]>
git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blob - mm/mincore.c
4 * Copyright (C) 1994-2006 Linus Torvalds
8 * The mincore() system call.
10 #include <linux/pagemap.h>
11 #include <linux/gfp.h>
13 #include <linux/mman.h>
14 #include <linux/syscalls.h>
15 #include <linux/swap.h>
16 #include <linux/swapops.h>
17 #include <linux/hugetlb.h>
19 #include <asm/uaccess.h>
20 #include <asm/pgtable.h>
22 static int mincore_hugetlb(pte_t
*pte
, unsigned long hmask
, unsigned long addr
,
23 unsigned long end
, struct mm_walk
*walk
)
25 #ifdef CONFIG_HUGETLB_PAGE
26 unsigned char present
;
27 unsigned char *vec
= walk
->private;
30 * Hugepages under user process are always in RAM and never
31 * swapped out, but theoretically it needs to be checked.
33 present
= pte
&& !huge_pte_none(huge_ptep_get(pte
));
34 for (; addr
!= end
; vec
++, addr
+= PAGE_SIZE
)
44 * Later we can get more picky about what "in core" means precisely.
45 * For now, simply check to see if the page is in the page cache,
46 * and is up to date; i.e. that no page-in operation would be required
47 * at this time if an application were to map and access this page.
49 static unsigned char mincore_page(struct address_space
*mapping
, pgoff_t pgoff
)
51 unsigned char present
= 0;
55 * When tmpfs swaps out a page from a file, any process mapping that
56 * file will not get a swp_entry_t in its pte, but rather it is like
57 * any other file mapping (ie. marked !present and faulted in with
58 * tmpfs's .fault). So swapped out tmpfs mappings are tested here.
61 if (shmem_mapping(mapping
)) {
62 page
= find_get_entry(mapping
, pgoff
);
64 * shmem/tmpfs may return swap: account for swapcache
67 if (radix_tree_exceptional_entry(page
)) {
68 swp_entry_t swp
= radix_to_swp_entry(page
);
69 page
= find_get_page(swap_address_space(swp
), swp
.val
);
72 page
= find_get_page(mapping
, pgoff
);
74 page
= find_get_page(mapping
, pgoff
);
77 present
= PageUptodate(page
);
78 page_cache_release(page
);
84 static int __mincore_unmapped_range(unsigned long addr
, unsigned long end
,
85 struct vm_area_struct
*vma
, unsigned char *vec
)
87 unsigned long nr
= (end
- addr
) >> PAGE_SHIFT
;
93 pgoff
= linear_page_index(vma
, addr
);
94 for (i
= 0; i
< nr
; i
++, pgoff
++)
95 vec
[i
] = mincore_page(vma
->vm_file
->f_mapping
, pgoff
);
97 for (i
= 0; i
< nr
; i
++)
103 static int mincore_unmapped_range(unsigned long addr
, unsigned long end
,
104 struct mm_walk
*walk
)
106 walk
->private += __mincore_unmapped_range(addr
, end
,
107 walk
->vma
, walk
->private);
111 static int mincore_pte_range(pmd_t
*pmd
, unsigned long addr
, unsigned long end
,
112 struct mm_walk
*walk
)
115 struct vm_area_struct
*vma
= walk
->vma
;
117 unsigned char *vec
= walk
->private;
118 int nr
= (end
- addr
) >> PAGE_SHIFT
;
120 if (pmd_trans_huge_lock(pmd
, vma
, &ptl
) == 1) {
126 if (pmd_trans_unstable(pmd
)) {
127 __mincore_unmapped_range(addr
, end
, vma
, vec
);
131 ptep
= pte_offset_map_lock(walk
->mm
, pmd
, addr
, &ptl
);
132 for (; addr
!= end
; ptep
++, addr
+= PAGE_SIZE
) {
136 __mincore_unmapped_range(addr
, addr
+ PAGE_SIZE
,
138 else if (pte_present(pte
))
140 else { /* pte is a swap entry */
141 swp_entry_t entry
= pte_to_swp_entry(pte
);
143 if (non_swap_entry(entry
)) {
145 * migration or hwpoison entries are always
151 *vec
= mincore_page(swap_address_space(entry
),
161 pte_unmap_unlock(ptep
- 1, ptl
);
169 * Do a chunk of "sys_mincore()". We've already checked
170 * all the arguments, we hold the mmap semaphore: we should
171 * just return the amount of info we're asked for.
173 static long do_mincore(unsigned long addr
, unsigned long pages
, unsigned char *vec
)
175 struct vm_area_struct
*vma
;
178 struct mm_walk mincore_walk
= {
179 .pmd_entry
= mincore_pte_range
,
180 .pte_hole
= mincore_unmapped_range
,
181 .hugetlb_entry
= mincore_hugetlb
,
185 vma
= find_vma(current
->mm
, addr
);
186 if (!vma
|| addr
< vma
->vm_start
)
188 mincore_walk
.mm
= vma
->vm_mm
;
189 end
= min(vma
->vm_end
, addr
+ (pages
<< PAGE_SHIFT
));
190 err
= walk_page_range(addr
, end
, &mincore_walk
);
193 return (end
- addr
) >> PAGE_SHIFT
;
197 * The mincore(2) system call.
199 * mincore() returns the memory residency status of the pages in the
200 * current process's address space specified by [addr, addr + len).
201 * The status is returned in a vector of bytes. The least significant
202 * bit of each byte is 1 if the referenced page is in memory, otherwise
205 * Because the status of a page can change after mincore() checks it
206 * but before it returns to the application, the returned vector may
207 * contain stale information. Only locked pages are guaranteed to
212 * -EFAULT - vec points to an illegal address
213 * -EINVAL - addr is not a multiple of PAGE_CACHE_SIZE
214 * -ENOMEM - Addresses in the range [addr, addr + len] are
215 * invalid for the address space of this process, or
216 * specify one or more pages which are not currently
218 * -EAGAIN - A kernel resource was temporarily unavailable.
220 SYSCALL_DEFINE3(mincore
, unsigned long, start
, size_t, len
,
221 unsigned char __user
*, vec
)
227 /* Check the start address: needs to be page-aligned.. */
228 if (start
& ~PAGE_CACHE_MASK
)
231 /* ..and we need to be passed a valid user-space range */
232 if (!access_ok(VERIFY_READ
, (void __user
*) start
, len
))
235 /* This also avoids any overflows on PAGE_CACHE_ALIGN */
236 pages
= len
>> PAGE_SHIFT
;
237 pages
+= (len
& ~PAGE_MASK
) != 0;
239 if (!access_ok(VERIFY_WRITE
, vec
, pages
))
242 tmp
= (void *) __get_free_page(GFP_USER
);
249 * Do at most PAGE_SIZE entries per iteration, due to
250 * the temporary buffer size.
252 down_read(¤t
->mm
->mmap_sem
);
253 retval
= do_mincore(start
, min(pages
, PAGE_SIZE
), tmp
);
254 up_read(¤t
->mm
->mmap_sem
);
258 if (copy_to_user(vec
, tmp
, retval
)) {
264 start
+= retval
<< PAGE_SHIFT
;
267 free_page((unsigned long) tmp
);