_pfn(orig_pte));
		entry = pte_mkyoung(orig_pte);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
1978
		if (ptep_set_access_flags(vma, address, page_table, entry,1))
1979
			update_mmu_cache(vma, address, entry);
1980 1981
		ret |= VM_FAULT_WRITE;
		goto unlock;
L
Linus Torvalds 已提交
1982 1983 1984 1985 1986
	}

	/*
	 * Ok, we need to copy. Oh, well..
	 */
N
Nick Piggin 已提交
1987
	page_cache_get(old_page);
H
Hugh Dickins 已提交
1988
gotten:
1989
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
1990 1991

	if (unlikely(anon_vma_prepare(vma)))
1992
		goto oom;
N
Nick Piggin 已提交
1993 1994 1995 1996
	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
1997 1998 1999 2000 2001 2002 2003 2004 2005
	/*
	 * 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
2006
	cow_user_page(new_page, old_page, address, vma);
N
Nick Piggin 已提交
2007
	__SetPageUptodate(new_page);
2008

K
KAMEZAWA Hiroyuki 已提交
2009
	if (mem_cgroup_newpage_charge(new_page, mm, GFP_KERNEL))
2010 2011
		goto oom_free_new;

L
Linus Torvalds 已提交
2012 2013 2014
	/*
	 * Re-check the pte - we dropped the lock
	 */
2015
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
2016
	if (likely(pte_same(*page_table, orig_pte))) {
H
Hugh Dickins 已提交
2017 2018 2019 2020 2021 2022
		if (old_page) {
			if (!PageAnon(old_page)) {
				dec_mm_counter(mm, file_rss);
				inc_mm_counter(mm, anon_rss);
			}
		} else
2023
			inc_mm_counter(mm, anon_rss);
2024
		flush_cache_page(vma, address, pte_pfn(orig_pte));
2025 2026
		entry = mk_pte(new_page, vma->vm_page_prot);
		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
2027 2028 2029 2030 2031 2032
		/*
		 * 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
2033
		ptep_clear_flush_notify(vma, address, page_table);
N
Nick Piggin 已提交
2034
		page_add_new_anon_rmap(new_page, vma, address);
2035 2036
		set_pte_at(mm, address, page_table, entry);
		update_mmu_cache(vma, address, entry);
N
Nick Piggin 已提交
2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059
		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.
			 */
2060
			page_remove_rmap(old_page);
N
Nick Piggin 已提交
2061 2062
		}

L
Linus Torvalds 已提交
2063 2064
		/* Free the old page.. */
		new_page = old_page;
N
Nick Piggin 已提交
2065
		ret |= VM_FAULT_WRITE;
2066 2067 2068
	} else
		mem_cgroup_uncharge_page(new_page);

H
Hugh Dickins 已提交
2069 2070 2071 2072
	if (new_page)
		page_cache_release(new_page);
	if (old_page)
		page_cache_release(old_page);
2073
unlock:
2074
	pte_unmap_unlock(page_table, ptl);
2075
	if (dirty_page) {
2076 2077 2078
		if (vma->vm_file)
			file_update_time(vma->vm_file);

2079 2080 2081 2082 2083 2084 2085 2086 2087
		/*
		 * 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);
2088
		set_page_dirty_balance(dirty_page, page_mkwrite);
2089 2090
		put_page(dirty_page);
	}
N
Nick Piggin 已提交
2091
	return ret;
2092
oom_free_new:
2093
	page_cache_release(new_page);
2094
oom:
H
Hugh Dickins 已提交
2095 2096
	if (old_page)
		page_cache_release(old_page);
L
Linus Torvalds 已提交
2097
	return VM_FAULT_OOM;
2098 2099 2100 2101

unwritable_page:
	page_cache_release(old_page);
	return VM_FAULT_SIGBUS;
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
}

/*
 * 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
2129
 * large vma, note the restart_addr from unmap_vmas when it breaks out:
L
Linus Torvalds 已提交
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
 * 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;

2156 2157
	/*
	 * files that support invalidating or truncating portions of the
N
Nick Piggin 已提交
2158
	 * file from under mmaped areas must have their ->fault function
N
Nick Piggin 已提交
2159 2160
	 * return a locked page (and set VM_FAULT_LOCKED in the return).
	 * This provides synchronisation against concurrent unmapping here.
2161 2162
	 */

L
Linus Torvalds 已提交
2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173
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;
		}
	}

2174 2175
	restart_addr = zap_page_range(vma, start_addr,
					end_addr - start_addr, details);
N
Nick Piggin 已提交
2176
	need_break = need_resched() || spin_needbreak(details->i_mmap_lock);
L
Linus Torvalds 已提交
2177

2178
	if (restart_addr >= end_addr) {
L
Linus Torvalds 已提交
2179 2180 2181 2182 2183 2184
		/* 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 */
2185
		vma->vm_truncate_count = restart_addr;
L
Linus Torvalds 已提交
2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251
		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;
	}
}

/**
2252
 * 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
2253
 * @mapping: the address space containing mmaps to be unmapped.
L
Linus Torvalds 已提交
2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289
 * @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);

2290
	/* Protect against endless unmapping loops */
L
Linus Torvalds 已提交
2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306
	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);

2307 2308 2309 2310
/**
 * 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
2311 2312 2313 2314 2315 2316 2317
 *
 * 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
2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328
	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
2329

C
Christoph Hellwig 已提交
2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351
		/*
		 * 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);
	}
2352

A
Al Viro 已提交
2353
	if (inode->i_op->truncate)
L
Linus Torvalds 已提交
2354 2355
		inode->i_op->truncate(inode);
	return 0;
C
Christoph Hellwig 已提交
2356

L
Linus Torvalds 已提交
2357 2358 2359 2360 2361 2362 2363
out_sig:
	send_sig(SIGXFSZ, current, 0);
out_big:
	return -EFBIG;
}
EXPORT_SYMBOL(vmtruncate);

2364 2365 2366 2367 2368 2369 2370 2371 2372
int vmtruncate_range(struct inode *inode, loff_t offset, loff_t end)
{
	struct address_space *mapping = inode->i_mapping;

	/*
	 * If the underlying filesystem is not going to provide
	 * a way to truncate a range of blocks (punch a hole) -
	 * we should return failure right now.
	 */
A
Al Viro 已提交
2373
	if (!inode->i_op->truncate_range)
2374 2375
		return -ENOSYS;

2376
	mutex_lock(&inode->i_mutex);
2377 2378 2379
	down_write(&inode->i_alloc_sem);
	unmap_mapping_range(mapping, offset, (end - offset), 1);
	truncate_inode_pages_range(mapping, offset, end);
2380
	unmap_mapping_range(mapping, offset, (end - offset), 1);
2381 2382
	inode->i_op->truncate_range(inode, offset, end);
	up_write(&inode->i_alloc_sem);
2383
	mutex_unlock(&inode->i_mutex);
2384 2385 2386 2387

	return 0;
}

L
Linus Torvalds 已提交
2388
/*
2389 2390 2391
 * 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
2392
 */
2393 2394 2395
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
2396
{
2397
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2398
	struct page *page;
2399
	swp_entry_t entry;
L
Linus Torvalds 已提交
2400
	pte_t pte;
2401
	struct mem_cgroup *ptr = NULL;
N
Nick Piggin 已提交
2402
	int ret = 0;
L
Linus Torvalds 已提交
2403

H
Hugh Dickins 已提交
2404
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
2405
		goto out;
2406 2407

	entry = pte_to_swp_entry(orig_pte);
2408 2409 2410 2411
	if (is_migration_entry(entry)) {
		migration_entry_wait(mm, pmd, address);
		goto out;
	}
2412
	delayacct_set_flag(DELAYACCT_PF_SWAPIN);
L
Linus Torvalds 已提交
2413 2414
	page = lookup_swap_cache(entry);
	if (!page) {
2415
		grab_swap_token(); /* Contend for token _before_ read-in */
2416 2417
		page = swapin_readahead(entry,
					GFP_HIGHUSER_MOVABLE, vma, address);
L
Linus Torvalds 已提交
2418 2419
		if (!page) {
			/*
2420 2421
			 * Back out if somebody else faulted in this pte
			 * while we released the pte lock.
L
Linus Torvalds 已提交
2422
			 */
2423
			page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
2424 2425
			if (likely(pte_same(*page_table, orig_pte)))
				ret = VM_FAULT_OOM;
2426
			delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
2427
			goto unlock;
L
Linus Torvalds 已提交
2428 2429 2430 2431
		}

		/* Had to read the page from swap area: Major fault */
		ret = VM_FAULT_MAJOR;
2432
		count_vm_event(PGMAJFAULT);
L
Linus Torvalds 已提交
2433 2434
	}

2435 2436 2437 2438 2439
	mark_page_accessed(page);

	lock_page(page);
	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);

K
KAMEZAWA Hiroyuki 已提交
2440
	if (mem_cgroup_try_charge_swapin(mm, page, GFP_KERNEL, &ptr)) {
2441
		ret = VM_FAULT_OOM;
2442
		unlock_page(page);
2443 2444 2445
		goto out;
	}

L
Linus Torvalds 已提交
2446
	/*
2447
	 * Back out if somebody else already faulted in this pte.
L
Linus Torvalds 已提交
2448
	 */
2449
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
H
Hugh Dickins 已提交
2450
	if (unlikely(!pte_same(*page_table, orig_pte)))
2451 2452 2453 2454 2455
		goto out_nomap;

	if (unlikely(!PageUptodate(page))) {
		ret = VM_FAULT_SIGBUS;
		goto out_nomap;
L
Linus Torvalds 已提交
2456 2457
	}

2458 2459 2460 2461 2462 2463 2464 2465
	/*
	 * The page isn't present yet, go ahead with the fault.
	 *
	 * Be careful about the sequence of operations here.
	 * To get its accounting right, reuse_swap_page() must be called
	 * while the page is counted on swap but not yet in mapcount i.e.
	 * before page_add_anon_rmap() and swap_free(); try_to_free_swap()
	 * must be called after the swap_free(), or it will never succeed.
2466 2467 2468 2469
	 * Because delete_from_swap_page() may be called by reuse_swap_page(),
	 * mem_cgroup_commit_charge_swapin() may not be able to find swp_entry
	 * in page->private. In this case, a record in swap_cgroup  is silently
	 * discarded at swap_free().
2470
	 */
L
Linus Torvalds 已提交
2471

2472
	inc_mm_counter(mm, anon_rss);
L
Linus Torvalds 已提交
2473
	pte = mk_pte(page, vma->vm_page_prot);
2474
	if (write_access && reuse_swap_page(page)) {
L
Linus Torvalds 已提交
2475 2476 2477 2478 2479 2480
		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);
2481 2482
	/* It's better to call commit-charge after rmap is established */
	mem_cgroup_commit_charge_swapin(page, ptr);
L
Linus Torvalds 已提交
2483

2484
	swap_free(entry);
N
Nick Piggin 已提交
2485
	if (vm_swap_full() || (vma->vm_flags & VM_LOCKED) || PageMlocked(page))
2486
		try_to_free_swap(page);
2487 2488
	unlock_page(page);

L
Linus Torvalds 已提交
2489
	if (write_access) {
2490 2491 2492
		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
2493 2494 2495 2496 2497
		goto out;
	}

	/* No need to invalidate - it was non-present before */
	update_mmu_cache(vma, address, pte);
2498
unlock:
2499
	pte_unmap_unlock(page_table, ptl);
L
Linus Torvalds 已提交
2500 2501
out:
	return ret;
2502
out_nomap:
2503
	mem_cgroup_cancel_charge_swapin(ptr);
2504
	pte_unmap_unlock(page_table, ptl);
2505 2506
	unlock_page(page);
	page_cache_release(page);
2507
	return ret;
L
Linus Torvalds 已提交
2508 2509 2510
}

/*
2511 2512 2513
 * 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
2514
 */
2515 2516 2517
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
2518
{
2519 2520
	struct page *page;
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2521 2522
	pte_t entry;

N
Nick Piggin 已提交
2523 2524
	/* Allocate our own private page. */
	pte_unmap(page_table);
2525

N
Nick Piggin 已提交
2526 2527 2528 2529 2530
	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
2531
	__SetPageUptodate(page);
2532

K
KAMEZAWA Hiroyuki 已提交
2533
	if (mem_cgroup_newpage_charge(page, mm, GFP_KERNEL))
2534 2535
		goto oom_free_page;

N
Nick Piggin 已提交
2536 2537
	entry = mk_pte(page, vma->vm_page_prot);
	entry = maybe_mkwrite(pte_mkdirty(entry), vma);
L
Linus Torvalds 已提交
2538

N
Nick Piggin 已提交
2539 2540 2541 2542 2543
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
	if (!pte_none(*page_table))
		goto release;
	inc_mm_counter(mm, anon_rss);
	page_add_new_anon_rmap(page, vma, address);
2544
	set_pte_at(mm, address, page_table, entry);
L
Linus Torvalds 已提交
2545 2546

	/* No need to invalidate - it was non-present before */
2547 2548
	update_mmu_cache(vma, address, entry);
unlock:
2549
	pte_unmap_unlock(page_table, ptl);
N
Nick Piggin 已提交
2550
	return 0;
2551
release:
2552
	mem_cgroup_uncharge_page(page);
2553 2554
	page_cache_release(page);
	goto unlock;
2555
oom_free_page:
2556
	page_cache_release(page);
2557
oom:
L
Linus Torvalds 已提交
2558 2559 2560 2561
	return VM_FAULT_OOM;
}

/*
2562
 * __do_fault() tries to create a new page mapping. It aggressively
L
Linus Torvalds 已提交
2563
 * tries to share with existing pages, but makes a separate copy if
2564 2565
 * the FAULT_FLAG_WRITE is set in the flags parameter in order to avoid
 * the next page fault.
L
Linus Torvalds 已提交
2566 2567 2568 2569
 *
 * As this is called only for pages that do not currently exist, we
 * do not need to flush old virtual caches or the TLB.
 *
2570
 * We enter with non-exclusive mmap_sem (to exclude vma changes,
2571
 * but allow concurrent faults), and pte neither mapped nor locked.
2572
 * We return with mmap_sem still held, but pte unmapped and unlocked.
L
Linus Torvalds 已提交
2573
 */
2574
static int __do_fault(struct mm_struct *mm, struct vm_area_struct *vma,
2575
		unsigned long address, pmd_t *pmd,
2576
		pgoff_t pgoff, unsigned int flags, pte_t orig_pte)
L
Linus Torvalds 已提交
2577
{
2578
	pte_t *page_table;
2579
	spinlock_t *ptl;
N
Nick Piggin 已提交
2580
	struct page *page;
L
Linus Torvalds 已提交
2581 2582
	pte_t entry;
	int anon = 0;
2583
	int charged = 0;
2584
	struct page *dirty_page = NULL;
N
Nick Piggin 已提交
2585 2586
	struct vm_fault vmf;
	int ret;
2587
	int page_mkwrite = 0;
2588

N
Nick Piggin 已提交
2589 2590 2591 2592
	vmf.virtual_address = (void __user *)(address & PAGE_MASK);
	vmf.pgoff = pgoff;
	vmf.flags = flags;
	vmf.page = NULL;
L
Linus Torvalds 已提交
2593

N
Nick Piggin 已提交
2594 2595 2596
	ret = vma->vm_ops->fault(vma, &vmf);
	if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE)))
		return ret;
L
Linus Torvalds 已提交
2597

2598
	/*
N
Nick Piggin 已提交
2599
	 * For consistency in subsequent calls, make the faulted page always
2600 2601
	 * locked.
	 */
N
Nick Piggin 已提交
2602
	if (unlikely(!(ret & VM_FAULT_LOCKED)))
N
Nick Piggin 已提交
2603
		lock_page(vmf.page);
2604
	else
N
Nick Piggin 已提交
2605
		VM_BUG_ON(!PageLocked(vmf.page));
2606

L
Linus Torvalds 已提交
2607 2608 2609
	/*
	 * Should we do an early C-O-W break?
	 */
N
Nick Piggin 已提交
2610
	page = vmf.page;
2611
	if (flags & FAULT_FLAG_WRITE) {
2612
		if (!(vma->vm_flags & VM_SHARED)) {
2613
			anon = 1;
2614
			if (unlikely(anon_vma_prepare(vma))) {
N
Nick Piggin 已提交
2615
				ret = VM_FAULT_OOM;
2616
				goto out;
2617
			}
N
Nick Piggin 已提交
2618 2619
			page = alloc_page_vma(GFP_HIGHUSER_MOVABLE,
						vma, address);
2620
			if (!page) {
N
Nick Piggin 已提交
2621
				ret = VM_FAULT_OOM;
2622
				goto out;
2623
			}
K
KAMEZAWA Hiroyuki 已提交
2624
			if (mem_cgroup_newpage_charge(page, mm, GFP_KERNEL)) {
2625 2626 2627 2628 2629
				ret = VM_FAULT_OOM;
				page_cache_release(page);
				goto out;
			}
			charged = 1;
N
Nick Piggin 已提交
2630 2631 2632 2633 2634 2635
			/*
			 * 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
2636
			copy_user_highpage(page, vmf.page, address, vma);
N
Nick Piggin 已提交
2637
			__SetPageUptodate(page);
2638
		} else {
2639 2640
			/*
			 * If the page will be shareable, see if the backing
2641
			 * address space wants to know that the page is about
2642 2643
			 * to become writable
			 */
2644 2645 2646
			if (vma->vm_ops->page_mkwrite) {
				unlock_page(page);
				if (vma->vm_ops->page_mkwrite(vma, page) < 0) {
N
Nick Piggin 已提交
2647 2648
					ret = VM_FAULT_SIGBUS;
					anon = 1; /* no anon but release vmf.page */
2649 2650 2651
					goto out_unlocked;
				}
				lock_page(page);
N
Nick Piggin 已提交
2652 2653 2654 2655 2656 2657 2658 2659
				/*
				 * 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
2660
					ret = 0;
N
Nick Piggin 已提交
2661 2662 2663
					anon = 1; /* no anon but release vmf.page */
					goto out;
				}
2664
				page_mkwrite = 1;
2665 2666
			}
		}
2667

L
Linus Torvalds 已提交
2668 2669
	}

2670
	page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
L
Linus Torvalds 已提交
2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682

	/*
	 * 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... */
2683
	if (likely(pte_same(*page_table, orig_pte))) {
2684 2685
		flush_icache_page(vma, page);
		entry = mk_pte(page, vma->vm_page_prot);
2686
		if (flags & FAULT_FLAG_WRITE)
L
Linus Torvalds 已提交
2687 2688
			entry = maybe_mkwrite(pte_mkdirty(entry), vma);
		if (anon) {
2689 2690
			inc_mm_counter(mm, anon_rss);
			page_add_new_anon_rmap(page, vma, address);
2691
		} else {
2692
			inc_mm_counter(mm, file_rss);
2693
			page_add_file_rmap(page);
2694
			if (flags & FAULT_FLAG_WRITE) {
2695
				dirty_page = page;
2696 2697
				get_page(dirty_page);
			}
2698
		}
2699
		set_pte_at(mm, address, page_table, entry);
2700 2701 2702

		/* no need to invalidate: a not-present page won't be cached */
		update_mmu_cache(vma, address, entry);
L
Linus Torvalds 已提交
2703
	} else {
2704 2705
		if (charged)
			mem_cgroup_uncharge_page(page);
2706 2707 2708
		if (anon)
			page_cache_release(page);
		else
2709
			anon = 1; /* no anon but release faulted_page */
L
Linus Torvalds 已提交
2710 2711
	}

2712
	pte_unmap_unlock(page_table, ptl);
2713 2714

out:
N
Nick Piggin 已提交
2715
	unlock_page(vmf.page);
2716
out_unlocked:
2717
	if (anon)
N
Nick Piggin 已提交
2718
		page_cache_release(vmf.page);
2719
	else if (dirty_page) {
2720 2721 2722
		if (vma->vm_file)
			file_update_time(vma->vm_file);

2723
		set_page_dirty_balance(dirty_page, page_mkwrite);
2724 2725
		put_page(dirty_page);
	}
2726

N
Nick Piggin 已提交
2727
	return ret;
2728
}
2729

2730 2731 2732 2733 2734
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)
2735
			- vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
2736 2737
	unsigned int flags = (write_access ? FAULT_FLAG_WRITE : 0);

2738 2739
	pte_unmap(page_table);
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
2740 2741
}

L
Linus Torvalds 已提交
2742 2743 2744 2745
/*
 * Fault of a previously existing named mapping. Repopulate the pte
 * from the encoded file_pte if possible. This enables swappable
 * nonlinear vmas.
2746 2747 2748 2749
 *
 * 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
2750
 */
N
Nick Piggin 已提交
2751
static int do_nonlinear_fault(struct mm_struct *mm, struct vm_area_struct *vma,
2752 2753
		unsigned long address, pte_t *page_table, pmd_t *pmd,
		int write_access, pte_t orig_pte)
L
Linus Torvalds 已提交
2754
{
N
Nick Piggin 已提交
2755 2756
	unsigned int flags = FAULT_FLAG_NONLINEAR |
				(write_access ? FAULT_FLAG_WRITE : 0);
2757
	pgoff_t pgoff;
L
Linus Torvalds 已提交
2758

H
Hugh Dickins 已提交
2759
	if (!pte_unmap_same(mm, pmd, page_table, orig_pte))
N
Nick Piggin 已提交
2760
		return 0;
L
Linus Torvalds 已提交
2761

2762
	if (unlikely(!(vma->vm_flags & VM_NONLINEAR))) {
2763 2764 2765
		/*
		 * Page table corrupted: show pte and kill process.
		 */
2766
		print_bad_pte(vma, address, orig_pte, NULL);
2767 2768 2769 2770
		return VM_FAULT_OOM;
	}

	pgoff = pte_to_pgoff(orig_pte);
2771
	return __do_fault(mm, vma, address, pmd, pgoff, flags, orig_pte);
L
Linus Torvalds 已提交
2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782
}

/*
 * 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
2783 2784 2785
 * 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
2786 2787
 */
static inline int handle_pte_fault(struct mm_struct *mm,
2788 2789
		struct vm_area_struct *vma, unsigned long address,
		pte_t *pte, pmd_t *pmd, int write_access)
L
Linus Torvalds 已提交
2790 2791
{
	pte_t entry;
2792
	spinlock_t *ptl;
L
Linus Torvalds 已提交
2793

2794
	entry = *pte;
L
Linus Torvalds 已提交
2795
	if (!pte_present(entry)) {
2796
		if (pte_none(entry)) {
J
Jes Sorensen 已提交
2797
			if (vma->vm_ops) {
N
Nick Piggin 已提交
2798
				if (likely(vma->vm_ops->fault))
2799 2800
					return do_linear_fault(mm, vma, address,
						pte, pmd, write_access, entry);
J
Jes Sorensen 已提交
2801 2802 2803
			}
			return do_anonymous_page(mm, vma, address,
						 pte, pmd, write_access);
2804
		}
L
Linus Torvalds 已提交
2805
		if (pte_file(entry))
N
Nick Piggin 已提交
2806
			return do_nonlinear_fault(mm, vma, address,
2807 2808 2809
					pte, pmd, write_access, entry);
		return do_swap_page(mm, vma, address,
					pte, pmd, write_access, entry);
L
Linus Torvalds 已提交
2810 2811
	}

H
Hugh Dickins 已提交
2812
	ptl = pte_lockptr(mm, pmd);
2813 2814 2815
	spin_lock(ptl);
	if (unlikely(!pte_same(*pte, entry)))
		goto unlock;
L
Linus Torvalds 已提交
2816 2817
	if (write_access) {
		if (!pte_write(entry))
2818 2819
			return do_wp_page(mm, vma, address,
					pte, pmd, ptl, entry);
L
Linus Torvalds 已提交
2820 2821 2822
		entry = pte_mkdirty(entry);
	}
	entry = pte_mkyoung(entry);
2823
	if (ptep_set_access_flags(vma, address, pte, entry, write_access)) {
2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834
		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);
	}
2835 2836
unlock:
	pte_unmap_unlock(pte, ptl);
N
Nick Piggin 已提交
2837
	return 0;
L
Linus Torvalds 已提交
2838 2839 2840 2841 2842
}

/*
 * By the time we get here, we already hold the mm semaphore
 */
N
Nick Piggin 已提交
2843
int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
L
Linus Torvalds 已提交
2844 2845 2846 2847 2848 2849 2850 2851 2852
		unsigned long address, int write_access)
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

	__set_current_state(TASK_RUNNING);

2853
	count_vm_event(PGFAULT);
L
Linus Torvalds 已提交
2854

2855 2856
	if (unlikely(is_vm_hugetlb_page(vma)))
		return hugetlb_fault(mm, vma, address, write_access);
L
Linus Torvalds 已提交
2857 2858 2859 2860

	pgd = pgd_offset(mm, address);
	pud = pud_alloc(mm, pgd, address);
	if (!pud)
H
Hugh Dickins 已提交
2861
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
2862 2863
	pmd = pmd_alloc(mm, pud, address);
	if (!pmd)
H
Hugh Dickins 已提交
2864
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
2865 2866
	pte = pte_alloc_map(mm, pmd, address);
	if (!pte)
H
Hugh Dickins 已提交
2867
		return VM_FAULT_OOM;
L
Linus Torvalds 已提交
2868

H
Hugh Dickins 已提交
2869
	return handle_pte_fault(mm, vma, address, pte, pmd, write_access);
L
Linus Torvalds 已提交
2870 2871 2872 2873 2874
}

#ifndef __PAGETABLE_PUD_FOLDED
/*
 * Allocate page upper directory.
H
Hugh Dickins 已提交
2875
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
2876
 */
2877
int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
L
Linus Torvalds 已提交
2878
{
H
Hugh Dickins 已提交
2879 2880
	pud_t *new = pud_alloc_one(mm, address);
	if (!new)
2881
		return -ENOMEM;
L
Linus Torvalds 已提交
2882

2883 2884
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
2885
	spin_lock(&mm->page_table_lock);
2886
	if (pgd_present(*pgd))		/* Another has populated it */
2887
		pud_free(mm, new);
2888 2889
	else
		pgd_populate(mm, pgd, new);
H
Hugh Dickins 已提交
2890
	spin_unlock(&mm->page_table_lock);
2891
	return 0;
L
Linus Torvalds 已提交
2892 2893 2894 2895 2896 2897
}
#endif /* __PAGETABLE_PUD_FOLDED */

#ifndef __PAGETABLE_PMD_FOLDED
/*
 * Allocate page middle directory.
H
Hugh Dickins 已提交
2898
 * We've already handled the fast-path in-line.
L
Linus Torvalds 已提交
2899
 */
2900
int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
L
Linus Torvalds 已提交
2901
{
H
Hugh Dickins 已提交
2902 2903
	pmd_t *new = pmd_alloc_one(mm, address);
	if (!new)
2904
		return -ENOMEM;
L
Linus Torvalds 已提交
2905

2906 2907
	smp_wmb(); /* See comment in __pte_alloc */

H
Hugh Dickins 已提交
2908
	spin_lock(&mm->page_table_lock);
L
Linus Torvalds 已提交
2909
#ifndef __ARCH_HAS_4LEVEL_HACK
2910
	if (pud_present(*pud))		/* Another has populated it */
2911
		pmd_free(mm, new);
2912 2913
	else
		pud_populate(mm, pud, new);
L
Linus Torvalds 已提交
2914
#else
2915
	if (pgd_present(*pud))		/* Another has populated it */
2916
		pmd_free(mm, new);
2917 2918
	else
		pgd_populate(mm, pud, new);
L
Linus Torvalds 已提交
2919
#endif /* __ARCH_HAS_4LEVEL_HACK */
H
Hugh Dickins 已提交
2920
	spin_unlock(&mm->page_table_lock);
2921
	return 0;
2922
}
L
Linus Torvalds 已提交
2923 2924 2925 2926 2927 2928 2929 2930 2931
#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
2932
		return -ENOMEM;
L
Linus Torvalds 已提交
2933
	write = (vma->vm_flags & VM_WRITE) != 0;
2934 2935
	BUG_ON(addr >= end);
	BUG_ON(end > vma->vm_end);
2936
	len = DIV_ROUND_UP(end, PAGE_SIZE) - addr/PAGE_SIZE;
L
Linus Torvalds 已提交
2937 2938
	ret = get_user_pages(current, current->mm, addr,
			len, write, 0, NULL, NULL);
2939
	if (ret < 0)
L
Linus Torvalds 已提交
2940
		return ret;
2941
	return ret == len ? 0 : -EFAULT;
L
Linus Torvalds 已提交
2942 2943 2944 2945 2946
}

#if !defined(__HAVE_ARCH_GATE_AREA)

#if defined(AT_SYSINFO_EHDR)
2947
static struct vm_area_struct gate_vma;
L
Linus Torvalds 已提交
2948 2949 2950 2951 2952 2953

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
2954 2955
	gate_vma.vm_flags = VM_READ | VM_MAYREAD | VM_EXEC | VM_MAYEXEC;
	gate_vma.vm_page_prot = __P101;
2956 2957 2958 2959 2960 2961 2962
	/*
	 * 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
2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986
	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 */
2987

2988
#ifdef CONFIG_HAVE_IOREMAP_PROT
2989 2990 2991
int follow_phys(struct vm_area_struct *vma,
		unsigned long address, unsigned int flags,
		unsigned long *prot, resource_size_t *phys)
2992 2993 2994 2995 2996 2997 2998 2999
{
	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;
3000
	int ret = -EINVAL;
3001

3002 3003
	if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
		goto out;
3004 3005 3006

	pgd = pgd_offset(mm, address);
	if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
3007
		goto out;
3008 3009 3010

	pud = pud_offset(pgd, address);
	if (pud_none(*pud) || unlikely(pud_bad(*pud)))
3011
		goto out;
3012 3013 3014

	pmd = pmd_offset(pud, address);
	if (pmd_none(*pmd) || unlikely(pmd_bad(*pmd)))
3015
		goto out;
3016 3017 3018

	/* We cannot handle huge page PFN maps. Luckily they don't exist. */
	if (pmd_huge(*pmd))
3019
		goto out;
3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033

	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));
3034 3035
	*phys = phys_addr;
	ret = 0;
3036 3037 3038 3039

unlock:
	pte_unmap_unlock(ptep, ptl);
out:
3040
	return ret;
3041 3042 3043 3044 3045 3046 3047
}

int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
			void *buf, int len, int write)
{
	resource_size_t phys_addr;
	unsigned long prot = 0;
K
KOSAKI Motohiro 已提交
3048
	void __iomem *maddr;
3049 3050
	int offset = addr & (PAGE_SIZE-1);

3051
	if (follow_phys(vma, addr, write, &prot, &phys_addr))
3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064
		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

3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080
/*
 * 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
3081
	/* ignore errors, just check how much was successfully transferred */
3082 3083 3084
	while (len) {
		int bytes, ret, offset;
		void *maddr;
3085
		struct page *page = NULL;
3086 3087 3088

		ret = get_user_pages(tsk, mm, addr, 1,
				write, 1, &page, &vma);
3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104
		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;
3105
		} else {
3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121
			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);
3122 3123 3124 3125 3126 3127 3128 3129 3130 3131
		}
		len -= bytes;
		buf += bytes;
		addr += bytes;
	}
	up_read(&mm->mmap_sem);
	mmput(mm);

	return buf - old_buf;
}
3132 3133 3134 3135 3136 3137 3138 3139 3140

/*
 * 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;

3141 3142 3143 3144 3145 3146 3147
	/*
	 * Do not print if we are in atomic
	 * contexts (in exception stacks, etc.):
	 */
	if (preempt_count())
		return;

3148 3149 3150 3151 3152 3153 3154 3155
	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;

3156
			p = d_path(&f->f_path, buf, PAGE_SIZE);
3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169
			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);
}
3170 3171 3172 3173

#ifdef CONFIG_PROVE_LOCKING
void might_fault(void)
{
3174 3175 3176 3177 3178 3179 3180 3181 3182
	/*
	 * Some code (nfs/sunrpc) uses socket ops on kernel memory while
	 * holding the mmap_sem, this is safe because kernel memory doesn't
	 * get paged out, therefore we'll never actually fault, and the
	 * below annotations will generate false positives.
	 */
	if (segment_eq(get_fs(), KERNEL_DS))
		return;

3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193
	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