file_table.c 6.8 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 *  linux/fs/file_table.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *  Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
 */

#include <linux/string.h>
#include <linux/slab.h>
#include <linux/file.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/fs.h>
#include <linux/security.h>
#include <linux/eventpoll.h>
16
#include <linux/rcupdate.h>
L
Linus Torvalds 已提交
17
#include <linux/mount.h>
18
#include <linux/capability.h>
L
Linus Torvalds 已提交
19
#include <linux/cdev.h>
R
Robert Love 已提交
20
#include <linux/fsnotify.h>
D
Dipankar Sarma 已提交
21 22 23 24
#include <linux/sysctl.h>
#include <linux/percpu_counter.h>

#include <asm/atomic.h>
L
Linus Torvalds 已提交
25 26 27 28 29 30 31

/* sysctl tunables... */
struct files_stat_struct files_stat = {
	.max_files = NR_FILE
};

/* public. Not pretty! */
D
Dipankar Sarma 已提交
32
__cacheline_aligned_in_smp DEFINE_SPINLOCK(files_lock);
L
Linus Torvalds 已提交
33

D
Dipankar Sarma 已提交
34
static struct percpu_counter nr_files __cacheline_aligned_in_smp;
L
Linus Torvalds 已提交
35

D
Dipankar Sarma 已提交
36
static inline void file_free_rcu(struct rcu_head *head)
L
Linus Torvalds 已提交
37
{
D
Dipankar Sarma 已提交
38 39
	struct file *f =  container_of(head, struct file, f_u.fu_rcuhead);
	kmem_cache_free(filp_cachep, f);
L
Linus Torvalds 已提交
40 41
}

D
Dipankar Sarma 已提交
42
static inline void file_free(struct file *f)
L
Linus Torvalds 已提交
43
{
D
Dipankar Sarma 已提交
44 45
	percpu_counter_dec(&nr_files);
	call_rcu(&f->f_u.fu_rcuhead, file_free_rcu);
L
Linus Torvalds 已提交
46 47
}

D
Dipankar Sarma 已提交
48 49 50 51
/*
 * Return the total number of open files in the system
 */
static int get_nr_files(void)
L
Linus Torvalds 已提交
52
{
D
Dipankar Sarma 已提交
53
	return percpu_counter_read_positive(&nr_files);
L
Linus Torvalds 已提交
54 55
}

D
Dipankar Sarma 已提交
56 57 58 59
/*
 * Return the maximum number of open files in the system
 */
int get_max_files(void)
60
{
D
Dipankar Sarma 已提交
61
	return files_stat.max_files;
62
}
D
Dipankar Sarma 已提交
63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81
EXPORT_SYMBOL_GPL(get_max_files);

/*
 * Handle nr_files sysctl
 */
#if defined(CONFIG_SYSCTL) && defined(CONFIG_PROC_FS)
int proc_nr_files(ctl_table *table, int write, struct file *filp,
                     void __user *buffer, size_t *lenp, loff_t *ppos)
{
	files_stat.nr_files = get_nr_files();
	return proc_dointvec(table, write, filp, buffer, lenp, ppos);
}
#else
int proc_nr_files(ctl_table *table, int write, struct file *filp,
                     void __user *buffer, size_t *lenp, loff_t *ppos)
{
	return -ENOSYS;
}
#endif
82

L
Linus Torvalds 已提交
83 84 85 86 87 88
/* Find an unused file structure and return a pointer to it.
 * Returns NULL, if there are no more free file structures or
 * we run out of memory.
 */
struct file *get_empty_filp(void)
{
89
	struct task_struct *tsk;
90
	static int old_max;
L
Linus Torvalds 已提交
91 92 93 94 95
	struct file * f;

	/*
	 * Privileged users can go above max_files
	 */
D
Dipankar Sarma 已提交
96 97 98 99 100 101 102 103
	if (get_nr_files() >= files_stat.max_files && !capable(CAP_SYS_ADMIN)) {
		/*
		 * percpu_counters are inaccurate.  Do an expensive check before
		 * we go and fail.
		 */
		if (percpu_counter_sum(&nr_files) >= files_stat.max_files)
			goto over;
	}
104 105 106 107 108

	f = kmem_cache_alloc(filp_cachep, GFP_KERNEL);
	if (f == NULL)
		goto fail;

D
Dipankar Sarma 已提交
109
	percpu_counter_inc(&nr_files);
110 111 112
	memset(f, 0, sizeof(*f));
	if (security_file_alloc(f))
		goto fail_sec;
L
Linus Torvalds 已提交
113

114 115
	tsk = current;
	INIT_LIST_HEAD(&f->f_u.fu_list);
116 117
	atomic_set(&f->f_count, 1);
	rwlock_init(&f->f_owner.lock);
118 119 120
	f->f_uid = tsk->fsuid;
	f->f_gid = tsk->fsgid;
	eventpoll_init_file(f);
121 122 123 124
	/* f->f_version: 0 */
	return f;

over:
L
Linus Torvalds 已提交
125
	/* Ran out of filps - report that */
D
Dipankar Sarma 已提交
126
	if (get_nr_files() > old_max) {
L
Linus Torvalds 已提交
127
		printk(KERN_INFO "VFS: file-max limit %d reached\n",
D
Dipankar Sarma 已提交
128 129
					get_max_files());
		old_max = get_nr_files();
L
Linus Torvalds 已提交
130
	}
131 132 133 134
	goto fail;

fail_sec:
	file_free(f);
L
Linus Torvalds 已提交
135 136 137 138 139 140 141 142
fail:
	return NULL;
}

EXPORT_SYMBOL(get_empty_filp);

void fastcall fput(struct file *file)
{
143
	if (atomic_dec_and_test(&file->f_count))
L
Linus Torvalds 已提交
144 145 146 147 148 149 150 151 152 153
		__fput(file);
}

EXPORT_SYMBOL(fput);

/* __fput is called from task context when aio completion releases the last
 * last use of a struct file *.  Do not use otherwise.
 */
void fastcall __fput(struct file *file)
{
154 155
	struct dentry *dentry = file->f_path.dentry;
	struct vfsmount *mnt = file->f_path.mnt;
L
Linus Torvalds 已提交
156 157 158
	struct inode *inode = dentry->d_inode;

	might_sleep();
R
Robert Love 已提交
159 160

	fsnotify_close(file);
L
Linus Torvalds 已提交
161 162 163 164 165 166 167 168 169 170
	/*
	 * The function eventpoll_release() should be the first called
	 * in the file cleanup chain.
	 */
	eventpoll_release(file);
	locks_remove_flock(file);

	if (file->f_op && file->f_op->release)
		file->f_op->release(inode, file);
	security_file_free(file);
171
	if (unlikely(S_ISCHR(inode->i_mode) && inode->i_cdev != NULL))
L
Linus Torvalds 已提交
172 173 174 175
		cdev_put(inode->i_cdev);
	fops_put(file->f_op);
	if (file->f_mode & FMODE_WRITE)
		put_write_access(inode);
176
	put_pid(file->f_owner.pid);
L
Linus Torvalds 已提交
177
	file_kill(file);
178 179
	file->f_path.dentry = NULL;
	file->f_path.mnt = NULL;
L
Linus Torvalds 已提交
180 181 182 183 184 185 186 187 188 189
	file_free(file);
	dput(dentry);
	mntput(mnt);
}

struct file fastcall *fget(unsigned int fd)
{
	struct file *file;
	struct files_struct *files = current->files;

190
	rcu_read_lock();
L
Linus Torvalds 已提交
191
	file = fcheck_files(files, fd);
192
	if (file) {
193
		if (!atomic_inc_not_zero(&file->f_count)) {
194 195 196 197 198 199 200
			/* File object ref couldn't be taken */
			rcu_read_unlock();
			return NULL;
		}
	}
	rcu_read_unlock();

L
Linus Torvalds 已提交
201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221
	return file;
}

EXPORT_SYMBOL(fget);

/*
 * Lightweight file lookup - no refcnt increment if fd table isn't shared. 
 * You can use this only if it is guranteed that the current task already 
 * holds a refcnt to that file. That check has to be done at fget() only
 * and a flag is returned to be passed to the corresponding fput_light().
 * There must not be a cloning between an fget_light/fput_light pair.
 */
struct file fastcall *fget_light(unsigned int fd, int *fput_needed)
{
	struct file *file;
	struct files_struct *files = current->files;

	*fput_needed = 0;
	if (likely((atomic_read(&files->count) == 1))) {
		file = fcheck_files(files, fd);
	} else {
222
		rcu_read_lock();
L
Linus Torvalds 已提交
223 224
		file = fcheck_files(files, fd);
		if (file) {
225
			if (atomic_inc_not_zero(&file->f_count))
226 227 228 229
				*fput_needed = 1;
			else
				/* Didn't get the reference, someone's freed */
				file = NULL;
L
Linus Torvalds 已提交
230
		}
231
		rcu_read_unlock();
L
Linus Torvalds 已提交
232
	}
233

L
Linus Torvalds 已提交
234 235 236 237 238 239
	return file;
}


void put_filp(struct file *file)
{
240
	if (atomic_dec_and_test(&file->f_count)) {
L
Linus Torvalds 已提交
241 242 243 244 245 246 247 248 249 250 251
		security_file_free(file);
		file_kill(file);
		file_free(file);
	}
}

void file_move(struct file *file, struct list_head *list)
{
	if (!list)
		return;
	file_list_lock();
E
Eric Dumazet 已提交
252
	list_move(&file->f_u.fu_list, list);
L
Linus Torvalds 已提交
253 254 255 256 257
	file_list_unlock();
}

void file_kill(struct file *file)
{
E
Eric Dumazet 已提交
258
	if (!list_empty(&file->f_u.fu_list)) {
L
Linus Torvalds 已提交
259
		file_list_lock();
E
Eric Dumazet 已提交
260
		list_del_init(&file->f_u.fu_list);
L
Linus Torvalds 已提交
261 262 263 264 265 266 267 268 269 270 271
		file_list_unlock();
	}
}

int fs_may_remount_ro(struct super_block *sb)
{
	struct list_head *p;

	/* Check that no files are currently opened for writing. */
	file_list_lock();
	list_for_each(p, &sb->s_files) {
E
Eric Dumazet 已提交
272
		struct file *file = list_entry(p, struct file, f_u.fu_list);
273
		struct inode *inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300

		/* File with pending delete? */
		if (inode->i_nlink == 0)
			goto too_bad;

		/* Writeable file? */
		if (S_ISREG(inode->i_mode) && (file->f_mode & FMODE_WRITE))
			goto too_bad;
	}
	file_list_unlock();
	return 1; /* Tis' cool bro. */
too_bad:
	file_list_unlock();
	return 0;
}

void __init files_init(unsigned long mempages)
{ 
	int n; 
	/* One file with associated inode and dcache is very roughly 1K. 
	 * Per default don't use more than 10% of our memory for files. 
	 */ 

	n = (mempages * (PAGE_SIZE / 1024)) / 10;
	files_stat.max_files = n; 
	if (files_stat.max_files < NR_FILE)
		files_stat.max_files = NR_FILE;
301
	files_defer_init();
302
	percpu_counter_init(&nr_files, 0);
L
Linus Torvalds 已提交
303
}