[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/fdtable.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/fs.h>
#include <linux/security.h>
#include <linux/cred.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 <linux/percpu.h>
#include <linux/task_work.h>
#include <linux/ima.h>
#include <linux/swap.h>

#include <linux/atomic.h>

#include "internal.h"

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

/* SLAB cache for file structures */
static struct kmem_cache *filp_cachep __read_mostly;

static struct percpu_counter nr_files __cacheline_aligned_in_smp;

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

	put_cred(f->f_cred);
	kmem_cache_free(filp_cachep, f);
}

static inline void file_free(struct file *f)
{
	security_file_free(f);
	if (!(f->f_mode & FMODE_NOACCOUNT))
		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 long get_nr_files(void)
{
	return percpu_counter_read_positive(&nr_files);
}

/*
 * Return the maximum number of open files in the system
 */
unsigned long 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(struct ctl_table *table, int write,
                     void __user *buffer, size_t *lenp, loff_t *ppos)
{
	files_stat.nr_files = get_nr_files();
	return proc_doulongvec_minmax(table, write, buffer, lenp, ppos);
}
#else
int proc_nr_files(struct ctl_table *table, int write,
                     void __user *buffer, size_t *lenp, loff_t *ppos)
{
	return -ENOSYS;
}
#endif

static struct file *__alloc_file(int flags, const struct cred *cred)
{
	struct file *f;
	int error;

	f = kmem_cache_zalloc(filp_cachep, GFP_KERNEL);
	if (unlikely(!f))
		return ERR_PTR(-ENOMEM);

	f->f_cred = get_cred(cred);
	error = security_file_alloc(f);
	if (unlikely(error)) {
		file_free_rcu(&f->f_u.fu_rcuhead);
		return ERR_PTR(error);
	}

	atomic_long_set(&f->f_count, 1);
	rwlock_init(&f->f_owner.lock);
	spin_lock_init(&f->f_lock);
	mutex_init(&f->f_pos_lock);
	eventpoll_init_file(f);
	f->f_flags = flags;
	f->f_mode = OPEN_FMODE(flags);
	/* f->f_version: 0 */

	return f;
}

/* Find an unused file structure and return a pointer to it.
 * Returns an error pointer if some error happend e.g. we over file
 * structures limit, run out of memory or operation is not permitted.
 *
 * Be very careful using this.  You are responsible for
 * getting write access to any mount that you might assign
 * to this filp, if it is opened for write.  If this is not
 * done, you will imbalance int the mount's writer count
 * and a warning at __fput() time.
 */
struct file *alloc_empty_file(int flags, const struct cred *cred)
{
	static long 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 = __alloc_file(flags, cred);
	if (!IS_ERR(f))
		percpu_counter_inc(&nr_files);

	return f;

over:
	/* Ran out of filps - report that */
	if (get_nr_files() > old_max) {
		pr_info("VFS: file-max limit %lu reached\n", get_max_files());
		old_max = get_nr_files();
	}
	return ERR_PTR(-ENFILE);
}

/*
 * Variant of alloc_empty_file() that doesn't check and modify nr_files.
 *
 * Should not be used unless there's a very good reason to do so.
 */
struct file *alloc_empty_file_noaccount(int flags, const struct cred *cred)
{
	struct file *f = __alloc_file(flags, cred);

	if (!IS_ERR(f))
		f->f_mode |= FMODE_NOACCOUNT;

	return f;
}

/**
 * alloc_file - allocate and initialize a 'struct file'
 *
 * @path: the (dentry, vfsmount) pair for the new file
 * @flags: O_... flags with which the new file will be opened
 * @fop: the 'struct file_operations' for the new file
 */
static struct file *alloc_file(const struct path *path, int flags,
		const struct file_operations *fop)
{
	struct file *file;

	file = alloc_empty_file(flags, current_cred());
	if (IS_ERR(file))
		return file;

	file->f_path = *path;
	file->f_inode = path->dentry->d_inode;
	file->f_mapping = path->dentry->d_inode->i_mapping;
	file->f_wb_err = filemap_sample_wb_err(file->f_mapping);
	if ((file->f_mode & FMODE_READ) &&
	     likely(fop->read || fop->read_iter))
		file->f_mode |= FMODE_CAN_READ;
	if ((file->f_mode & FMODE_WRITE) &&
	     likely(fop->write || fop->write_iter))
		file->f_mode |= FMODE_CAN_WRITE;
	file->f_mode |= FMODE_OPENED;
	file->f_op = fop;
	if ((file->f_mode & (FMODE_READ | FMODE_WRITE)) == FMODE_READ)
		i_readcount_inc(path->dentry->d_inode);
	return file;
}

struct file *alloc_file_pseudo(struct inode *inode, struct vfsmount *mnt,
				const char *name, int flags,
				const struct file_operations *fops)
{
	static const struct dentry_operations anon_ops = {
		.d_dname = simple_dname
	};
	struct qstr this = QSTR_INIT(name, strlen(name));
	struct path path;
	struct file *file;

	path.dentry = d_alloc_pseudo(mnt->mnt_sb, &this);
	if (!path.dentry)
		return ERR_PTR(-ENOMEM);
	if (!mnt->mnt_sb->s_d_op)
		d_set_d_op(path.dentry, &anon_ops);
	path.mnt = mntget(mnt);
	d_instantiate(path.dentry, inode);
	file = alloc_file(&path, flags, fops);
	if (IS_ERR(file)) {
		ihold(inode);
		path_put(&path);
	}
	return file;
}
EXPORT_SYMBOL(alloc_file_pseudo);

struct file *alloc_file_clone(struct file *base, int flags,
				const struct file_operations *fops)
{
	struct file *f = alloc_file(&base->f_path, flags, fops);
	if (!IS_ERR(f)) {
		path_get(&f->f_path);
		f->f_mapping = base->f_mapping;
	}
	return f;
}

/* the real guts of fput() - releasing the last reference to file
 */
static void __fput(struct file *file)
{
	struct dentry *dentry = file->f_path.dentry;
	struct vfsmount *mnt = file->f_path.mnt;
	struct inode *inode = file->f_inode;

	if (unlikely(!(file->f_mode & FMODE_OPENED)))
		goto out;

	might_sleep();

	fsnotify_close(file);
	/*
	 * The function eventpoll_release() should be the first called
	 * in the file cleanup chain.
	 */
	eventpoll_release(file);
	locks_remove_file(file);

	ima_file_free(file);
	if (unlikely(file->f_flags & FASYNC)) {
		if (file->f_op->fasync)
			file->f_op->fasync(-1, file, 0);
	}
	if (file->f_op->release)
		file->f_op->release(inode, file);
	if (unlikely(S_ISCHR(inode->i_mode) && inode->i_cdev != NULL &&
		     !(file->f_mode & FMODE_PATH))) {
		cdev_put(inode->i_cdev);
	}
	fops_put(file->f_op);
	put_pid(file->f_owner.pid);
	if ((file->f_mode & (FMODE_READ | FMODE_WRITE)) == FMODE_READ)
		i_readcount_dec(inode);
	if (file->f_mode & FMODE_WRITER) {
		put_write_access(inode);
		__mnt_drop_write(mnt);
	}
	dput(dentry);
	mntput(mnt);
out:
	file_free(file);
}

static LLIST_HEAD(delayed_fput_list);
static void delayed_fput(struct work_struct *unused)
{
	struct llist_node *node = llist_del_all(&delayed_fput_list);
	struct file *f, *t;

	llist_for_each_entry_safe(f, t, node, f_u.fu_llist)
		__fput(f);
}

static void ____fput(struct callback_head *work)
{
	__fput(container_of(work, struct file, f_u.fu_rcuhead));
}

/*
 * If kernel thread really needs to have the final fput() it has done
 * to complete, call this.  The only user right now is the boot - we
 * *do* need to make sure our writes to binaries on initramfs has
 * not left us with opened struct file waiting for __fput() - execve()
 * won't work without that.  Please, don't add more callers without
 * very good reasons; in particular, never call that with locks
 * held and never call that from a thread that might need to do
 * some work on any kind of umount.
 */
void flush_delayed_fput(void)
{
	delayed_fput(NULL);
}

static DECLARE_DELAYED_WORK(delayed_fput_work, delayed_fput);

void fput(struct file *file)
{
	if (atomic_long_dec_and_test(&file->f_count)) {
		struct task_struct *task = current;

		if (likely(!in_interrupt() && !(task->flags & PF_KTHREAD))) {
			init_task_work(&file->f_u.fu_rcuhead, ____fput);
			if (!task_work_add(task, &file->f_u.fu_rcuhead, true))
				return;
			/*
			 * After this task has run exit_task_work(),
			 * task_work_add() will fail.  Fall through to delayed
			 * fput to avoid leaking *file.
			 */
		}

		if (llist_add(&file->f_u.fu_llist, &delayed_fput_list))
			schedule_delayed_work(&delayed_fput_work, 1);
	}
}

/*
 * synchronous analog of fput(); for kernel threads that might be needed
 * in some umount() (and thus can't use flush_delayed_fput() without
 * risking deadlocks), need to wait for completion of __fput() and know
 * for this specific struct file it won't involve anything that would
 * need them.  Use only if you really need it - at the very least,
 * don't blindly convert fput() by kernel thread to that.
 */
void __fput_sync(struct file *file)
{
	if (atomic_long_dec_and_test(&file->f_count)) {
		struct task_struct *task = current;
		BUG_ON(!(task->flags & PF_KTHREAD));
		__fput(file);
	}
}

EXPORT_SYMBOL(fput);

void __init files_init(void)
{
	filp_cachep = kmem_cache_create("filp", sizeof(struct file), 0,
			SLAB_HWCACHE_ALIGN | SLAB_PANIC | SLAB_ACCOUNT, NULL);
	percpu_counter_init(&nr_files, 0, GFP_KERNEL);
}

/*
 * One file with associated inode and dcache is very roughly 1K. Per default
 * do not use more than 10% of our memory for files.
 */
void __init files_maxfiles_init(void)
{
	unsigned long n;
	unsigned long memreserve = (totalram_pages - nr_free_pages()) * 3/2;

	memreserve = min(memreserve, totalram_pages - 1);
	n = ((totalram_pages - memreserve) * (PAGE_SIZE / 1024)) / 10;

	files_stat.max_files = max_t(unsigned long, n, NR_FILE);
}
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>
9f3acc31	11	#include <linux/fdtable.h>
1da177e4 LT	12	#include <linux/init.h>
1da177e4 LT	13	#include <linux/module.h>
1da177e4 LT	14	#include <linux/fs.h>
1da177e4 LT	15	#include <linux/security.h>
5b825c3a	16	#include <linux/cred.h>
1da177e4	17	#include <linux/eventpoll.h>
ab2af1f5	18	#include <linux/rcupdate.h>
1da177e4	19	#include <linux/mount.h>
16f7e0fe	20	#include <linux/capability.h>
1da177e4	21	#include <linux/cdev.h>
0eeca283	22	#include <linux/fsnotify.h>
529bf6be DS	23	#include <linux/sysctl.h>
529bf6be DS	24	#include <linux/percpu_counter.h>
6416ccb7	25	#include <linux/percpu.h>
4a9d4b02	26	#include <linux/task_work.h>
0552f879	27	#include <linux/ima.h>
4248b0da	28	#include <linux/swap.h>
529bf6be	29
60063497	30	#include <linux/atomic.h>
1da177e4	31
e81e3f4d EP	32	#include "internal.h"
e81e3f4d EP	33
1da177e4 LT	34	/* sysctl tunables... */
	35	struct files_stat_struct files_stat = {
	36	.max_files = NR_FILE
	37	};
	38
b6b3fdea ED	39	/* SLAB cache for file structures */
	40	static struct kmem_cache *filp_cachep __read_mostly;
	41
529bf6be	42	static struct percpu_counter nr_files __cacheline_aligned_in_smp;
1da177e4	43
5c33b183	44	static void file_free_rcu(struct rcu_head *head)
1da177e4	45	{
d76b0d9b DH	46	struct file *f = container_of(head, struct file, f_u.fu_rcuhead);
	47
	48	put_cred(f->f_cred);
529bf6be	49	kmem_cache_free(filp_cachep, f);
1da177e4 LT	50	}
1da177e4 LT	51
529bf6be	52	static inline void file_free(struct file *f)
1da177e4	53	{
e8cff84f	54	security_file_free(f);
d3b1084d MS	55	if (!(f->f_mode & FMODE_NOACCOUNT))
d3b1084d MS	56	percpu_counter_dec(&nr_files);
529bf6be	57	call_rcu(&f->f_u.fu_rcuhead, file_free_rcu);
1da177e4 LT	58	}
1da177e4 LT	59
529bf6be DS	60	/*
	61	* Return the total number of open files in the system
	62	*/
518de9b3	63	static long get_nr_files(void)
1da177e4	64	{
529bf6be	65	return percpu_counter_read_positive(&nr_files);
1da177e4 LT	66	}
1da177e4 LT	67
529bf6be DS	68	/*
	69	* Return the maximum number of open files in the system
	70	*/
518de9b3	71	unsigned long get_max_files(void)
ab2af1f5	72	{
529bf6be	73	return files_stat.max_files;
ab2af1f5	74	}
529bf6be DS	75	EXPORT_SYMBOL_GPL(get_max_files);
	76
	77	/*
	78	* Handle nr_files sysctl
	79	*/
	80	#if defined(CONFIG_SYSCTL) && defined(CONFIG_PROC_FS)
1f7e0616	81	int proc_nr_files(struct ctl_table *table, int write,
529bf6be DS	82	void __user buffer, size_t lenp, loff_t *ppos)
	83	{
	84	files_stat.nr_files = get_nr_files();
518de9b3	85	return proc_doulongvec_minmax(table, write, buffer, lenp, ppos);
529bf6be DS	86	}
529bf6be DS	87	#else
1f7e0616	88	int proc_nr_files(struct ctl_table *table, int write,
529bf6be DS	89	void __user buffer, size_t lenp, loff_t *ppos)
	90	{
	91	return -ENOSYS;
	92	}
	93	#endif
ab2af1f5	94
d3b1084d MS	95	static struct file __alloc_file(int flags, const struct cred cred)
	96	{
	97	struct file *f;
	98	int error;
	99
	100	f = kmem_cache_zalloc(filp_cachep, GFP_KERNEL);
	101	if (unlikely(!f))
	102	return ERR_PTR(-ENOMEM);
	103
	104	f->f_cred = get_cred(cred);
	105	error = security_file_alloc(f);
	106	if (unlikely(error)) {
	107	file_free_rcu(&f->f_u.fu_rcuhead);
	108	return ERR_PTR(error);
	109	}
	110
	111	atomic_long_set(&f->f_count, 1);
	112	rwlock_init(&f->f_owner.lock);
	113	spin_lock_init(&f->f_lock);
	114	mutex_init(&f->f_pos_lock);
	115	eventpoll_init_file(f);
	116	f->f_flags = flags;
	117	f->f_mode = OPEN_FMODE(flags);
	118	/* f->f_version: 0 */
	119
	120	return f;
	121	}
	122
1da177e4	123	/* Find an unused file structure and return a pointer to it.
1afc99be AV	124	* Returns an error pointer if some error happend e.g. we over file
1afc99be AV	125	* structures limit, run out of memory or operation is not permitted.
430e285e DH	126	*
	127	* Be very careful using this. You are responsible for
	128	* getting write access to any mount that you might assign
	129	* to this filp, if it is opened for write. If this is not
	130	* done, you will imbalance int the mount's writer count
	131	* and a warning at __fput() time.
1da177e4	132	*/
ea73ea72	133	struct file alloc_empty_file(int flags, const struct cred cred)
1da177e4	134	{
518de9b3	135	static long old_max;
1afc99be	136	struct file *f;
1da177e4 LT	137
	138	/*
	139	* Privileged users can go above max_files
	140	*/
529bf6be DS	141	if (get_nr_files() >= files_stat.max_files && !capable(CAP_SYS_ADMIN)) {
	142	/*
	143	* percpu_counters are inaccurate. Do an expensive check before
	144	* we go and fail.
	145	*/
52d9f3b4	146	if (percpu_counter_sum_positive(&nr_files) >= files_stat.max_files)
529bf6be DS	147	goto over;
529bf6be DS	148	}
af4d2ecb	149
d3b1084d MS	150	f = __alloc_file(flags, cred);
	151	if (!IS_ERR(f))
	152	percpu_counter_inc(&nr_files);
1da177e4	153
af4d2ecb KK	154	return f;
	155
	156	over:
1da177e4	157	/* Ran out of filps - report that */
529bf6be	158	if (get_nr_files() > old_max) {
518de9b3	159	pr_info("VFS: file-max limit %lu reached\n", get_max_files());
529bf6be	160	old_max = get_nr_files();
1da177e4	161	}
1afc99be	162	return ERR_PTR(-ENFILE);
1da177e4 LT	163	}
1da177e4 LT	164
d3b1084d MS	165	/*
	166	* Variant of alloc_empty_file() that doesn't check and modify nr_files.
	167	*
	168	* Should not be used unless there's a very good reason to do so.
	169	*/
	170	struct file alloc_empty_file_noaccount(int flags, const struct cred cred)
	171	{
	172	struct file *f = __alloc_file(flags, cred);
	173
	174	if (!IS_ERR(f))
	175	f->f_mode \|= FMODE_NOACCOUNT;
	176
	177	return f;
	178	}
	179
ce8d2cdf DH	180	/**
ce8d2cdf DH	181	* alloc_file - allocate and initialize a 'struct file'
a457606a EB	182	*
a457606a EB	183	* @path: the (dentry, vfsmount) pair for the new file
c9c554f2	184	* @flags: O_... flags with which the new file will be opened
ce8d2cdf	185	* @fop: the 'struct file_operations' for the new file
ce8d2cdf	186	*/
ee1904ba	187	static struct file alloc_file(const struct path path, int flags,
2c48b9c4	188	const struct file_operations *fop)
ce8d2cdf DH	189	{
ce8d2cdf DH	190	struct file *file;
ce8d2cdf	191
ea73ea72	192	file = alloc_empty_file(flags, current_cred());
1afc99be	193	if (IS_ERR(file))
39b65252	194	return file;
ce8d2cdf	195
2c48b9c4	196	file->f_path = *path;
dd37978c	197	file->f_inode = path->dentry->d_inode;
2c48b9c4	198	file->f_mapping = path->dentry->d_inode->i_mapping;
5660e13d	199	file->f_wb_err = filemap_sample_wb_err(file->f_mapping);
c9c554f2	200	if ((file->f_mode & FMODE_READ) &&
84363182	201	likely(fop->read \|\| fop->read_iter))
c9c554f2 AV	202	file->f_mode \|= FMODE_CAN_READ;
c9c554f2 AV	203	if ((file->f_mode & FMODE_WRITE) &&
84363182	204	likely(fop->write \|\| fop->write_iter))
c9c554f2	205	file->f_mode \|= FMODE_CAN_WRITE;
f5d11409	206	file->f_mode \|= FMODE_OPENED;
ce8d2cdf	207	file->f_op = fop;
c9c554f2	208	if ((file->f_mode & (FMODE_READ \| FMODE_WRITE)) == FMODE_READ)
890275b5	209	i_readcount_inc(path->dentry->d_inode);
3d1e4631	210	return file;
ce8d2cdf	211	}
ce8d2cdf	212
d93aa9d8 AV	213	struct file alloc_file_pseudo(struct inode inode, struct vfsmount *mnt,
	214	const char *name, int flags,
	215	const struct file_operations *fops)
	216	{
	217	static const struct dentry_operations anon_ops = {
	218	.d_dname = simple_dname
	219	};
	220	struct qstr this = QSTR_INIT(name, strlen(name));
	221	struct path path;
	222	struct file *file;
	223
	224	path.dentry = d_alloc_pseudo(mnt->mnt_sb, &this);
	225	if (!path.dentry)
	226	return ERR_PTR(-ENOMEM);
	227	if (!mnt->mnt_sb->s_d_op)
	228	d_set_d_op(path.dentry, &anon_ops);
	229	path.mnt = mntget(mnt);
	230	d_instantiate(path.dentry, inode);
	231	file = alloc_file(&path, flags, fops);
	232	if (IS_ERR(file)) {
	233	ihold(inode);
	234	path_put(&path);
	235	}
	236	return file;
	237	}
	238	EXPORT_SYMBOL(alloc_file_pseudo);
	239
183266f2 AV	240	struct file alloc_file_clone(struct file base, int flags,
	241	const struct file_operations *fops)
	242	{
	243	struct file *f = alloc_file(&base->f_path, flags, fops);
	244	if (!IS_ERR(f)) {
	245	path_get(&f->f_path);
	246	f->f_mapping = base->f_mapping;
	247	}
	248	return f;
	249	}
	250
d7065da0	251	/* the real guts of fput() - releasing the last reference to file
1da177e4	252	*/
d7065da0	253	static void __fput(struct file *file)
1da177e4	254	{
0f7fc9e4 JJS	255	struct dentry *dentry = file->f_path.dentry;
0f7fc9e4 JJS	256	struct vfsmount *mnt = file->f_path.mnt;
c77cecee	257	struct inode *inode = file->f_inode;
1da177e4	258
4d27f326 AV	259	if (unlikely(!(file->f_mode & FMODE_OPENED)))
	260	goto out;
	261
1da177e4	262	might_sleep();
0eeca283 RL	263
0eeca283 RL	264	fsnotify_close(file);
1da177e4 LT	265	/*
	266	* The function eventpoll_release() should be the first called
	267	* in the file cleanup chain.
	268	*/
	269	eventpoll_release(file);
78ed8a13	270	locks_remove_file(file);
1da177e4	271
bb02b186	272	ima_file_free(file);
233e70f4	273	if (unlikely(file->f_flags & FASYNC)) {
72c2d531	274	if (file->f_op->fasync)
233e70f4 AV	275	file->f_op->fasync(-1, file, 0);
233e70f4 AV	276	}
72c2d531	277	if (file->f_op->release)
1da177e4	278	file->f_op->release(inode, file);
60ed8cf7 MS	279	if (unlikely(S_ISCHR(inode->i_mode) && inode->i_cdev != NULL &&
60ed8cf7 MS	280	!(file->f_mode & FMODE_PATH))) {
1da177e4	281	cdev_put(inode->i_cdev);
60ed8cf7	282	}
1da177e4	283	fops_put(file->f_op);
609d7fa9	284	put_pid(file->f_owner.pid);
890275b5 MZ	285	if ((file->f_mode & (FMODE_READ \| FMODE_WRITE)) == FMODE_READ)
890275b5 MZ	286	i_readcount_dec(inode);
83f936c7 AV	287	if (file->f_mode & FMODE_WRITER) {
	288	put_write_access(inode);
	289	__mnt_drop_write(mnt);
	290	}
1da177e4 LT	291	dput(dentry);
1da177e4 LT	292	mntput(mnt);
4d27f326 AV	293	out:
4d27f326 AV	294	file_free(file);
1da177e4 LT	295	}
1da177e4 LT	296
4f5e65a1	297	static LLIST_HEAD(delayed_fput_list);
4a9d4b02 AV	298	static void delayed_fput(struct work_struct *unused)
4a9d4b02 AV	299	{
4f5e65a1	300	struct llist_node *node = llist_del_all(&delayed_fput_list);
b9ea557e	301	struct file f, t;
4f5e65a1	302
b9ea557e BP	303	llist_for_each_entry_safe(f, t, node, f_u.fu_llist)
b9ea557e BP	304	__fput(f);
4a9d4b02 AV	305	}
	306
	307	static void ____fput(struct callback_head *work)
	308	{
	309	__fput(container_of(work, struct file, f_u.fu_rcuhead));
	310	}
	311
	312	/*
	313	* If kernel thread really needs to have the final fput() it has done
	314	* to complete, call this. The only user right now is the boot - we
	315	* do need to make sure our writes to binaries on initramfs has
	316	* not left us with opened struct file waiting for __fput() - execve()
	317	* won't work without that. Please, don't add more callers without
	318	* very good reasons; in particular, never call that with locks
	319	* held and never call that from a thread that might need to do
	320	* some work on any kind of umount.
	321	*/
	322	void flush_delayed_fput(void)
	323	{
	324	delayed_fput(NULL);
	325	}
	326
c7314d74	327	static DECLARE_DELAYED_WORK(delayed_fput_work, delayed_fput);
4a9d4b02	328
d7065da0 AV	329	void fput(struct file *file)
d7065da0 AV	330	{
4a9d4b02 AV	331	if (atomic_long_dec_and_test(&file->f_count)) {
4a9d4b02 AV	332	struct task_struct *task = current;
e7b2c406	333
e7b2c406 ON	334	if (likely(!in_interrupt() && !(task->flags & PF_KTHREAD))) {
	335	init_task_work(&file->f_u.fu_rcuhead, ____fput);
	336	if (!task_work_add(task, &file->f_u.fu_rcuhead, true))
	337	return;
64372501 AM	338	/*
64372501 AM	339	* After this task has run exit_task_work(),
be49b30a	340	* task_work_add() will fail. Fall through to delayed
64372501 AM	341	* fput to avoid leaking *file.
64372501 AM	342	*/
4a9d4b02	343	}
4f5e65a1 ON	344
4f5e65a1 ON	345	if (llist_add(&file->f_u.fu_llist, &delayed_fput_list))
c7314d74	346	schedule_delayed_work(&delayed_fput_work, 1);
4a9d4b02 AV	347	}
	348	}
	349
	350	/*
	351	* synchronous analog of fput(); for kernel threads that might be needed
	352	* in some umount() (and thus can't use flush_delayed_fput() without
	353	* risking deadlocks), need to wait for completion of __fput() and know
	354	* for this specific struct file it won't involve anything that would
	355	* need them. Use only if you really need it - at the very least,
	356	* don't blindly convert fput() by kernel thread to that.
	357	*/
	358	void __fput_sync(struct file *file)
	359	{
	360	if (atomic_long_dec_and_test(&file->f_count)) {
	361	struct task_struct *task = current;
4a9d4b02	362	BUG_ON(!(task->flags & PF_KTHREAD));
d7065da0	363	__fput(file);
4a9d4b02	364	}
d7065da0 AV	365	}
	366
	367	EXPORT_SYMBOL(fput);
	368
4248b0da	369	void __init files_init(void)
b9ea557e	370	{
b6b3fdea	371	filp_cachep = kmem_cache_create("filp", sizeof(struct file), 0,
f3f7c093	372	SLAB_HWCACHE_ALIGN \| SLAB_PANIC \| SLAB_ACCOUNT, NULL);
4248b0da MG	373	percpu_counter_init(&nr_files, 0, GFP_KERNEL);
4248b0da MG	374	}
b6b3fdea	375
4248b0da MG	376	/*
	377	* One file with associated inode and dcache is very roughly 1K. Per default
	378	* do not use more than 10% of our memory for files.
	379	*/
	380	void __init files_maxfiles_init(void)
	381	{
	382	unsigned long n;
	383	unsigned long memreserve = (totalram_pages - nr_free_pages()) * 3/2;
	384
	385	memreserve = min(memreserve, totalram_pages - 1);
	386	n = ((totalram_pages - memreserve) * (PAGE_SIZE / 1024)) / 10;
1da177e4	387
518de9b3	388	files_stat.max_files = max_t(unsigned long, n, NR_FILE);
b9ea557e	389	}