Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * linux/mm/mincore.c | |
3 | * | |
2f77d107 | 4 | * Copyright (C) 1994-2006 Linus Torvalds |
1da177e4 LT |
5 | */ |
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> | |
42da9cbd NP |
15 | #include <linux/swap.h> |
16 | #include <linux/swapops.h> | |
1da177e4 LT |
17 | |
18 | #include <asm/uaccess.h> | |
19 | #include <asm/pgtable.h> | |
20 | ||
21 | /* | |
22 | * Later we can get more picky about what "in core" means precisely. | |
23 | * For now, simply check to see if the page is in the page cache, | |
24 | * and is up to date; i.e. that no page-in operation would be required | |
25 | * at this time if an application were to map and access this page. | |
26 | */ | |
42da9cbd | 27 | static unsigned char mincore_page(struct address_space *mapping, pgoff_t pgoff) |
1da177e4 LT |
28 | { |
29 | unsigned char present = 0; | |
42da9cbd | 30 | struct page *page; |
1da177e4 | 31 | |
42da9cbd NP |
32 | /* |
33 | * When tmpfs swaps out a page from a file, any process mapping that | |
34 | * file will not get a swp_entry_t in its pte, but rather it is like | |
35 | * any other file mapping (ie. marked !present and faulted in with | |
36 | * tmpfs's .nopage). So swapped out tmpfs mappings are tested here. | |
37 | * | |
38 | * However when tmpfs moves the page from pagecache and into swapcache, | |
39 | * it is still in core, but the find_get_page below won't find it. | |
40 | * No big deal, but make a note of it. | |
41 | */ | |
42 | page = find_get_page(mapping, pgoff); | |
1da177e4 LT |
43 | if (page) { |
44 | present = PageUptodate(page); | |
45 | page_cache_release(page); | |
46 | } | |
47 | ||
48 | return present; | |
49 | } | |
50 | ||
2f77d107 LT |
51 | /* |
52 | * Do a chunk of "sys_mincore()". We've already checked | |
53 | * all the arguments, we hold the mmap semaphore: we should | |
54 | * just return the amount of info we're asked for. | |
55 | */ | |
56 | static long do_mincore(unsigned long addr, unsigned char *vec, unsigned long pages) | |
1da177e4 | 57 | { |
42da9cbd NP |
58 | pgd_t *pgd; |
59 | pud_t *pud; | |
60 | pmd_t *pmd; | |
61 | pte_t *ptep; | |
62 | spinlock_t *ptl; | |
63 | unsigned long nr; | |
64 | int i; | |
65 | pgoff_t pgoff; | |
2f77d107 | 66 | struct vm_area_struct *vma = find_vma(current->mm, addr); |
1da177e4 | 67 | |
2f77d107 | 68 | /* |
4fb23e43 LT |
69 | * find_vma() didn't find anything above us, or we're |
70 | * in an unmapped hole in the address space: ENOMEM. | |
2f77d107 | 71 | */ |
4fb23e43 LT |
72 | if (!vma || addr < vma->vm_start) |
73 | return -ENOMEM; | |
1da177e4 | 74 | |
2f77d107 | 75 | /* |
42da9cbd NP |
76 | * Calculate how many pages there are left in the last level of the |
77 | * PTE array for our address. | |
2f77d107 | 78 | */ |
42da9cbd | 79 | nr = PTRS_PER_PTE - ((addr >> PAGE_SHIFT) & (PTRS_PER_PTE-1)); |
2f77d107 LT |
80 | if (nr > pages) |
81 | nr = pages; | |
1da177e4 | 82 | |
42da9cbd NP |
83 | pgd = pgd_offset(vma->vm_mm, addr); |
84 | if (pgd_none_or_clear_bad(pgd)) | |
85 | goto none_mapped; | |
86 | pud = pud_offset(pgd, addr); | |
87 | if (pud_none_or_clear_bad(pud)) | |
88 | goto none_mapped; | |
89 | pmd = pmd_offset(pud, addr); | |
90 | if (pmd_none_or_clear_bad(pmd)) | |
91 | goto none_mapped; | |
92 | ||
93 | ptep = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl); | |
94 | for (i = 0; i < nr; i++, ptep++, addr += PAGE_SIZE) { | |
95 | unsigned char present; | |
96 | pte_t pte = *ptep; | |
1da177e4 | 97 | |
42da9cbd NP |
98 | if (pte_present(pte)) { |
99 | present = 1; | |
100 | ||
101 | } else if (pte_none(pte)) { | |
102 | if (vma->vm_file) { | |
103 | pgoff = linear_page_index(vma, addr); | |
104 | present = mincore_page(vma->vm_file->f_mapping, | |
105 | pgoff); | |
106 | } else | |
107 | present = 0; | |
108 | ||
109 | } else if (pte_file(pte)) { | |
110 | pgoff = pte_to_pgoff(pte); | |
111 | present = mincore_page(vma->vm_file->f_mapping, pgoff); | |
112 | ||
113 | } else { /* pte is a swap entry */ | |
114 | swp_entry_t entry = pte_to_swp_entry(pte); | |
115 | if (is_migration_entry(entry)) { | |
116 | /* migration entries are always uptodate */ | |
117 | present = 1; | |
118 | } else { | |
119 | pgoff = entry.val; | |
120 | present = mincore_page(&swapper_space, pgoff); | |
121 | } | |
122 | } | |
123 | } | |
124 | pte_unmap_unlock(ptep-1, ptl); | |
125 | ||
126 | return nr; | |
127 | ||
128 | none_mapped: | |
129 | if (vma->vm_file) { | |
130 | pgoff = linear_page_index(vma, addr); | |
131 | for (i = 0; i < nr; i++, pgoff++) | |
132 | vec[i] = mincore_page(vma->vm_file->f_mapping, pgoff); | |
133 | } | |
1da177e4 | 134 | |
2f77d107 | 135 | return nr; |
1da177e4 LT |
136 | } |
137 | ||
138 | /* | |
139 | * The mincore(2) system call. | |
140 | * | |
141 | * mincore() returns the memory residency status of the pages in the | |
142 | * current process's address space specified by [addr, addr + len). | |
143 | * The status is returned in a vector of bytes. The least significant | |
144 | * bit of each byte is 1 if the referenced page is in memory, otherwise | |
145 | * it is zero. | |
146 | * | |
147 | * Because the status of a page can change after mincore() checks it | |
148 | * but before it returns to the application, the returned vector may | |
149 | * contain stale information. Only locked pages are guaranteed to | |
150 | * remain in memory. | |
151 | * | |
152 | * return values: | |
153 | * zero - success | |
154 | * -EFAULT - vec points to an illegal address | |
155 | * -EINVAL - addr is not a multiple of PAGE_CACHE_SIZE | |
156 | * -ENOMEM - Addresses in the range [addr, addr + len] are | |
157 | * invalid for the address space of this process, or | |
158 | * specify one or more pages which are not currently | |
159 | * mapped | |
160 | * -EAGAIN - A kernel resource was temporarily unavailable. | |
161 | */ | |
162 | asmlinkage long sys_mincore(unsigned long start, size_t len, | |
163 | unsigned char __user * vec) | |
164 | { | |
2f77d107 LT |
165 | long retval; |
166 | unsigned long pages; | |
167 | unsigned char *tmp; | |
1da177e4 | 168 | |
2f77d107 LT |
169 | /* Check the start address: needs to be page-aligned.. */ |
170 | if (start & ~PAGE_CACHE_MASK) | |
171 | return -EINVAL; | |
1da177e4 | 172 | |
2f77d107 LT |
173 | /* ..and we need to be passed a valid user-space range */ |
174 | if (!access_ok(VERIFY_READ, (void __user *) start, len)) | |
175 | return -ENOMEM; | |
1da177e4 | 176 | |
2f77d107 LT |
177 | /* This also avoids any overflows on PAGE_CACHE_ALIGN */ |
178 | pages = len >> PAGE_SHIFT; | |
179 | pages += (len & ~PAGE_MASK) != 0; | |
1da177e4 | 180 | |
2f77d107 LT |
181 | if (!access_ok(VERIFY_WRITE, vec, pages)) |
182 | return -EFAULT; | |
1da177e4 | 183 | |
2f77d107 LT |
184 | tmp = (void *) __get_free_page(GFP_USER); |
185 | if (!tmp) | |
4fb23e43 | 186 | return -EAGAIN; |
2f77d107 LT |
187 | |
188 | retval = 0; | |
189 | while (pages) { | |
190 | /* | |
191 | * Do at most PAGE_SIZE entries per iteration, due to | |
192 | * the temporary buffer size. | |
193 | */ | |
194 | down_read(¤t->mm->mmap_sem); | |
825020c3 | 195 | retval = do_mincore(start, tmp, min(pages, PAGE_SIZE)); |
2f77d107 LT |
196 | up_read(¤t->mm->mmap_sem); |
197 | ||
198 | if (retval <= 0) | |
199 | break; | |
200 | if (copy_to_user(vec, tmp, retval)) { | |
201 | retval = -EFAULT; | |
202 | break; | |
1da177e4 | 203 | } |
2f77d107 LT |
204 | pages -= retval; |
205 | vec += retval; | |
206 | start += retval << PAGE_SHIFT; | |
207 | retval = 0; | |
1da177e4 | 208 | } |
2f77d107 LT |
209 | free_page((unsigned long) tmp); |
210 | return retval; | |
1da177e4 | 211 | } |