fadvise.c 4.38 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5
/*
 * mm/fadvise.c
 *
 * Copyright (C) 2002, Linus Torvalds
 *
6
 * 11Jan2003	Andrew Morton
Linus Torvalds's avatar
Linus Torvalds committed
7 8 9 10 11 12 13 14 15 16 17
 *		Initial version.
 */

#include <linux/kernel.h>
#include <linux/file.h>
#include <linux/fs.h>
#include <linux/mm.h>
#include <linux/pagemap.h>
#include <linux/backing-dev.h>
#include <linux/pagevec.h>
#include <linux/fadvise.h>
18
#include <linux/writeback.h>
Linus Torvalds's avatar
Linus Torvalds committed
19
#include <linux/syscalls.h>
20
#include <linux/swap.h>
Linus Torvalds's avatar
Linus Torvalds committed
21 22 23 24 25 26 27

#include <asm/unistd.h>

/*
 * POSIX_FADV_WILLNEED could set PG_Referenced, and POSIX_FADV_NOREUSE could
 * deactivate the pages and clear PG_Referenced.
 */
28
SYSCALL_DEFINE4(fadvise64_64, int, fd, loff_t, offset, loff_t, len, int, advice)
Linus Torvalds's avatar
Linus Torvalds committed
29
{
30
	struct fd f = fdget(fd);
Matthew Wilcox's avatar
Matthew Wilcox committed
31
	struct inode *inode;
Linus Torvalds's avatar
Linus Torvalds committed
32 33
	struct address_space *mapping;
	struct backing_dev_info *bdi;
34
	loff_t endbyte;			/* inclusive */
Linus Torvalds's avatar
Linus Torvalds committed
35 36 37 38 39
	pgoff_t start_index;
	pgoff_t end_index;
	unsigned long nrpages;
	int ret = 0;

40
	if (!f.file)
Linus Torvalds's avatar
Linus Torvalds committed
41 42
		return -EBADF;

Matthew Wilcox's avatar
Matthew Wilcox committed
43 44
	inode = file_inode(f.file);
	if (S_ISFIFO(inode->i_mode)) {
45 46 47 48
		ret = -ESPIPE;
		goto out;
	}

49
	mapping = f.file->f_mapping;
Linus Torvalds's avatar
Linus Torvalds committed
50 51 52 53 54
	if (!mapping || len < 0) {
		ret = -EINVAL;
		goto out;
	}

55 56 57
	bdi = inode_to_bdi(mapping->host);

	if (IS_DAX(inode) || (bdi == &noop_backing_dev_info)) {
58 59 60 61 62 63 64 65 66 67 68 69
		switch (advice) {
		case POSIX_FADV_NORMAL:
		case POSIX_FADV_RANDOM:
		case POSIX_FADV_SEQUENTIAL:
		case POSIX_FADV_WILLNEED:
		case POSIX_FADV_NOREUSE:
		case POSIX_FADV_DONTNEED:
			/* no bad return value, but ignore advice */
			break;
		default:
			ret = -EINVAL;
		}
70
		goto out;
71
	}
72

Linus Torvalds's avatar
Linus Torvalds committed
73 74 75 76
	/* Careful about overflows. Len == 0 means "as much as possible" */
	endbyte = offset + len;
	if (!len || endbyte < len)
		endbyte = -1;
77 78
	else
		endbyte--;		/* inclusive */
Linus Torvalds's avatar
Linus Torvalds committed
79 80 81

	switch (advice) {
	case POSIX_FADV_NORMAL:
82 83 84 85
		f.file->f_ra.ra_pages = bdi->ra_pages;
		spin_lock(&f.file->f_lock);
		f.file->f_mode &= ~FMODE_RANDOM;
		spin_unlock(&f.file->f_lock);
Linus Torvalds's avatar
Linus Torvalds committed
86 87
		break;
	case POSIX_FADV_RANDOM:
88 89 90
		spin_lock(&f.file->f_lock);
		f.file->f_mode |= FMODE_RANDOM;
		spin_unlock(&f.file->f_lock);
Linus Torvalds's avatar
Linus Torvalds committed
91 92
		break;
	case POSIX_FADV_SEQUENTIAL:
93 94 95 96
		f.file->f_ra.ra_pages = bdi->ra_pages * 2;
		spin_lock(&f.file->f_lock);
		f.file->f_mode &= ~FMODE_RANDOM;
		spin_unlock(&f.file->f_lock);
Linus Torvalds's avatar
Linus Torvalds committed
97 98 99
		break;
	case POSIX_FADV_WILLNEED:
		/* First and last PARTIAL page! */
100 101
		start_index = offset >> PAGE_SHIFT;
		end_index = endbyte >> PAGE_SHIFT;
Linus Torvalds's avatar
Linus Torvalds committed
102 103 104 105 106

		/* Careful about overflow on the "+1" */
		nrpages = end_index - start_index + 1;
		if (!nrpages)
			nrpages = ~0UL;
107 108 109 110 111

		/*
		 * Ignore return value because fadvise() shall return
		 * success even if filesystem can't retrieve a hint,
		 */
112
		force_page_cache_readahead(mapping, f.file, start_index,
113
					   nrpages);
Linus Torvalds's avatar
Linus Torvalds committed
114
		break;
115 116
	case POSIX_FADV_NOREUSE:
		break;
Linus Torvalds's avatar
Linus Torvalds committed
117
	case POSIX_FADV_DONTNEED:
118
		if (!inode_write_congested(mapping->host))
119 120
			__filemap_fdatawrite_range(mapping, offset, endbyte,
						   WB_SYNC_NONE);
Linus Torvalds's avatar
Linus Torvalds committed
121

122 123 124 125 126
		/*
		 * First and last FULL page! Partial pages are deliberately
		 * preserved on the expectation that it is better to preserve
		 * needed memory than to discard unneeded memory.
		 */
127 128
		start_index = (offset+(PAGE_SIZE-1)) >> PAGE_SHIFT;
		end_index = (endbyte >> PAGE_SHIFT);
129 130 131 132 133 134 135 136 137 138 139
		if ((endbyte & ~PAGE_MASK) != ~PAGE_MASK) {
			/* First page is tricky as 0 - 1 = -1, but pgoff_t
			 * is unsigned, so the end_index >= start_index
			 * check below would be true and we'll discard the whole
			 * file cache which is not what was asked.
			 */
			if (end_index == 0)
				break;

			end_index--;
		}
Linus Torvalds's avatar
Linus Torvalds committed
140

141
		if (end_index >= start_index) {
142 143 144 145 146 147 148 149 150 151 152 153 154 155
			unsigned long count;

			/*
			 * It's common to FADV_DONTNEED right after
			 * the read or write that instantiates the
			 * pages, in which case there will be some
			 * sitting on the local LRU cache. Try to
			 * avoid the expensive remote drain and the
			 * second cache tree walk below by flushing
			 * them out right away.
			 */
			lru_add_drain();

			count = invalidate_mapping_pages(mapping,
156 157 158 159 160 161 162 163 164 165 166
						start_index, end_index);

			/*
			 * If fewer pages were invalidated than expected then
			 * it is possible that some of the pages were on
			 * a per-cpu pagevec for a remote CPU. Drain all
			 * pagevecs and try again.
			 */
			if (count < (end_index - start_index + 1)) {
				lru_add_drain_all();
				invalidate_mapping_pages(mapping, start_index,
167
						end_index);
168 169
			}
		}
170
		break;
Linus Torvalds's avatar
Linus Torvalds committed
171 172 173 174
	default:
		ret = -EINVAL;
	}
out:
175
	fdput(f);
Linus Torvalds's avatar
Linus Torvalds committed
176 177 178 179 180
	return ret;
}

#ifdef __ARCH_WANT_SYS_FADVISE64

181
SYSCALL_DEFINE4(fadvise64, int, fd, loff_t, offset, size_t, len, int, advice)
Linus Torvalds's avatar
Linus Torvalds committed
182 183 184 185 186
{
	return sys_fadvise64_64(fd, offset, len, advice);
}

#endif