[deliverable/linux.git] / fs / overlayfs / readdir.c

/*
 *
 * Copyright (C) 2011 Novell Inc.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License version 2 as published by
 * the Free Software Foundation.
 */

#include <linux/fs.h>
#include <linux/slab.h>
#include <linux/namei.h>
#include <linux/file.h>
#include <linux/xattr.h>
#include <linux/rbtree.h>
#include <linux/security.h>
#include <linux/cred.h>
#include "overlayfs.h"

struct ovl_cache_entry {
	unsigned int len;
	unsigned int type;
	u64 ino;
	struct list_head l_node;
	struct rb_node node;
	bool is_whiteout;
	char name[];
};

struct ovl_dir_cache {
	long refcount;
	u64 version;
	struct list_head entries;
};

struct ovl_readdir_data {
	struct dir_context ctx;
	bool is_merge;
	struct rb_root root;
	struct list_head *list;
	struct list_head middle;
	struct dentry *dir;
	int count;
	int err;
};

struct ovl_dir_file {
	bool is_real;
	bool is_upper;
	struct ovl_dir_cache *cache;
	struct list_head *cursor;
	struct file *realfile;
	struct file *upperfile;
};

static struct ovl_cache_entry *ovl_cache_entry_from_node(struct rb_node *n)
{
	return container_of(n, struct ovl_cache_entry, node);
}

static struct ovl_cache_entry *ovl_cache_entry_find(struct rb_root *root,
						    const char *name, int len)
{
	struct rb_node *node = root->rb_node;
	int cmp;

	while (node) {
		struct ovl_cache_entry *p = ovl_cache_entry_from_node(node);

		cmp = strncmp(name, p->name, len);
		if (cmp > 0)
			node = p->node.rb_right;
		else if (cmp < 0 || len < p->len)
			node = p->node.rb_left;
		else
			return p;
	}

	return NULL;
}

static struct ovl_cache_entry *ovl_cache_entry_new(struct dentry *dir,
						   const char *name, int len,
						   u64 ino, unsigned int d_type)
{
	struct ovl_cache_entry *p;
	size_t size = offsetof(struct ovl_cache_entry, name[len + 1]);

	p = kmalloc(size, GFP_KERNEL);
	if (!p)
		return NULL;

	memcpy(p->name, name, len);
	p->name[len] = '\0';
	p->len = len;
	p->type = d_type;
	p->ino = ino;
	p->is_whiteout = false;

	if (d_type == DT_CHR) {
		struct dentry *dentry;
		const struct cred *old_cred;
		struct cred *override_cred;

		override_cred = prepare_creds();
		if (!override_cred) {
			kfree(p);
			return NULL;
		}

		/*
		 * CAP_DAC_OVERRIDE for lookup
		 */
		cap_raise(override_cred->cap_effective, CAP_DAC_OVERRIDE);
		old_cred = override_creds(override_cred);

		dentry = lookup_one_len(name, dir, len);
		if (!IS_ERR(dentry)) {
			p->is_whiteout = ovl_is_whiteout(dentry);
			dput(dentry);
		}
		revert_creds(old_cred);
		put_cred(override_cred);
	}
	return p;
}

static int ovl_cache_entry_add_rb(struct ovl_readdir_data *rdd,
				  const char *name, int len, u64 ino,
				  unsigned int d_type)
{
	struct rb_node **newp = &rdd->root.rb_node;
	struct rb_node *parent = NULL;
	struct ovl_cache_entry *p;

	while (*newp) {
		int cmp;
		struct ovl_cache_entry *tmp;

		parent = *newp;
		tmp = ovl_cache_entry_from_node(*newp);
		cmp = strncmp(name, tmp->name, len);
		if (cmp > 0)
			newp = &tmp->node.rb_right;
		else if (cmp < 0 || len < tmp->len)
			newp = &tmp->node.rb_left;
		else
			return 0;
	}

	p = ovl_cache_entry_new(rdd->dir, name, len, ino, d_type);
	if (p == NULL)
		return -ENOMEM;

	list_add_tail(&p->l_node, rdd->list);
	rb_link_node(&p->node, parent, newp);
	rb_insert_color(&p->node, &rdd->root);

	return 0;
}

static int ovl_fill_lower(struct ovl_readdir_data *rdd,
			  const char *name, int namelen,
			  loff_t offset, u64 ino, unsigned int d_type)
{
	struct ovl_cache_entry *p;

	p = ovl_cache_entry_find(&rdd->root, name, namelen);
	if (p) {
		list_move_tail(&p->l_node, &rdd->middle);
	} else {
		p = ovl_cache_entry_new(rdd->dir, name, namelen, ino, d_type);
		if (p == NULL)
			rdd->err = -ENOMEM;
		else
			list_add_tail(&p->l_node, &rdd->middle);
	}

	return rdd->err;
}

void ovl_cache_free(struct list_head *list)
{
	struct ovl_cache_entry *p;
	struct ovl_cache_entry *n;

	list_for_each_entry_safe(p, n, list, l_node)
		kfree(p);

	INIT_LIST_HEAD(list);
}

static void ovl_cache_put(struct ovl_dir_file *od, struct dentry *dentry)
{
	struct ovl_dir_cache *cache = od->cache;

	WARN_ON(cache->refcount <= 0);
	cache->refcount--;
	if (!cache->refcount) {
		if (ovl_dir_cache(dentry) == cache)
			ovl_set_dir_cache(dentry, NULL);

		ovl_cache_free(&cache->entries);
		kfree(cache);
	}
}

static int ovl_fill_merge(struct dir_context *ctx, const char *name,
			  int namelen, loff_t offset, u64 ino,
			  unsigned int d_type)
{
	struct ovl_readdir_data *rdd =
		container_of(ctx, struct ovl_readdir_data, ctx);

	rdd->count++;
	if (!rdd->is_merge)
		return ovl_cache_entry_add_rb(rdd, name, namelen, ino, d_type);
	else
		return ovl_fill_lower(rdd, name, namelen, offset, ino, d_type);
}

static inline int ovl_dir_read(struct path *realpath,
			       struct ovl_readdir_data *rdd)
{
	struct file *realfile;
	int err;

	realfile = ovl_path_open(realpath, O_RDONLY | O_DIRECTORY);
	if (IS_ERR(realfile))
		return PTR_ERR(realfile);

	rdd->dir = realpath->dentry;
	rdd->ctx.pos = 0;
	do {
		rdd->count = 0;
		rdd->err = 0;
		err = iterate_dir(realfile, &rdd->ctx);
		if (err >= 0)
			err = rdd->err;
	} while (!err && rdd->count);
	fput(realfile);

	return err;
}

static void ovl_dir_reset(struct file *file)
{
	struct ovl_dir_file *od = file->private_data;
	struct ovl_dir_cache *cache = od->cache;
	struct dentry *dentry = file->f_path.dentry;
	enum ovl_path_type type = ovl_path_type(dentry);

	if (cache && ovl_dentry_version_get(dentry) != cache->version) {
		ovl_cache_put(od, dentry);
		od->cache = NULL;
		od->cursor = NULL;
	}
	WARN_ON(!od->is_real && !OVL_TYPE_MERGE(type));
	if (od->is_real && OVL_TYPE_MERGE(type))
		od->is_real = false;
}

static int ovl_dir_read_merged(struct dentry *dentry, struct list_head *list)
{
	int err;
	struct path realpath;
	struct ovl_readdir_data rdd = {
		.ctx.actor = ovl_fill_merge,
		.list = list,
		.root = RB_ROOT,
		.is_merge = false,
	};
	int idx, next;

	for (idx = 0; idx != -1; idx = next) {
		next = ovl_path_next(idx, dentry, &realpath);

		if (next != -1) {
			err = ovl_dir_read(&realpath, &rdd);
			if (err)
				break;
		} else {
			/*
			 * Insert lowest layer entries before upper ones, this
			 * allows offsets to be reasonably constant
			 */
			list_add(&rdd.middle, rdd.list);
			rdd.is_merge = true;
			err = ovl_dir_read(&realpath, &rdd);
			list_del(&rdd.middle);
		}
	}
	return err;
}

static void ovl_seek_cursor(struct ovl_dir_file *od, loff_t pos)
{
	struct list_head *p;
	loff_t off = 0;

	list_for_each(p, &od->cache->entries) {
		if (off >= pos)
			break;
		off++;
	}
	/* Cursor is safe since the cache is stable */
	od->cursor = p;
}

static struct ovl_dir_cache *ovl_cache_get(struct dentry *dentry)
{
	int res;
	struct ovl_dir_cache *cache;

	cache = ovl_dir_cache(dentry);
	if (cache && ovl_dentry_version_get(dentry) == cache->version) {
		cache->refcount++;
		return cache;
	}
	ovl_set_dir_cache(dentry, NULL);

	cache = kzalloc(sizeof(struct ovl_dir_cache), GFP_KERNEL);
	if (!cache)
		return ERR_PTR(-ENOMEM);

	cache->refcount = 1;
	INIT_LIST_HEAD(&cache->entries);

	res = ovl_dir_read_merged(dentry, &cache->entries);
	if (res) {
		ovl_cache_free(&cache->entries);
		kfree(cache);
		return ERR_PTR(res);
	}

	cache->version = ovl_dentry_version_get(dentry);
	ovl_set_dir_cache(dentry, cache);

	return cache;
}

static int ovl_iterate(struct file *file, struct dir_context *ctx)
{
	struct ovl_dir_file *od = file->private_data;
	struct dentry *dentry = file->f_path.dentry;
	struct ovl_cache_entry *p;

	if (!ctx->pos)
		ovl_dir_reset(file);

	if (od->is_real)
		return iterate_dir(od->realfile, ctx);

	if (!od->cache) {
		struct ovl_dir_cache *cache;

		cache = ovl_cache_get(dentry);
		if (IS_ERR(cache))
			return PTR_ERR(cache);

		od->cache = cache;
		ovl_seek_cursor(od, ctx->pos);
	}

	while (od->cursor != &od->cache->entries) {
		p = list_entry(od->cursor, struct ovl_cache_entry, l_node);
		if (!p->is_whiteout)
			if (!dir_emit(ctx, p->name, p->len, p->ino, p->type))
				break;
		od->cursor = p->l_node.next;
		ctx->pos++;
	}
	return 0;
}

static loff_t ovl_dir_llseek(struct file *file, loff_t offset, int origin)
{
	loff_t res;
	struct ovl_dir_file *od = file->private_data;

	mutex_lock(&file_inode(file)->i_mutex);
	if (!file->f_pos)
		ovl_dir_reset(file);

	if (od->is_real) {
		res = vfs_llseek(od->realfile, offset, origin);
		file->f_pos = od->realfile->f_pos;
	} else {
		res = -EINVAL;

		switch (origin) {
		case SEEK_CUR:
			offset += file->f_pos;
			break;
		case SEEK_SET:
			break;
		default:
			goto out_unlock;
		}
		if (offset < 0)
			goto out_unlock;

		if (offset != file->f_pos) {
			file->f_pos = offset;
			if (od->cache)
				ovl_seek_cursor(od, offset);
		}
		res = offset;
	}
out_unlock:
	mutex_unlock(&file_inode(file)->i_mutex);

	return res;
}

static int ovl_dir_fsync(struct file *file, loff_t start, loff_t end,
			 int datasync)
{
	struct ovl_dir_file *od = file->private_data;
	struct dentry *dentry = file->f_path.dentry;
	struct file *realfile = od->realfile;

	/*
	 * Need to check if we started out being a lower dir, but got copied up
	 */
	if (!od->is_upper && OVL_TYPE_UPPER(ovl_path_type(dentry))) {
		struct inode *inode = file_inode(file);

		realfile = lockless_dereference(od->upperfile);
		if (!realfile) {
			struct path upperpath;

			ovl_path_upper(dentry, &upperpath);
			realfile = ovl_path_open(&upperpath, O_RDONLY);
			smp_mb__before_spinlock();
			mutex_lock(&inode->i_mutex);
			if (!od->upperfile) {
				if (IS_ERR(realfile)) {
					mutex_unlock(&inode->i_mutex);
					return PTR_ERR(realfile);
				}
				od->upperfile = realfile;
			} else {
				/* somebody has beaten us to it */
				if (!IS_ERR(realfile))
					fput(realfile);
				realfile = od->upperfile;
			}
			mutex_unlock(&inode->i_mutex);
		}
	}

	return vfs_fsync_range(realfile, start, end, datasync);
}

static int ovl_dir_release(struct inode *inode, struct file *file)
{
	struct ovl_dir_file *od = file->private_data;

	if (od->cache) {
		mutex_lock(&inode->i_mutex);
		ovl_cache_put(od, file->f_path.dentry);
		mutex_unlock(&inode->i_mutex);
	}
	fput(od->realfile);
	if (od->upperfile)
		fput(od->upperfile);
	kfree(od);

	return 0;
}

static int ovl_dir_open(struct inode *inode, struct file *file)
{
	struct path realpath;
	struct file *realfile;
	struct ovl_dir_file *od;
	enum ovl_path_type type;

	od = kzalloc(sizeof(struct ovl_dir_file), GFP_KERNEL);
	if (!od)
		return -ENOMEM;

	type = ovl_path_real(file->f_path.dentry, &realpath);
	realfile = ovl_path_open(&realpath, file->f_flags);
	if (IS_ERR(realfile)) {
		kfree(od);
		return PTR_ERR(realfile);
	}
	od->realfile = realfile;
	od->is_real = !OVL_TYPE_MERGE(type);
	od->is_upper = OVL_TYPE_UPPER(type);
	file->private_data = od;

	return 0;
}

const struct file_operations ovl_dir_operations = {
	.read		= generic_read_dir,
	.open		= ovl_dir_open,
	.iterate	= ovl_iterate,
	.llseek		= ovl_dir_llseek,
	.fsync		= ovl_dir_fsync,
	.release	= ovl_dir_release,
};

int ovl_check_empty_dir(struct dentry *dentry, struct list_head *list)
{
	int err;
	struct ovl_cache_entry *p;

	err = ovl_dir_read_merged(dentry, list);
	if (err)
		return err;

	err = 0;

	list_for_each_entry(p, list, l_node) {
		if (p->is_whiteout)
			continue;

		if (p->name[0] == '.') {
			if (p->len == 1)
				continue;
			if (p->len == 2 && p->name[1] == '.')
				continue;
		}
		err = -ENOTEMPTY;
		break;
	}

	return err;
}

void ovl_cleanup_whiteouts(struct dentry *upper, struct list_head *list)
{
	struct ovl_cache_entry *p;

	mutex_lock_nested(&upper->d_inode->i_mutex, I_MUTEX_CHILD);
	list_for_each_entry(p, list, l_node) {
		struct dentry *dentry;

		if (!p->is_whiteout)
			continue;

		dentry = lookup_one_len(p->name, upper, p->len);
		if (IS_ERR(dentry)) {
			pr_err("overlayfs: lookup '%s/%.*s' failed (%i)\n",
			       upper->d_name.name, p->len, p->name,
			       (int) PTR_ERR(dentry));
			continue;
		}
		ovl_cleanup(upper->d_inode, dentry);
		dput(dentry);
	}
	mutex_unlock(&upper->d_inode->i_mutex);
}
Commit	Line	Data
e9be9d5e MS	1	/*
	2	*
	3	* Copyright (C) 2011 Novell Inc.
	4	*
	5	* This program is free software; you can redistribute it and/or modify it
	6	* under the terms of the GNU General Public License version 2 as published by
	7	* the Free Software Foundation.
	8	*/
	9
	10	#include <linux/fs.h>
	11	#include <linux/slab.h>
	12	#include <linux/namei.h>
	13	#include <linux/file.h>
	14	#include <linux/xattr.h>
	15	#include <linux/rbtree.h>
	16	#include <linux/security.h>
	17	#include <linux/cred.h>
	18	#include "overlayfs.h"
	19
	20	struct ovl_cache_entry {
e9be9d5e MS	21	unsigned int len;
	22	unsigned int type;
	23	u64 ino;
e9be9d5e MS	24	struct list_head l_node;
e9be9d5e MS	25	struct rb_node node;
c2096537	26	bool is_whiteout;
68bf8611	27	char name[];
e9be9d5e MS	28	};
	29
	30	struct ovl_dir_cache {
	31	long refcount;
	32	u64 version;
	33	struct list_head entries;
	34	};
	35
	36	struct ovl_readdir_data {
	37	struct dir_context ctx;
	38	bool is_merge;
49be4fb9	39	struct rb_root root;
e9be9d5e	40	struct list_head *list;
db6ec212	41	struct list_head middle;
49c21e1c	42	struct dentry *dir;
e9be9d5e MS	43	int count;
	44	int err;
	45	};
	46
	47	struct ovl_dir_file {
	48	bool is_real;
	49	bool is_upper;
	50	struct ovl_dir_cache *cache;
4330397e	51	struct list_head *cursor;
e9be9d5e MS	52	struct file *realfile;
	53	struct file *upperfile;
	54	};
	55
	56	static struct ovl_cache_entry ovl_cache_entry_from_node(struct rb_node n)
	57	{
	58	return container_of(n, struct ovl_cache_entry, node);
	59	}
	60
	61	static struct ovl_cache_entry ovl_cache_entry_find(struct rb_root root,
	62	const char *name, int len)
	63	{
	64	struct rb_node *node = root->rb_node;
	65	int cmp;
	66
	67	while (node) {
	68	struct ovl_cache_entry *p = ovl_cache_entry_from_node(node);
	69
	70	cmp = strncmp(name, p->name, len);
	71	if (cmp > 0)
	72	node = p->node.rb_right;
	73	else if (cmp < 0 \|\| len < p->len)
	74	node = p->node.rb_left;
	75	else
	76	return p;
	77	}
	78
	79	return NULL;
	80	}
	81
3e01cee3 MS	82	static struct ovl_cache_entry ovl_cache_entry_new(struct dentry dir,
3e01cee3 MS	83	const char *name, int len,
e9be9d5e MS	84	u64 ino, unsigned int d_type)
	85	{
	86	struct ovl_cache_entry *p;
68bf8611	87	size_t size = offsetof(struct ovl_cache_entry, name[len + 1]);
e9be9d5e	88
68bf8611	89	p = kmalloc(size, GFP_KERNEL);
3e01cee3 MS	90	if (!p)
	91	return NULL;
	92
	93	memcpy(p->name, name, len);
	94	p->name[len] = '\0';
	95	p->len = len;
	96	p->type = d_type;
	97	p->ino = ino;
	98	p->is_whiteout = false;
3e01cee3 MS	99
	100	if (d_type == DT_CHR) {
	101	struct dentry *dentry;
	102	const struct cred *old_cred;
	103	struct cred *override_cred;
	104
	105	override_cred = prepare_creds();
	106	if (!override_cred) {
	107	kfree(p);
	108	return NULL;
	109	}
e9be9d5e	110
3e01cee3 MS	111	/*
	112	* CAP_DAC_OVERRIDE for lookup
	113	*/
	114	cap_raise(override_cred->cap_effective, CAP_DAC_OVERRIDE);
	115	old_cred = override_creds(override_cred);
	116
	117	dentry = lookup_one_len(name, dir, len);
	118	if (!IS_ERR(dentry)) {
	119	p->is_whiteout = ovl_is_whiteout(dentry);
	120	dput(dentry);
	121	}
	122	revert_creds(old_cred);
	123	put_cred(override_cred);
	124	}
e9be9d5e MS	125	return p;
	126	}
	127
	128	static int ovl_cache_entry_add_rb(struct ovl_readdir_data *rdd,
	129	const char *name, int len, u64 ino,
	130	unsigned int d_type)
	131	{
49be4fb9	132	struct rb_node **newp = &rdd->root.rb_node;
e9be9d5e MS	133	struct rb_node *parent = NULL;
	134	struct ovl_cache_entry *p;
	135
	136	while (*newp) {
	137	int cmp;
	138	struct ovl_cache_entry *tmp;
	139
	140	parent = *newp;
	141	tmp = ovl_cache_entry_from_node(*newp);
	142	cmp = strncmp(name, tmp->name, len);
	143	if (cmp > 0)
	144	newp = &tmp->node.rb_right;
	145	else if (cmp < 0 \|\| len < tmp->len)
	146	newp = &tmp->node.rb_left;
	147	else
	148	return 0;
	149	}
	150
3e01cee3	151	p = ovl_cache_entry_new(rdd->dir, name, len, ino, d_type);
e9be9d5e MS	152	if (p == NULL)
	153	return -ENOMEM;
	154
	155	list_add_tail(&p->l_node, rdd->list);
	156	rb_link_node(&p->node, parent, newp);
49be4fb9	157	rb_insert_color(&p->node, &rdd->root);
e9be9d5e MS	158
	159	return 0;
	160	}
	161
	162	static int ovl_fill_lower(struct ovl_readdir_data *rdd,
	163	const char *name, int namelen,
	164	loff_t offset, u64 ino, unsigned int d_type)
	165	{
	166	struct ovl_cache_entry *p;
	167
49be4fb9	168	p = ovl_cache_entry_find(&rdd->root, name, namelen);
e9be9d5e	169	if (p) {
db6ec212	170	list_move_tail(&p->l_node, &rdd->middle);
e9be9d5e	171	} else {
3e01cee3	172	p = ovl_cache_entry_new(rdd->dir, name, namelen, ino, d_type);
e9be9d5e MS	173	if (p == NULL)
	174	rdd->err = -ENOMEM;
	175	else
db6ec212	176	list_add_tail(&p->l_node, &rdd->middle);
e9be9d5e MS	177	}
	178
	179	return rdd->err;
	180	}
	181
	182	void ovl_cache_free(struct list_head *list)
	183	{
	184	struct ovl_cache_entry *p;
	185	struct ovl_cache_entry *n;
	186
	187	list_for_each_entry_safe(p, n, list, l_node)
	188	kfree(p);
	189
	190	INIT_LIST_HEAD(list);
	191	}
	192
	193	static void ovl_cache_put(struct ovl_dir_file od, struct dentry dentry)
	194	{
	195	struct ovl_dir_cache *cache = od->cache;
	196
e9be9d5e MS	197	WARN_ON(cache->refcount <= 0);
	198	cache->refcount--;
	199	if (!cache->refcount) {
	200	if (ovl_dir_cache(dentry) == cache)
	201	ovl_set_dir_cache(dentry, NULL);
	202
	203	ovl_cache_free(&cache->entries);
	204	kfree(cache);
	205	}
	206	}
	207
ac7576f4 MS	208	static int ovl_fill_merge(struct dir_context ctx, const char name,
	209	int namelen, loff_t offset, u64 ino,
	210	unsigned int d_type)
e9be9d5e	211	{
ac7576f4 MS	212	struct ovl_readdir_data *rdd =
ac7576f4 MS	213	container_of(ctx, struct ovl_readdir_data, ctx);
e9be9d5e MS	214
	215	rdd->count++;
	216	if (!rdd->is_merge)
	217	return ovl_cache_entry_add_rb(rdd, name, namelen, ino, d_type);
	218	else
	219	return ovl_fill_lower(rdd, name, namelen, offset, ino, d_type);
	220	}
	221
	222	static inline int ovl_dir_read(struct path *realpath,
	223	struct ovl_readdir_data *rdd)
	224	{
	225	struct file *realfile;
	226	int err;
	227
	228	realfile = ovl_path_open(realpath, O_RDONLY \| O_DIRECTORY);
	229	if (IS_ERR(realfile))
	230	return PTR_ERR(realfile);
	231
3e01cee3	232	rdd->dir = realpath->dentry;
e9be9d5e MS	233	rdd->ctx.pos = 0;
	234	do {
	235	rdd->count = 0;
	236	rdd->err = 0;
	237	err = iterate_dir(realfile, &rdd->ctx);
	238	if (err >= 0)
	239	err = rdd->err;
	240	} while (!err && rdd->count);
	241	fput(realfile);
	242
	243	return err;
	244	}
	245
	246	static void ovl_dir_reset(struct file *file)
	247	{
	248	struct ovl_dir_file *od = file->private_data;
	249	struct ovl_dir_cache *cache = od->cache;
	250	struct dentry *dentry = file->f_path.dentry;
	251	enum ovl_path_type type = ovl_path_type(dentry);
	252
	253	if (cache && ovl_dentry_version_get(dentry) != cache->version) {
	254	ovl_cache_put(od, dentry);
	255	od->cache = NULL;
4330397e	256	od->cursor = NULL;
e9be9d5e	257	}
1afaba1e MS	258	WARN_ON(!od->is_real && !OVL_TYPE_MERGE(type));
1afaba1e MS	259	if (od->is_real && OVL_TYPE_MERGE(type))
e9be9d5e MS	260	od->is_real = false;
	261	}
	262
c9f00fdb	263	static int ovl_dir_read_merged(struct dentry dentry, struct list_head list)
e9be9d5e MS	264	{
e9be9d5e MS	265	int err;
9d7459d8	266	struct path realpath;
e9be9d5e MS	267	struct ovl_readdir_data rdd = {
	268	.ctx.actor = ovl_fill_merge,
	269	.list = list,
49be4fb9	270	.root = RB_ROOT,
e9be9d5e MS	271	.is_merge = false,
e9be9d5e MS	272	};
9d7459d8	273	int idx, next;
e9be9d5e	274
9d7459d8 MS	275	for (idx = 0; idx != -1; idx = next) {
9d7459d8 MS	276	next = ovl_path_next(idx, dentry, &realpath);
c9f00fdb	277
9d7459d8	278	if (next != -1) {
9d7459d8	279	err = ovl_dir_read(&realpath, &rdd);
e9be9d5e	280	if (err)
9d7459d8 MS	281	break;
	282	} else {
	283	/*
	284	* Insert lowest layer entries before upper ones, this
	285	* allows offsets to be reasonably constant
	286	*/
	287	list_add(&rdd.middle, rdd.list);
	288	rdd.is_merge = true;
	289	err = ovl_dir_read(&realpath, &rdd);
	290	list_del(&rdd.middle);
e9be9d5e MS	291	}
e9be9d5e MS	292	}
e9be9d5e	293	return err;
e9be9d5e MS	294	}
	295
	296	static void ovl_seek_cursor(struct ovl_dir_file *od, loff_t pos)
	297	{
4330397e	298	struct list_head *p;
e9be9d5e MS	299	loff_t off = 0;
e9be9d5e MS	300
4330397e	301	list_for_each(p, &od->cache->entries) {
e9be9d5e MS	302	if (off >= pos)
	303	break;
	304	off++;
	305	}
4330397e	306	/* Cursor is safe since the cache is stable */
4330397e	307	od->cursor = p;
e9be9d5e MS	308	}
	309
	310	static struct ovl_dir_cache ovl_cache_get(struct dentry dentry)
	311	{
	312	int res;
e9be9d5e MS	313	struct ovl_dir_cache *cache;
	314
	315	cache = ovl_dir_cache(dentry);
	316	if (cache && ovl_dentry_version_get(dentry) == cache->version) {
	317	cache->refcount++;
	318	return cache;
	319	}
	320	ovl_set_dir_cache(dentry, NULL);
	321
	322	cache = kzalloc(sizeof(struct ovl_dir_cache), GFP_KERNEL);
	323	if (!cache)
	324	return ERR_PTR(-ENOMEM);
	325
	326	cache->refcount = 1;
	327	INIT_LIST_HEAD(&cache->entries);
	328
c9f00fdb	329	res = ovl_dir_read_merged(dentry, &cache->entries);
e9be9d5e MS	330	if (res) {
	331	ovl_cache_free(&cache->entries);
	332	kfree(cache);
	333	return ERR_PTR(res);
	334	}
	335
	336	cache->version = ovl_dentry_version_get(dentry);
	337	ovl_set_dir_cache(dentry, cache);
	338
	339	return cache;
	340	}
	341
	342	static int ovl_iterate(struct file file, struct dir_context ctx)
	343	{
	344	struct ovl_dir_file *od = file->private_data;
	345	struct dentry *dentry = file->f_path.dentry;
4330397e	346	struct ovl_cache_entry *p;
e9be9d5e MS	347
	348	if (!ctx->pos)
	349	ovl_dir_reset(file);
	350
	351	if (od->is_real)
	352	return iterate_dir(od->realfile, ctx);
	353
	354	if (!od->cache) {
	355	struct ovl_dir_cache *cache;
	356
	357	cache = ovl_cache_get(dentry);
	358	if (IS_ERR(cache))
	359	return PTR_ERR(cache);
	360
	361	od->cache = cache;
	362	ovl_seek_cursor(od, ctx->pos);
	363	}
	364
4330397e	365	while (od->cursor != &od->cache->entries) {
	366	p = list_entry(od->cursor, struct ovl_cache_entry, l_node);
	367	if (!p->is_whiteout)
	368	if (!dir_emit(ctx, p->name, p->len, p->ino, p->type))
	369	break;
	370	od->cursor = p->l_node.next;
	371	ctx->pos++;
e9be9d5e MS	372	}
	373	return 0;
	374	}
	375
	376	static loff_t ovl_dir_llseek(struct file *file, loff_t offset, int origin)
	377	{
	378	loff_t res;
	379	struct ovl_dir_file *od = file->private_data;
	380
	381	mutex_lock(&file_inode(file)->i_mutex);
	382	if (!file->f_pos)
	383	ovl_dir_reset(file);
	384
	385	if (od->is_real) {
	386	res = vfs_llseek(od->realfile, offset, origin);
	387	file->f_pos = od->realfile->f_pos;
	388	} else {
	389	res = -EINVAL;
	390
	391	switch (origin) {
	392	case SEEK_CUR:
	393	offset += file->f_pos;
	394	break;
	395	case SEEK_SET:
	396	break;
	397	default:
	398	goto out_unlock;
	399	}
	400	if (offset < 0)
	401	goto out_unlock;
	402
	403	if (offset != file->f_pos) {
	404	file->f_pos = offset;
	405	if (od->cache)
	406	ovl_seek_cursor(od, offset);
	407	}
	408	res = offset;
	409	}
	410	out_unlock:
	411	mutex_unlock(&file_inode(file)->i_mutex);
	412
	413	return res;
	414	}
	415
	416	static int ovl_dir_fsync(struct file *file, loff_t start, loff_t end,
	417	int datasync)
	418	{
	419	struct ovl_dir_file *od = file->private_data;
	420	struct dentry *dentry = file->f_path.dentry;
	421	struct file *realfile = od->realfile;
	422
	423	/*
	424	* Need to check if we started out being a lower dir, but got copied up
	425	*/
1afaba1e	426	if (!od->is_upper && OVL_TYPE_UPPER(ovl_path_type(dentry))) {
e9be9d5e MS	427	struct inode *inode = file_inode(file);
e9be9d5e MS	428
7676895f	429	realfile = lockless_dereference(od->upperfile);
e9be9d5e MS	430	if (!realfile) {
	431	struct path upperpath;
	432
	433	ovl_path_upper(dentry, &upperpath);
	434	realfile = ovl_path_open(&upperpath, O_RDONLY);
d45f00ae	435	smp_mb__before_spinlock();
3d268c9b AV	436	mutex_lock(&inode->i_mutex);
	437	if (!od->upperfile) {
	438	if (IS_ERR(realfile)) {
	439	mutex_unlock(&inode->i_mutex);
	440	return PTR_ERR(realfile);
	441	}
	442	od->upperfile = realfile;
	443	} else {
	444	/* somebody has beaten us to it */
	445	if (!IS_ERR(realfile))
	446	fput(realfile);
	447	realfile = od->upperfile;
e9be9d5e	448	}
3d268c9b	449	mutex_unlock(&inode->i_mutex);
e9be9d5e	450	}
e9be9d5e MS	451	}
	452
	453	return vfs_fsync_range(realfile, start, end, datasync);
	454	}
	455
	456	static int ovl_dir_release(struct inode inode, struct file file)
	457	{
	458	struct ovl_dir_file *od = file->private_data;
	459
	460	if (od->cache) {
	461	mutex_lock(&inode->i_mutex);
	462	ovl_cache_put(od, file->f_path.dentry);
	463	mutex_unlock(&inode->i_mutex);
	464	}
	465	fput(od->realfile);
	466	if (od->upperfile)
	467	fput(od->upperfile);
	468	kfree(od);
	469
	470	return 0;
	471	}
	472
	473	static int ovl_dir_open(struct inode inode, struct file file)
	474	{
	475	struct path realpath;
	476	struct file *realfile;
	477	struct ovl_dir_file *od;
	478	enum ovl_path_type type;
	479
	480	od = kzalloc(sizeof(struct ovl_dir_file), GFP_KERNEL);
	481	if (!od)
	482	return -ENOMEM;
	483
	484	type = ovl_path_real(file->f_path.dentry, &realpath);
	485	realfile = ovl_path_open(&realpath, file->f_flags);
	486	if (IS_ERR(realfile)) {
	487	kfree(od);
	488	return PTR_ERR(realfile);
	489	}
e9be9d5e	490	od->realfile = realfile;
1afaba1e MS	491	od->is_real = !OVL_TYPE_MERGE(type);
1afaba1e MS	492	od->is_upper = OVL_TYPE_UPPER(type);
e9be9d5e MS	493	file->private_data = od;
	494
	495	return 0;
	496	}
	497
	498	const struct file_operations ovl_dir_operations = {
	499	.read = generic_read_dir,
	500	.open = ovl_dir_open,
	501	.iterate = ovl_iterate,
	502	.llseek = ovl_dir_llseek,
	503	.fsync = ovl_dir_fsync,
	504	.release = ovl_dir_release,
	505	};
	506
	507	int ovl_check_empty_dir(struct dentry dentry, struct list_head list)
	508	{
	509	int err;
e9be9d5e MS	510	struct ovl_cache_entry *p;
e9be9d5e MS	511
c9f00fdb	512	err = ovl_dir_read_merged(dentry, list);
e9be9d5e MS	513	if (err)
	514	return err;
	515
	516	err = 0;
	517
	518	list_for_each_entry(p, list, l_node) {
	519	if (p->is_whiteout)
	520	continue;
	521
	522	if (p->name[0] == '.') {
	523	if (p->len == 1)
	524	continue;
	525	if (p->len == 2 && p->name[1] == '.')
	526	continue;
	527	}
	528	err = -ENOTEMPTY;
	529	break;
	530	}
	531
	532	return err;
	533	}
	534
	535	void ovl_cleanup_whiteouts(struct dentry upper, struct list_head list)
	536	{
	537	struct ovl_cache_entry *p;
	538
d1b72cc6	539	mutex_lock_nested(&upper->d_inode->i_mutex, I_MUTEX_CHILD);
e9be9d5e MS	540	list_for_each_entry(p, list, l_node) {
	541	struct dentry *dentry;
	542
	543	if (!p->is_whiteout)
	544	continue;
	545
	546	dentry = lookup_one_len(p->name, upper, p->len);
	547	if (IS_ERR(dentry)) {
	548	pr_err("overlayfs: lookup '%s/%.*s' failed (%i)\n",
	549	upper->d_name.name, p->len, p->name,
	550	(int) PTR_ERR(dentry));
	551	continue;
	552	}
	553	ovl_cleanup(upper->d_inode, dentry);
	554	dput(dentry);
	555	}
	556	mutex_unlock(&upper->d_inode->i_mutex);
	557	}