1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * linux/mm/mincore.c 4 * 5 * Copyright (C) 1994-2006 Linus Torvalds 6 */ 7 8 /* 9 * The mincore() system call. 10 */ 11 #include <linux/pagemap.h> 12 #include <linux/gfp.h> 13 #include <linux/pagewalk.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 24 static 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 * Hugepages under user process are always in RAM and never 33 * swapped out, but theoretically it needs to be checked. 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 * Later we can get more picky about what "in core" means precisely. 47 * For now, simply check to see if the page is in the page cache, 48 * and is up to date; i.e. that no page-in operation would be required 49 * at this time if an application were to map and access this page. 50 */ 51 static unsigned char mincore_page(struct address_space *mapping, pgoff_t pgoff) 52 { 53 unsigned char present = 0; 54 struct page *page; 55 56 /* 57 * When tmpfs swaps out a page from a file, any process mapping that 58 * file will not get a swp_entry_t in its pte, but rather it is like 59 * any other file mapping (ie. marked !present and faulted in with 60 * tmpfs's .fault). So swapped out tmpfs mappings are tested here. 61 */ 62 #ifdef CONFIG_SWAP 63 if (shmem_mapping(mapping)) { 64 page = find_get_entry(mapping, pgoff); 65 /* 66 * shmem/tmpfs may return swap: account for swapcache 67 * page too. 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 /* Prevent swap device to being swapoff under us */ 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 95 static 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 } 113 114 static 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 } 121 122 static 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 { /* pte is a swap entry */ 153 swp_entry_t entry = pte_to_swp_entry(pte); 154 155 if (non_swap_entry(entry)) { 156 /* 157 * migration or hwpoison entries are always 158 * uptodate 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); 174 out: 175 walk->private += nr; 176 cond_resched(); 177 return 0; 178 } 179 180 static 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 * Reveal pagecache information only for non-anonymous mappings that 188 * correspond to the files the calling process could (if tried) open 189 * for writing; otherwise we'd be including shared non-exclusive 190 * mappings, which opens a side channel. 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 static const struct mm_walk_ops mincore_walk_ops = { 197 .pmd_entry = mincore_pte_range, 198 .pte_hole = mincore_unmapped_range, 199 .hugetlb_entry = mincore_hugetlb, 200 }; 201 202 /* 203 * Do a chunk of "sys_mincore()". We've already checked 204 * all the arguments, we hold the mmap semaphore: we should 205 * just return the amount of info we're asked for. 206 */ 207 static long do_mincore(unsigned long addr, unsigned long pages, unsigned char *vec) 208 { 209 struct vm_area_struct *vma; 210 unsigned long end; 211 int err; 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 err = walk_page_range(vma->vm_mm, addr, end, &mincore_walk_ops, vec); 223 if (err < 0) 224 return err; 225 return (end - addr) >> PAGE_SHIFT; 226 } 227 228 /* 229 * The mincore(2) system call. 230 * 231 * mincore() returns the memory residency status of the pages in the 232 * current process's address space specified by [addr, addr + len). 233 * The status is returned in a vector of bytes. The least significant 234 * bit of each byte is 1 if the referenced page is in memory, otherwise 235 * it is zero. 236 * 237 * Because the status of a page can change after mincore() checks it 238 * but before it returns to the application, the returned vector may 239 * contain stale information. Only locked pages are guaranteed to 240 * remain in memory. 241 * 242 * return values: 243 * zero - success 244 * -EFAULT - vec points to an illegal address 245 * -EINVAL - addr is not a multiple of PAGE_SIZE 246 * -ENOMEM - Addresses in the range [addr, addr + len] are 247 * invalid for the address space of this process, or 248 * specify one or more pages which are not currently 249 * mapped 250 * -EAGAIN - A kernel resource was temporarily unavailable. 251 */ 252 SYSCALL_DEFINE3(mincore, unsigned long, start, size_t, len, 253 unsigned char __user *, vec) 254 { 255 long retval; 256 unsigned long pages; 257 unsigned char *tmp; 258 259 start = untagged_addr(start); 260 261 /* Check the start address: needs to be page-aligned.. */ 262 if (start & ~PAGE_MASK) 263 return -EINVAL; 264 265 /* ..and we need to be passed a valid user-space range */ 266 if (!access_ok((void __user *) start, len)) 267 return -ENOMEM; 268 269 /* This also avoids any overflows on PAGE_ALIGN */ 270 pages = len >> PAGE_SHIFT; 271 pages += (offset_in_page(len)) != 0; 272 273 if (!access_ok(vec, pages)) 274 return -EFAULT; 275 276 tmp = (void *) __get_free_page(GFP_USER); 277 if (!tmp) 278 return -EAGAIN; 279 280 retval = 0; 281 while (pages) { 282 /* 283 * Do at most PAGE_SIZE entries per iteration, due to 284 * the temporary buffer size. 285 */ 286 down_read(¤t->mm->mmap_sem); 287 retval = do_mincore(start, min(pages, PAGE_SIZE), tmp); 288 up_read(¤t->mm->mmap_sem); 289 290 if (retval <= 0) 291 break; 292 if (copy_to_user(vec, tmp, retval)) { 293 retval = -EFAULT; 294 break; 295 } 296 pages -= retval; 297 vec += retval; 298 start += retval << PAGE_SHIFT; 299 retval = 0; 300 } 301 free_page((unsigned long) tmp); 302 return retval; 303 } 304