blob: 8aca6f7167bbb25f858075c90c64d766fcc1e4c3 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * linux/mm/mincore.c
3 *
Linus Torvalds2f77d102006-12-16 09:44:32 -08004 * Copyright (C) 1994-2006 Linus Torvalds
Linus Torvalds1da177e2005-04-16 15:20:36 -07005 */
6
7/*
8 * The mincore() system call.
9 */
10#include <linux/slab.h>
11#include <linux/pagemap.h>
12#include <linux/mm.h>
13#include <linux/mman.h>
14#include <linux/syscalls.h>
15
16#include <asm/uaccess.h>
17#include <asm/pgtable.h>
18
19/*
20 * Later we can get more picky about what "in core" means precisely.
21 * For now, simply check to see if the page is in the page cache,
22 * and is up to date; i.e. that no page-in operation would be required
23 * at this time if an application were to map and access this page.
24 */
25static unsigned char mincore_page(struct vm_area_struct * vma,
26 unsigned long pgoff)
27{
28 unsigned char present = 0;
29 struct address_space * as = vma->vm_file->f_mapping;
30 struct page * page;
31
32 page = find_get_page(as, pgoff);
33 if (page) {
34 present = PageUptodate(page);
35 page_cache_release(page);
36 }
37
38 return present;
39}
40
Linus Torvalds2f77d102006-12-16 09:44:32 -080041/*
42 * Do a chunk of "sys_mincore()". We've already checked
43 * all the arguments, we hold the mmap semaphore: we should
44 * just return the amount of info we're asked for.
45 */
46static long do_mincore(unsigned long addr, unsigned char *vec, unsigned long pages)
Linus Torvalds1da177e2005-04-16 15:20:36 -070047{
Linus Torvalds2f77d102006-12-16 09:44:32 -080048 unsigned long i, nr, pgoff;
49 struct vm_area_struct *vma = find_vma(current->mm, addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -070050
Linus Torvalds2f77d102006-12-16 09:44:32 -080051 /*
Linus Torvalds4fb23e42006-12-16 16:01:50 -080052 * find_vma() didn't find anything above us, or we're
53 * in an unmapped hole in the address space: ENOMEM.
Linus Torvalds2f77d102006-12-16 09:44:32 -080054 */
Linus Torvalds4fb23e42006-12-16 16:01:50 -080055 if (!vma || addr < vma->vm_start)
56 return -ENOMEM;
Linus Torvalds2f77d102006-12-16 09:44:32 -080057
58 /*
59 * Ok, got it. But check whether it's a segment we support
60 * mincore() on. Right now, we don't do any anonymous mappings.
Linus Torvalds4fb23e42006-12-16 16:01:50 -080061 *
62 * FIXME: This is just stupid. And returning ENOMEM is
63 * stupid too. We should just look at the page tables. But
64 * this is what we've traditionally done, so we'll just
65 * continue doing it.
Linus Torvalds2f77d102006-12-16 09:44:32 -080066 */
67 if (!vma->vm_file)
68 return -ENOMEM;
69
70 /*
71 * Calculate how many pages there are left in the vma, and
72 * what the pgoff is for our address.
73 */
74 nr = (vma->vm_end - addr) >> PAGE_SHIFT;
75 if (nr > pages)
76 nr = pages;
77
78 pgoff = (addr - vma->vm_start) >> PAGE_SHIFT;
79 pgoff += vma->vm_pgoff;
80
81 /* And then we just fill the sucker in.. */
82 for (i = 0 ; i < nr; i++, pgoff++)
83 vec[i] = mincore_page(vma, pgoff);
84
85 return nr;
Linus Torvalds1da177e2005-04-16 15:20:36 -070086}
87
88/*
89 * The mincore(2) system call.
90 *
91 * mincore() returns the memory residency status of the pages in the
92 * current process's address space specified by [addr, addr + len).
93 * The status is returned in a vector of bytes. The least significant
94 * bit of each byte is 1 if the referenced page is in memory, otherwise
95 * it is zero.
96 *
97 * Because the status of a page can change after mincore() checks it
98 * but before it returns to the application, the returned vector may
99 * contain stale information. Only locked pages are guaranteed to
100 * remain in memory.
101 *
102 * return values:
103 * zero - success
104 * -EFAULT - vec points to an illegal address
105 * -EINVAL - addr is not a multiple of PAGE_CACHE_SIZE
106 * -ENOMEM - Addresses in the range [addr, addr + len] are
107 * invalid for the address space of this process, or
108 * specify one or more pages which are not currently
109 * mapped
110 * -EAGAIN - A kernel resource was temporarily unavailable.
111 */
112asmlinkage long sys_mincore(unsigned long start, size_t len,
113 unsigned char __user * vec)
114{
Linus Torvalds2f77d102006-12-16 09:44:32 -0800115 long retval;
116 unsigned long pages;
117 unsigned char *tmp;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700118
Linus Torvalds2f77d102006-12-16 09:44:32 -0800119 /* Check the start address: needs to be page-aligned.. */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700120 if (start & ~PAGE_CACHE_MASK)
Linus Torvalds2f77d102006-12-16 09:44:32 -0800121 return -EINVAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700122
Linus Torvalds2f77d102006-12-16 09:44:32 -0800123 /* ..and we need to be passed a valid user-space range */
124 if (!access_ok(VERIFY_READ, (void __user *) start, len))
125 return -ENOMEM;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700126
Linus Torvalds2f77d102006-12-16 09:44:32 -0800127 /* This also avoids any overflows on PAGE_CACHE_ALIGN */
128 pages = len >> PAGE_SHIFT;
129 pages += (len & ~PAGE_MASK) != 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700130
Linus Torvalds2f77d102006-12-16 09:44:32 -0800131 if (!access_ok(VERIFY_WRITE, vec, pages))
132 return -EFAULT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700133
Linus Torvalds2f77d102006-12-16 09:44:32 -0800134 tmp = (void *) __get_free_page(GFP_USER);
135 if (!tmp)
Linus Torvalds4fb23e42006-12-16 16:01:50 -0800136 return -EAGAIN;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700137
Linus Torvalds2f77d102006-12-16 09:44:32 -0800138 retval = 0;
139 while (pages) {
140 /*
141 * Do at most PAGE_SIZE entries per iteration, due to
142 * the temporary buffer size.
143 */
144 down_read(&current->mm->mmap_sem);
Oleg Nesterov825020c2006-12-17 18:52:47 +0300145 retval = do_mincore(start, tmp, min(pages, PAGE_SIZE));
Linus Torvalds2f77d102006-12-16 09:44:32 -0800146 up_read(&current->mm->mmap_sem);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700147
Linus Torvalds2f77d102006-12-16 09:44:32 -0800148 if (retval <= 0)
149 break;
150 if (copy_to_user(vec, tmp, retval)) {
151 retval = -EFAULT;
152 break;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700153 }
Linus Torvalds2f77d102006-12-16 09:44:32 -0800154 pages -= retval;
155 vec += retval;
156 start += retval << PAGE_SHIFT;
157 retval = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700158 }
Linus Torvalds2f77d102006-12-16 09:44:32 -0800159 free_page((unsigned long) tmp);
160 return retval;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700161}