4 * Copyright (C) 1994-2006 Linus Torvalds
8 * The mincore() system call.
10 #include <linux/slab.h>
11 #include <linux/pagemap.h>
13 #include <linux/mman.h>
14 #include <linux/syscalls.h>
15 #include <linux/swap.h>
16 #include <linux/swapops.h>
18 #include <asm/uaccess.h>
19 #include <asm/pgtable.h>
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.
27 static unsigned char mincore_page(struct address_space *mapping, pgoff_t pgoff)
29 unsigned char present = 0;
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 .fault). So swapped out tmpfs mappings are tested here.
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.
42 page = find_get_page(mapping, pgoff);
44 present = PageUptodate(page);
45 page_cache_release(page);
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.
56 static long do_mincore(unsigned long addr, unsigned char *vec, unsigned long pages)
66 struct vm_area_struct *vma = find_vma(current->mm, addr);
69 * find_vma() didn't find anything above us, or we're
70 * in an unmapped hole in the address space: ENOMEM.
72 if (!vma || addr < vma->vm_start)
76 * Calculate how many pages there are left in the last level of the
77 * PTE array for our address.
79 nr = PTRS_PER_PTE - ((addr >> PAGE_SHIFT) & (PTRS_PER_PTE-1));
82 * Don't overrun this vma
84 nr = min(nr, (vma->vm_end - addr) >> PAGE_SHIFT);
87 * Don't return more than the caller asked for
91 pgd = pgd_offset(vma->vm_mm, addr);
92 if (pgd_none_or_clear_bad(pgd))
94 pud = pud_offset(pgd, addr);
95 if (pud_none_or_clear_bad(pud))
97 pmd = pmd_offset(pud, addr);
98 if (pmd_none_or_clear_bad(pmd))
101 ptep = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
102 for (i = 0; i < nr; i++, ptep++, addr += PAGE_SIZE) {
103 unsigned char present;
106 if (pte_present(pte)) {
109 } else if (pte_none(pte)) {
111 pgoff = linear_page_index(vma, addr);
112 present = mincore_page(vma->vm_file->f_mapping,
117 } else if (pte_file(pte)) {
118 pgoff = pte_to_pgoff(pte);
119 present = mincore_page(vma->vm_file->f_mapping, pgoff);
121 } else { /* pte is a swap entry */
122 swp_entry_t entry = pte_to_swp_entry(pte);
123 if (is_migration_entry(entry)) {
124 /* migration entries are always uptodate */
129 present = mincore_page(&swapper_space, pgoff);
139 pte_unmap_unlock(ptep-1, ptl);
145 pgoff = linear_page_index(vma, addr);
146 for (i = 0; i < nr; i++, pgoff++)
147 vec[i] = mincore_page(vma->vm_file->f_mapping, pgoff);
149 for (i = 0; i < nr; i++)
157 * The mincore(2) system call.
159 * mincore() returns the memory residency status of the pages in the
160 * current process's address space specified by [addr, addr + len).
161 * The status is returned in a vector of bytes. The least significant
162 * bit of each byte is 1 if the referenced page is in memory, otherwise
165 * Because the status of a page can change after mincore() checks it
166 * but before it returns to the application, the returned vector may
167 * contain stale information. Only locked pages are guaranteed to
172 * -EFAULT - vec points to an illegal address
173 * -EINVAL - addr is not a multiple of PAGE_CACHE_SIZE
174 * -ENOMEM - Addresses in the range [addr, addr + len] are
175 * invalid for the address space of this process, or
176 * specify one or more pages which are not currently
178 * -EAGAIN - A kernel resource was temporarily unavailable.
180 SYSCALL_DEFINE3(mincore, unsigned long, start, size_t, len,
181 unsigned char __user *, vec)
187 /* Check the start address: needs to be page-aligned.. */
188 if (start & ~PAGE_CACHE_MASK)
191 /* ..and we need to be passed a valid user-space range */
192 if (!access_ok(VERIFY_READ, (void __user *) start, len))
195 /* This also avoids any overflows on PAGE_CACHE_ALIGN */
196 pages = len >> PAGE_SHIFT;
197 pages += (len & ~PAGE_MASK) != 0;
199 if (!access_ok(VERIFY_WRITE, vec, pages))
202 tmp = (void *) __get_free_page(GFP_USER);
209 * Do at most PAGE_SIZE entries per iteration, due to
210 * the temporary buffer size.
212 down_read(¤t->mm->mmap_sem);
213 retval = do_mincore(start, tmp, min(pages, PAGE_SIZE));
214 up_read(¤t->mm->mmap_sem);
218 if (copy_to_user(vec, tmp, retval)) {
224 start += retval << PAGE_SHIFT;
227 free_page((unsigned long) tmp);