1
2
3
4
5
6
7
8
9
10
11#include <linux/pagemap.h>
12#include <linux/gfp.h>
13#include <linux/mm.h>
14#include <linux/mman.h>
15#include <linux/syscalls.h>
16#include <linux/swap.h>
17#include <linux/swapops.h>
18#include <linux/shmem_fs.h>
19#include <linux/hugetlb.h>
20
21#include <linux/uaccess.h>
22#include <asm/pgtable.h>
23
24static int mincore_hugetlb(pte_t *pte, unsigned long hmask, unsigned long addr,
25 unsigned long end, struct mm_walk *walk)
26{
27#ifdef CONFIG_HUGETLB_PAGE
28 unsigned char present;
29 unsigned char *vec = walk->private;
30
31
32
33
34
35 present = pte && !huge_pte_none(huge_ptep_get(pte));
36 for (; addr != end; vec++, addr += PAGE_SIZE)
37 *vec = present;
38 walk->private = vec;
39#else
40 BUG();
41#endif
42 return 0;
43}
44
45
46
47
48
49
50
51static unsigned char mincore_page(struct address_space *mapping, pgoff_t pgoff)
52{
53 unsigned char present = 0;
54 struct page *page;
55
56
57
58
59
60
61
62#ifdef CONFIG_SWAP
63 if (shmem_mapping(mapping)) {
64 page = find_get_entry(mapping, pgoff);
65
66
67
68
69 if (xa_is_value(page)) {
70 swp_entry_t swp = radix_to_swp_entry(page);
71 struct swap_info_struct *si;
72
73
74 si = get_swap_device(swp);
75 if (si) {
76 page = find_get_page(swap_address_space(swp),
77 swp_offset(swp));
78 put_swap_device(si);
79 } else
80 page = NULL;
81 }
82 } else
83 page = find_get_page(mapping, pgoff);
84#else
85 page = find_get_page(mapping, pgoff);
86#endif
87 if (page) {
88 present = PageUptodate(page);
89 put_page(page);
90 }
91
92 return present;
93}
94
95static int __mincore_unmapped_range(unsigned long addr, unsigned long end,
96 struct vm_area_struct *vma, unsigned char *vec)
97{
98 unsigned long nr = (end - addr) >> PAGE_SHIFT;
99 int i;
100
101 if (vma->vm_file) {
102 pgoff_t pgoff;
103
104 pgoff = linear_page_index(vma, addr);
105 for (i = 0; i < nr; i++, pgoff++)
106 vec[i] = mincore_page(vma->vm_file->f_mapping, pgoff);
107 } else {
108 for (i = 0; i < nr; i++)
109 vec[i] = 0;
110 }
111 return nr;
112}
113
114static int mincore_unmapped_range(unsigned long addr, unsigned long end,
115 struct mm_walk *walk)
116{
117 walk->private += __mincore_unmapped_range(addr, end,
118 walk->vma, walk->private);
119 return 0;
120}
121
122static int mincore_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
123 struct mm_walk *walk)
124{
125 spinlock_t *ptl;
126 struct vm_area_struct *vma = walk->vma;
127 pte_t *ptep;
128 unsigned char *vec = walk->private;
129 int nr = (end - addr) >> PAGE_SHIFT;
130
131 ptl = pmd_trans_huge_lock(pmd, vma);
132 if (ptl) {
133 memset(vec, 1, nr);
134 spin_unlock(ptl);
135 goto out;
136 }
137
138 if (pmd_trans_unstable(pmd)) {
139 __mincore_unmapped_range(addr, end, vma, vec);
140 goto out;
141 }
142
143 ptep = pte_offset_map_lock(walk->mm, pmd, addr, &ptl);
144 for (; addr != end; ptep++, addr += PAGE_SIZE) {
145 pte_t pte = *ptep;
146
147 if (pte_none(pte))
148 __mincore_unmapped_range(addr, addr + PAGE_SIZE,
149 vma, vec);
150 else if (pte_present(pte))
151 *vec = 1;
152 else {
153 swp_entry_t entry = pte_to_swp_entry(pte);
154
155 if (non_swap_entry(entry)) {
156
157
158
159
160 *vec = 1;
161 } else {
162#ifdef CONFIG_SWAP
163 *vec = mincore_page(swap_address_space(entry),
164 swp_offset(entry));
165#else
166 WARN_ON(1);
167 *vec = 1;
168#endif
169 }
170 }
171 vec++;
172 }
173 pte_unmap_unlock(ptep - 1, ptl);
174out:
175 walk->private += nr;
176 cond_resched();
177 return 0;
178}
179
180static inline bool can_do_mincore(struct vm_area_struct *vma)
181{
182 if (vma_is_anonymous(vma))
183 return true;
184 if (!vma->vm_file)
185 return false;
186
187
188
189
190
191
192 return inode_owner_or_capable(file_inode(vma->vm_file)) ||
193 inode_permission(file_inode(vma->vm_file), MAY_WRITE) == 0;
194}
195
196
197
198
199
200
201static long do_mincore(unsigned long addr, unsigned long pages, unsigned char *vec)
202{
203 struct vm_area_struct *vma;
204 unsigned long end;
205 int err;
206 struct mm_walk mincore_walk = {
207 .pmd_entry = mincore_pte_range,
208 .pte_hole = mincore_unmapped_range,
209 .hugetlb_entry = mincore_hugetlb,
210 .private = vec,
211 };
212
213 vma = find_vma(current->mm, addr);
214 if (!vma || addr < vma->vm_start)
215 return -ENOMEM;
216 end = min(vma->vm_end, addr + (pages << PAGE_SHIFT));
217 if (!can_do_mincore(vma)) {
218 unsigned long pages = DIV_ROUND_UP(end - addr, PAGE_SIZE);
219 memset(vec, 1, pages);
220 return pages;
221 }
222 mincore_walk.mm = vma->vm_mm;
223 err = walk_page_range(addr, end, &mincore_walk);
224 if (err < 0)
225 return err;
226 return (end - addr) >> PAGE_SHIFT;
227}
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253SYSCALL_DEFINE3(mincore, unsigned long, start, size_t, len,
254 unsigned char __user *, vec)
255{
256 long retval;
257 unsigned long pages;
258 unsigned char *tmp;
259
260
261 if (start & ~PAGE_MASK)
262 return -EINVAL;
263
264
265 if (!access_ok((void __user *) start, len))
266 return -ENOMEM;
267
268
269 pages = len >> PAGE_SHIFT;
270 pages += (offset_in_page(len)) != 0;
271
272 if (!access_ok(vec, pages))
273 return -EFAULT;
274
275 tmp = (void *) __get_free_page(GFP_USER);
276 if (!tmp)
277 return -EAGAIN;
278
279 retval = 0;
280 while (pages) {
281
282
283
284
285 down_read(¤t->mm->mmap_sem);
286 retval = do_mincore(start, min(pages, PAGE_SIZE), tmp);
287 up_read(¤t->mm->mmap_sem);
288
289 if (retval <= 0)
290 break;
291 if (copy_to_user(vec, tmp, retval)) {
292 retval = -EFAULT;
293 break;
294 }
295 pages -= retval;
296 vec += retval;
297 start += retval << PAGE_SHIFT;
298 retval = 0;
299 }
300 free_page((unsigned long) tmp);
301 return retval;
302}
303