1// SPDX-License-Identifier: GPL-2.0
2/*
3 * mm/fadvise.c
4 *
5 * Copyright (C) 2002, Linus Torvalds
6 *
7 * 11Jan2003 Andrew Morton
8 * Initial version.
9 */
10
11#include <linux/kernel.h>
12#include <linux/file.h>
13#include <linux/fs.h>
14#include <linux/mm.h>
15#include <linux/pagemap.h>
16#include <linux/backing-dev.h>
17#include <linux/pagevec.h>
18#include <linux/fadvise.h>
19#include <linux/writeback.h>
20#include <linux/syscalls.h>
21#include <linux/swap.h>
22
23#include <asm/unistd.h>
24
25/*
26 * POSIX_FADV_WILLNEED could set PG_Referenced, and POSIX_FADV_NOREUSE could
27 * deactivate the pages and clear PG_Referenced.
28 */
29
30static int generic_fadvise(struct file *file, loff_t offset, loff_t len,
31 int advice)
32{
33 struct inode *inode;
34 struct address_space *mapping;
35 struct backing_dev_info *bdi;
36 loff_t endbyte; /* inclusive */
37 pgoff_t start_index;
38 pgoff_t end_index;
39 unsigned long nrpages;
40
41 inode = file_inode(file);
42 if (S_ISFIFO(inode->i_mode))
43 return -ESPIPE;
44
45 mapping = file->f_mapping;
46 if (!mapping || len < 0)
47 return -EINVAL;
48
49 bdi = inode_to_bdi(mapping->host);
50
51 if (IS_DAX(inode) || (bdi == &noop_backing_dev_info)) {
52 switch (advice) {
53 case POSIX_FADV_NORMAL:
54 case POSIX_FADV_RANDOM:
55 case POSIX_FADV_SEQUENTIAL:
56 case POSIX_FADV_WILLNEED:
57 case POSIX_FADV_NOREUSE:
58 case POSIX_FADV_DONTNEED:
59 /* no bad return value, but ignore advice */
60 break;
61 default:
62 return -EINVAL;
63 }
64 return 0;
65 }
66
67 /*
68 * Careful about overflows. Len == 0 means "as much as possible". Use
69 * unsigned math because signed overflows are undefined and UBSan
70 * complains.
71 */
72 endbyte = (u64)offset + (u64)len;
73 if (!len || endbyte < len)
74 endbyte = -1;
75 else
76 endbyte--; /* inclusive */
77
78 switch (advice) {
79 case POSIX_FADV_NORMAL:
80 file->f_ra.ra_pages = bdi->ra_pages;
81 spin_lock(&file->f_lock);
82 file->f_mode &= ~FMODE_RANDOM;
83 spin_unlock(&file->f_lock);
84 break;
85 case POSIX_FADV_RANDOM:
86 spin_lock(&file->f_lock);
87 file->f_mode |= FMODE_RANDOM;
88 spin_unlock(&file->f_lock);
89 break;
90 case POSIX_FADV_SEQUENTIAL:
91 file->f_ra.ra_pages = bdi->ra_pages * 2;
92 spin_lock(&file->f_lock);
93 file->f_mode &= ~FMODE_RANDOM;
94 spin_unlock(&file->f_lock);
95 break;
96 case POSIX_FADV_WILLNEED:
97 /* First and last PARTIAL page! */
98 start_index = offset >> PAGE_SHIFT;
99 end_index = endbyte >> PAGE_SHIFT;
100
101 /* Careful about overflow on the "+1" */
102 nrpages = end_index - start_index + 1;
103 if (!nrpages)
104 nrpages = ~0UL;
105
106 /*
107 * Ignore return value because fadvise() shall return
108 * success even if filesystem can't retrieve a hint,
109 */
110 force_page_cache_readahead(mapping, file, start_index, nrpages);
111 break;
112 case POSIX_FADV_NOREUSE:
113 break;
114 case POSIX_FADV_DONTNEED:
115 if (!inode_write_congested(mapping->host))
116 __filemap_fdatawrite_range(mapping, offset, endbyte,
117 WB_SYNC_NONE);
118
119 /*
120 * First and last FULL page! Partial pages are deliberately
121 * preserved on the expectation that it is better to preserve
122 * needed memory than to discard unneeded memory.
123 */
124 start_index = (offset+(PAGE_SIZE-1)) >> PAGE_SHIFT;
125 end_index = (endbyte >> PAGE_SHIFT);
126 /*
127 * The page at end_index will be inclusively discarded according
128 * by invalidate_mapping_pages(), so subtracting 1 from
129 * end_index means we will skip the last page. But if endbyte
130 * is page aligned or is at the end of file, we should not skip
131 * that page - discarding the last page is safe enough.
132 */
133 if ((endbyte & ~PAGE_MASK) != ~PAGE_MASK &&
134 endbyte != inode->i_size - 1) {
135 /* First page is tricky as 0 - 1 = -1, but pgoff_t
136 * is unsigned, so the end_index >= start_index
137 * check below would be true and we'll discard the whole
138 * file cache which is not what was asked.
139 */
140 if (end_index == 0)
141 break;
142
143 end_index--;
144 }
145
146 if (end_index >= start_index) {
147 unsigned long count;
148
149 /*
150 * It's common to FADV_DONTNEED right after
151 * the read or write that instantiates the
152 * pages, in which case there will be some
153 * sitting on the local LRU cache. Try to
154 * avoid the expensive remote drain and the
155 * second cache tree walk below by flushing
156 * them out right away.
157 */
158 lru_add_drain();
159
160 count = invalidate_mapping_pages(mapping,
161 start_index, end_index);
162
163 /*
164 * If fewer pages were invalidated than expected then
165 * it is possible that some of the pages were on
166 * a per-cpu pagevec for a remote CPU. Drain all
167 * pagevecs and try again.
168 */
169 if (count < (end_index - start_index + 1)) {
170 lru_add_drain_all();
171 invalidate_mapping_pages(mapping, start_index,
172 end_index);
173 }
174 }
175 break;
176 default:
177 return -EINVAL;
178 }
179 return 0;
180}
181
182int vfs_fadvise(struct file *file, loff_t offset, loff_t len, int advice)
183{
184 if (file->f_op->fadvise)
185 return file->f_op->fadvise(file, offset, len, advice);
186
187 return generic_fadvise(file, offset, len, advice);
188}
189EXPORT_SYMBOL(vfs_fadvise);
190
191#ifdef CONFIG_ADVISE_SYSCALLS
192
193int ksys_fadvise64_64(int fd, loff_t offset, loff_t len, int advice)
194{
195 struct fd f = fdget(fd);
196 int ret;
197
198 if (!f.file)
199 return -EBADF;
200
201 ret = vfs_fadvise(f.file, offset, len, advice);
202
203 fdput(f);
204 return ret;
205}
206
207SYSCALL_DEFINE4(fadvise64_64, int, fd, loff_t, offset, loff_t, len, int, advice)
208{
209 return ksys_fadvise64_64(fd, offset, len, advice);
210}
211
212#ifdef __ARCH_WANT_SYS_FADVISE64
213
214SYSCALL_DEFINE4(fadvise64, int, fd, loff_t, offset, size_t, len, int, advice)
215{
216 return ksys_fadvise64_64(fd, offset, len, advice);
217}
218
219#endif
220#endif
221