[linux.git] / fs / file_table.c

/*
 *  linux/fs/file_table.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *  Copyright (C) 1997 David S. Miller ([email protected])
 */

#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>
#include <linux/rcupdate.h>
#include <linux/mount.h>
#include <linux/capability.h>
#include <linux/cdev.h>
#include <linux/fsnotify.h>
#include <linux/sysctl.h>
#include <linux/percpu_counter.h>

#include <asm/atomic.h>

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

/* public. Not pretty! */
__cacheline_aligned_in_smp DEFINE_SPINLOCK(files_lock);

static struct percpu_counter nr_files __cacheline_aligned_in_smp;

static inline void file_free_rcu(struct rcu_head *head)
{
	struct file *f =  container_of(head, struct file, f_u.fu_rcuhead);
	kmem_cache_free(filp_cachep, f);
}

static inline void file_free(struct file *f)
{
	percpu_counter_dec(&nr_files);
	call_rcu(&f->f_u.fu_rcuhead, file_free_rcu);
}

/*
 * Return the total number of open files in the system
 */
static int get_nr_files(void)
{
	return percpu_counter_read_positive(&nr_files);
}

/*
 * Return the maximum number of open files in the system
 */
int get_max_files(void)
{
	return files_stat.max_files;
}
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

/* 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)
{
	struct task_struct *tsk;
	static int old_max;
	struct file * f;

	/*
	 * Privileged users can go above max_files
	 */
	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_positive(&nr_files) >= files_stat.max_files)
			goto over;
	}

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

	percpu_counter_inc(&nr_files);
	if (security_file_alloc(f))
		goto fail_sec;

	tsk = current;
	INIT_LIST_HEAD(&f->f_u.fu_list);
	atomic_set(&f->f_count, 1);
	rwlock_init(&f->f_owner.lock);
	f->f_uid = tsk->fsuid;
	f->f_gid = tsk->fsgid;
	eventpoll_init_file(f);
	/* f->f_version: 0 */
	return f;

over:
	/* Ran out of filps - report that */
	if (get_nr_files() > old_max) {
		printk(KERN_INFO "VFS: file-max limit %d reached\n",
					get_max_files());
		old_max = get_nr_files();
	}
	goto fail;

fail_sec:
	file_free(f);
fail:
	return NULL;
}

EXPORT_SYMBOL(get_empty_filp);

/**
 * alloc_file - allocate and initialize a 'struct file'
 * @mnt: the vfsmount on which the file will reside
 * @dentry: the dentry representing the new file
 * @mode: the mode with which the new file will be opened
 * @fop: the 'struct file_operations' for the new file
 *
 * Use this instead of get_empty_filp() to get a new
 * 'struct file'.  Do so because of the same initialization
 * pitfalls reasons listed for init_file().  This is a
 * preferred interface to using init_file().
 *
 * If all the callers of init_file() are eliminated, its
 * code should be moved into this function.
 */
struct file *alloc_file(struct vfsmount *mnt, struct dentry *dentry,
		mode_t mode, const struct file_operations *fop)
{
	struct file *file;
	struct path;

	file = get_empty_filp();
	if (!file)
		return NULL;

	init_file(file, mnt, dentry, mode, fop);
	return file;
}
EXPORT_SYMBOL(alloc_file);

/**
 * init_file - initialize a 'struct file'
 * @file: the already allocated 'struct file' to initialized
 * @mnt: the vfsmount on which the file resides
 * @dentry: the dentry representing this file
 * @mode: the mode the file is opened with
 * @fop: the 'struct file_operations' for this file
 *
 * Use this instead of setting the members directly.  Doing so
 * avoids making mistakes like forgetting the mntget() or
 * forgetting to take a write on the mnt.
 *
 * Note: This is a crappy interface.  It is here to make
 * merging with the existing users of get_empty_filp()
 * who have complex failure logic easier.  All users
 * of this should be moving to alloc_file().
 */
int init_file(struct file *file, struct vfsmount *mnt, struct dentry *dentry,
	   mode_t mode, const struct file_operations *fop)
{
	int error = 0;
	file->f_path.dentry = dentry;
	file->f_path.mnt = mntget(mnt);
	file->f_mapping = dentry->d_inode->i_mapping;
	file->f_mode = mode;
	file->f_op = fop;
	return error;
}
EXPORT_SYMBOL(init_file);

void fput(struct file *file)
{
	if (atomic_dec_and_test(&file->f_count))
		__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 __fput(struct file *file)
{
	struct dentry *dentry = file->f_path.dentry;
	struct vfsmount *mnt = file->f_path.mnt;
	struct inode *inode = dentry->d_inode;

	might_sleep();

	fsnotify_close(file);
	/*
	 * 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);
	if (unlikely(S_ISCHR(inode->i_mode) && inode->i_cdev != NULL))
		cdev_put(inode->i_cdev);
	fops_put(file->f_op);
	if (file->f_mode & FMODE_WRITE)
		put_write_access(inode);
	put_pid(file->f_owner.pid);
	file_kill(file);
	file->f_path.dentry = NULL;
	file->f_path.mnt = NULL;
	file_free(file);
	dput(dentry);
	mntput(mnt);
}

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

	rcu_read_lock();
	file = fcheck_files(files, fd);
	if (file) {
		if (!atomic_inc_not_zero(&file->f_count)) {
			/* File object ref couldn't be taken */
			rcu_read_unlock();
			return NULL;
		}
	}
	rcu_read_unlock();

	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 *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 {
		rcu_read_lock();
		file = fcheck_files(files, fd);
		if (file) {
			if (atomic_inc_not_zero(&file->f_count))
				*fput_needed = 1;
			else
				/* Didn't get the reference, someone's freed */
				file = NULL;
		}
		rcu_read_unlock();
	}

	return file;
}


void put_filp(struct file *file)
{
	if (atomic_dec_and_test(&file->f_count)) {
		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();
	list_move(&file->f_u.fu_list, list);
	file_list_unlock();
}

void file_kill(struct file *file)
{
	if (!list_empty(&file->f_u.fu_list)) {
		file_list_lock();
		list_del_init(&file->f_u.fu_list);
		file_list_unlock();
	}
}

int fs_may_remount_ro(struct super_block *sb)
{
	struct file *file;

	/* Check that no files are currently opened for writing. */
	file_list_lock();
	list_for_each_entry(file, &sb->s_files, f_u.fu_list) {
		struct inode *inode = file->f_path.dentry->d_inode;

		/* 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;
	files_defer_init();
	percpu_counter_init(&nr_files, 0);
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/fs/file_table.c
	3	*
	4	* Copyright (C) 1991, 1992 Linus Torvalds
	5	* Copyright (C) 1997 David S. Miller ([email protected])
	6	*/
	7
	8	#include <linux/string.h>
	9	#include <linux/slab.h>
	10	#include <linux/file.h>
	11	#include <linux/init.h>
	12	#include <linux/module.h>
1da177e4 LT	13	#include <linux/fs.h>
	14	#include <linux/security.h>
	15	#include <linux/eventpoll.h>
ab2af1f5	16	#include <linux/rcupdate.h>
1da177e4	17	#include <linux/mount.h>
16f7e0fe	18	#include <linux/capability.h>
1da177e4	19	#include <linux/cdev.h>
0eeca283	20	#include <linux/fsnotify.h>
529bf6be DS	21	#include <linux/sysctl.h>
	22	#include <linux/percpu_counter.h>
	23
	24	#include <asm/atomic.h>
1da177e4 LT	25
	26	/* sysctl tunables... */
	27	struct files_stat_struct files_stat = {
	28	.max_files = NR_FILE
	29	};
	30
1da177e4	31	/* public. Not pretty! */
529bf6be	32	__cacheline_aligned_in_smp DEFINE_SPINLOCK(files_lock);
1da177e4	33
529bf6be	34	static struct percpu_counter nr_files __cacheline_aligned_in_smp;
1da177e4	35
529bf6be	36	static inline void file_free_rcu(struct rcu_head *head)
1da177e4	37	{
529bf6be DS	38	struct file *f = container_of(head, struct file, f_u.fu_rcuhead);
529bf6be DS	39	kmem_cache_free(filp_cachep, f);
1da177e4 LT	40	}
1da177e4 LT	41
529bf6be	42	static inline void file_free(struct file *f)
1da177e4	43	{
529bf6be DS	44	percpu_counter_dec(&nr_files);
529bf6be DS	45	call_rcu(&f->f_u.fu_rcuhead, file_free_rcu);
1da177e4 LT	46	}
1da177e4 LT	47
529bf6be DS	48	/*
	49	* Return the total number of open files in the system
	50	*/
	51	static int get_nr_files(void)
1da177e4	52	{
529bf6be	53	return percpu_counter_read_positive(&nr_files);
1da177e4 LT	54	}
1da177e4 LT	55
529bf6be DS	56	/*
	57	* Return the maximum number of open files in the system
	58	*/
	59	int get_max_files(void)
ab2af1f5	60	{
529bf6be	61	return files_stat.max_files;
ab2af1f5	62	}
529bf6be DS	63	EXPORT_SYMBOL_GPL(get_max_files);
	64
	65	/*
	66	* Handle nr_files sysctl
	67	*/
	68	#if defined(CONFIG_SYSCTL) && defined(CONFIG_PROC_FS)
	69	int proc_nr_files(ctl_table table, int write, struct file filp,
	70	void __user buffer, size_t lenp, loff_t *ppos)
	71	{
	72	files_stat.nr_files = get_nr_files();
	73	return proc_dointvec(table, write, filp, buffer, lenp, ppos);
	74	}
	75	#else
	76	int proc_nr_files(ctl_table table, int write, struct file filp,
	77	void __user buffer, size_t lenp, loff_t *ppos)
	78	{
	79	return -ENOSYS;
	80	}
	81	#endif
ab2af1f5	82
1da177e4 LT	83	/* Find an unused file structure and return a pointer to it.
	84	* Returns NULL, if there are no more free file structures or
	85	* we run out of memory.
	86	*/
	87	struct file *get_empty_filp(void)
	88	{
5a6b7951	89	struct task_struct *tsk;
af4d2ecb	90	static int old_max;
1da177e4 LT	91	struct file * f;
	92
	93	/*
	94	* Privileged users can go above max_files
	95	*/
529bf6be DS	96	if (get_nr_files() >= files_stat.max_files && !capable(CAP_SYS_ADMIN)) {
	97	/*
	98	* percpu_counters are inaccurate. Do an expensive check before
	99	* we go and fail.
	100	*/
52d9f3b4	101	if (percpu_counter_sum_positive(&nr_files) >= files_stat.max_files)
529bf6be DS	102	goto over;
529bf6be DS	103	}
af4d2ecb	104
4975e45f	105	f = kmem_cache_zalloc(filp_cachep, GFP_KERNEL);
af4d2ecb KK	106	if (f == NULL)
	107	goto fail;
	108
529bf6be	109	percpu_counter_inc(&nr_files);
af4d2ecb KK	110	if (security_file_alloc(f))
af4d2ecb KK	111	goto fail_sec;
1da177e4	112
5a6b7951 BL	113	tsk = current;
5a6b7951 BL	114	INIT_LIST_HEAD(&f->f_u.fu_list);
af4d2ecb	115	atomic_set(&f->f_count, 1);
af4d2ecb	116	rwlock_init(&f->f_owner.lock);
5a6b7951 BL	117	f->f_uid = tsk->fsuid;
	118	f->f_gid = tsk->fsgid;
	119	eventpoll_init_file(f);
af4d2ecb	120	/* f->f_version: 0 */
af4d2ecb KK	121	return f;
	122
	123	over:
1da177e4	124	/* Ran out of filps - report that */
529bf6be	125	if (get_nr_files() > old_max) {
1da177e4	126	printk(KERN_INFO "VFS: file-max limit %d reached\n",
529bf6be DS	127	get_max_files());
529bf6be DS	128	old_max = get_nr_files();
1da177e4	129	}
af4d2ecb KK	130	goto fail;
	131
	132	fail_sec:
	133	file_free(f);
1da177e4 LT	134	fail:
	135	return NULL;
	136	}
	137
	138	EXPORT_SYMBOL(get_empty_filp);
	139
ce8d2cdf DH	140	/**
	141	* alloc_file - allocate and initialize a 'struct file'
	142	* @mnt: the vfsmount on which the file will reside
	143	* @dentry: the dentry representing the new file
	144	* @mode: the mode with which the new file will be opened
	145	* @fop: the 'struct file_operations' for the new file
	146	*
	147	* Use this instead of get_empty_filp() to get a new
	148	* 'struct file'. Do so because of the same initialization
	149	* pitfalls reasons listed for init_file(). This is a
	150	* preferred interface to using init_file().
	151	*
	152	* If all the callers of init_file() are eliminated, its
	153	* code should be moved into this function.
	154	*/
	155	struct file alloc_file(struct vfsmount mnt, struct dentry *dentry,
	156	mode_t mode, const struct file_operations *fop)
	157	{
	158	struct file *file;
	159	struct path;
	160
	161	file = get_empty_filp();
	162	if (!file)
	163	return NULL;
	164
	165	init_file(file, mnt, dentry, mode, fop);
	166	return file;
	167	}
	168	EXPORT_SYMBOL(alloc_file);
	169
	170	/**
	171	* init_file - initialize a 'struct file'
	172	* @file: the already allocated 'struct file' to initialized
	173	* @mnt: the vfsmount on which the file resides
	174	* @dentry: the dentry representing this file
	175	* @mode: the mode the file is opened with
	176	* @fop: the 'struct file_operations' for this file
	177	*
	178	* Use this instead of setting the members directly. Doing so
	179	* avoids making mistakes like forgetting the mntget() or
	180	* forgetting to take a write on the mnt.
	181	*
	182	* Note: This is a crappy interface. It is here to make
	183	* merging with the existing users of get_empty_filp()
	184	* who have complex failure logic easier. All users
	185	* of this should be moving to alloc_file().
	186	*/
	187	int init_file(struct file file, struct vfsmount mnt, struct dentry *dentry,
	188	mode_t mode, const struct file_operations *fop)
	189	{
	190	int error = 0;
	191	file->f_path.dentry = dentry;
	192	file->f_path.mnt = mntget(mnt);
	193	file->f_mapping = dentry->d_inode->i_mapping;
	194	file->f_mode = mode;
	195	file->f_op = fop;
	196	return error;
	197	}
	198	EXPORT_SYMBOL(init_file);
	199
fc9b52cd	200	void fput(struct file *file)
1da177e4	201	{
095975da	202	if (atomic_dec_and_test(&file->f_count))
1da177e4 LT	203	__fput(file);
	204	}
	205
	206	EXPORT_SYMBOL(fput);
	207
	208	/* __fput is called from task context when aio completion releases the last
	209	* last use of a struct file *. Do not use otherwise.
	210	*/
fc9b52cd	211	void __fput(struct file *file)
1da177e4	212	{
0f7fc9e4 JJS	213	struct dentry *dentry = file->f_path.dentry;
0f7fc9e4 JJS	214	struct vfsmount *mnt = file->f_path.mnt;
1da177e4 LT	215	struct inode *inode = dentry->d_inode;
	216
	217	might_sleep();
0eeca283 RL	218
0eeca283 RL	219	fsnotify_close(file);
1da177e4 LT	220	/*
	221	* The function eventpoll_release() should be the first called
	222	* in the file cleanup chain.
	223	*/
	224	eventpoll_release(file);
	225	locks_remove_flock(file);
	226
	227	if (file->f_op && file->f_op->release)
	228	file->f_op->release(inode, file);
	229	security_file_free(file);
577c4eb0	230	if (unlikely(S_ISCHR(inode->i_mode) && inode->i_cdev != NULL))
1da177e4 LT	231	cdev_put(inode->i_cdev);
	232	fops_put(file->f_op);
	233	if (file->f_mode & FMODE_WRITE)
	234	put_write_access(inode);
609d7fa9	235	put_pid(file->f_owner.pid);
1da177e4	236	file_kill(file);
0f7fc9e4 JJS	237	file->f_path.dentry = NULL;
0f7fc9e4 JJS	238	file->f_path.mnt = NULL;
1da177e4 LT	239	file_free(file);
	240	dput(dentry);
	241	mntput(mnt);
	242	}
	243
fc9b52cd	244	struct file *fget(unsigned int fd)
1da177e4 LT	245	{
	246	struct file *file;
	247	struct files_struct *files = current->files;
	248
ab2af1f5	249	rcu_read_lock();
1da177e4	250	file = fcheck_files(files, fd);
ab2af1f5	251	if (file) {
095975da	252	if (!atomic_inc_not_zero(&file->f_count)) {
ab2af1f5 DS	253	/* File object ref couldn't be taken */
	254	rcu_read_unlock();
	255	return NULL;
	256	}
	257	}
	258	rcu_read_unlock();
	259
1da177e4 LT	260	return file;
	261	}
	262
	263	EXPORT_SYMBOL(fget);
	264
	265	/*
	266	* Lightweight file lookup - no refcnt increment if fd table isn't shared.
	267	* You can use this only if it is guranteed that the current task already
	268	* holds a refcnt to that file. That check has to be done at fget() only
	269	* and a flag is returned to be passed to the corresponding fput_light().
	270	* There must not be a cloning between an fget_light/fput_light pair.
	271	*/
fc9b52cd	272	struct file fget_light(unsigned int fd, int fput_needed)
1da177e4 LT	273	{
	274	struct file *file;
	275	struct files_struct *files = current->files;
	276
	277	*fput_needed = 0;
	278	if (likely((atomic_read(&files->count) == 1))) {
	279	file = fcheck_files(files, fd);
	280	} else {
ab2af1f5	281	rcu_read_lock();
1da177e4 LT	282	file = fcheck_files(files, fd);
1da177e4 LT	283	if (file) {
095975da	284	if (atomic_inc_not_zero(&file->f_count))
ab2af1f5 DS	285	*fput_needed = 1;
	286	else
	287	/* Didn't get the reference, someone's freed */
	288	file = NULL;
1da177e4	289	}
ab2af1f5	290	rcu_read_unlock();
1da177e4	291	}
ab2af1f5	292
1da177e4 LT	293	return file;
	294	}
	295
	296
	297	void put_filp(struct file *file)
	298	{
095975da	299	if (atomic_dec_and_test(&file->f_count)) {
1da177e4 LT	300	security_file_free(file);
	301	file_kill(file);
	302	file_free(file);
	303	}
	304	}
	305
	306	void file_move(struct file file, struct list_head list)
	307	{
	308	if (!list)
	309	return;
	310	file_list_lock();
2f512016	311	list_move(&file->f_u.fu_list, list);
1da177e4 LT	312	file_list_unlock();
	313	}
	314
	315	void file_kill(struct file *file)
	316	{
2f512016	317	if (!list_empty(&file->f_u.fu_list)) {
1da177e4	318	file_list_lock();
2f512016	319	list_del_init(&file->f_u.fu_list);
1da177e4 LT	320	file_list_unlock();
	321	}
	322	}
	323
	324	int fs_may_remount_ro(struct super_block *sb)
	325	{
cfdaf9e5	326	struct file *file;
1da177e4 LT	327
	328	/* Check that no files are currently opened for writing. */
	329	file_list_lock();
cfdaf9e5	330	list_for_each_entry(file, &sb->s_files, f_u.fu_list) {
0f7fc9e4	331	struct inode *inode = file->f_path.dentry->d_inode;
1da177e4 LT	332
	333	/* File with pending delete? */
	334	if (inode->i_nlink == 0)
	335	goto too_bad;
	336
	337	/* Writeable file? */
	338	if (S_ISREG(inode->i_mode) && (file->f_mode & FMODE_WRITE))
	339	goto too_bad;
	340	}
	341	file_list_unlock();
	342	return 1; /* Tis' cool bro. */
	343	too_bad:
	344	file_list_unlock();
	345	return 0;
	346	}
	347
	348	void __init files_init(unsigned long mempages)
	349	{
	350	int n;
	351	/* One file with associated inode and dcache is very roughly 1K.
	352	* Per default don't use more than 10% of our memory for files.
	353	*/
	354
	355	n = (mempages * (PAGE_SIZE / 1024)) / 10;
	356	files_stat.max_files = n;
	357	if (files_stat.max_files < NR_FILE)
	358	files_stat.max_files = NR_FILE;
ab2af1f5	359	files_defer_init();
0216bfcf	360	percpu_counter_init(&nr_files, 0);
1da177e4	361	}