[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;
	bool is_cursor;
	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;
	int count;
	int err;
};

struct ovl_dir_file {
	bool is_real;
	bool is_upper;
	struct ovl_dir_cache *cache;
	struct ovl_cache_entry 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(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) {
		memcpy(p->name, name, len);
		p->name[len] = '\0';
		p->len = len;
		p->type = d_type;
		p->ino = ino;
		p->is_whiteout = false;
	}

	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(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(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;

	list_del(&od->cursor.l_node);
	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(void *buf, const char *name, int namelen,
			  loff_t offset, u64 ino, unsigned int d_type)
{
	struct ovl_readdir_data *rdd = buf;

	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->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;
	}
	WARN_ON(!od->is_real && type != OVL_PATH_MERGE);
	if (od->is_real && type == OVL_PATH_MERGE)
		od->is_real = false;
}

static int ovl_dir_mark_whiteouts(struct dentry *dir,
				  struct ovl_readdir_data *rdd)
{
	struct ovl_cache_entry *p;
	struct dentry *dentry;
	const struct cred *old_cred;
	struct cred *override_cred;

	override_cred = prepare_creds();
	if (!override_cred) {
		ovl_cache_free(rdd->list);
		return -ENOMEM;
	}

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

	mutex_lock(&dir->d_inode->i_mutex);
	list_for_each_entry(p, rdd->list, l_node) {
		if (p->is_cursor)
			continue;

		if (p->type != DT_CHR)
			continue;

		dentry = lookup_one_len(p->name, dir, p->len);
		if (IS_ERR(dentry))
			continue;

		p->is_whiteout = ovl_is_whiteout(dentry);
		dput(dentry);
	}
	mutex_unlock(&dir->d_inode->i_mutex);

	revert_creds(old_cred);
	put_cred(override_cred);

	return 0;
}

static inline int ovl_dir_read_merged(struct path *upperpath,
				      struct path *lowerpath,
				      struct list_head *list)
{
	int err;
	struct ovl_readdir_data rdd = {
		.ctx.actor = ovl_fill_merge,
		.list = list,
		.root = RB_ROOT,
		.is_merge = false,
	};

	if (upperpath->dentry) {
		err = ovl_dir_read(upperpath, &rdd);
		if (err)
			goto out;

		if (lowerpath->dentry) {
			err = ovl_dir_mark_whiteouts(upperpath->dentry, &rdd);
			if (err)
				goto out;
		}
	}
	if (lowerpath->dentry) {
		/*
		 * Insert lowerpath entries before upperpath ones, this allows
		 * offsets to be reasonably constant
		 */
		list_add(&rdd.middle, rdd.list);
		rdd.is_merge = true;
		err = ovl_dir_read(lowerpath, &rdd);
		list_del(&rdd.middle);
	}
out:
	return err;
}

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

	list_for_each_entry(p, &od->cache->entries, l_node) {
		if (p->is_cursor)
			continue;
		if (off >= pos)
			break;
		off++;
	}
	list_move_tail(&od->cursor.l_node, &p->l_node);
}

static struct ovl_dir_cache *ovl_cache_get(struct dentry *dentry)
{
	int res;
	struct path lowerpath;
	struct path upperpath;
	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);

	ovl_path_lower(dentry, &lowerpath);
	ovl_path_upper(dentry, &upperpath);

	res = ovl_dir_read_merged(&upperpath, &lowerpath, &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;

	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.l_node.next != &od->cache->entries) {
		struct ovl_cache_entry *p;

		p = list_entry(od->cursor.l_node.next, struct ovl_cache_entry, l_node);
		/* Skip cursors */
		if (!p->is_cursor) {
			if (!p->is_whiteout) {
				if (!dir_emit(ctx, p->name, p->len, p->ino, p->type))
					break;
			}
			ctx->pos++;
		}
		list_move(&od->cursor.l_node, &p->l_node);
	}
	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_path_type(dentry) == OVL_PATH_MERGE) {
		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);
	}
	INIT_LIST_HEAD(&od->cursor.l_node);
	od->realfile = realfile;
	od->is_real = (type != OVL_PATH_MERGE);
	od->is_upper = (type != OVL_PATH_LOWER);
	od->cursor.is_cursor = true;
	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 path lowerpath;
	struct path upperpath;
	struct ovl_cache_entry *p;

	ovl_path_upper(dentry, &upperpath);
	ovl_path_lower(dentry, &lowerpath);

	err = ovl_dir_read_merged(&upperpath, &lowerpath, 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_PARENT);
	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 MS	26	bool is_whiteout;
c2096537 MS	27	bool is_cursor;
68bf8611	28	char name[];
e9be9d5e MS	29	};
	30
	31	struct ovl_dir_cache {
	32	long refcount;
	33	u64 version;
	34	struct list_head entries;
	35	};
	36
	37	struct ovl_readdir_data {
	38	struct dir_context ctx;
	39	bool is_merge;
49be4fb9	40	struct rb_root root;
e9be9d5e	41	struct list_head *list;
db6ec212	42	struct list_head middle;
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;
	51	struct ovl_cache_entry cursor;
	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
	82	static struct ovl_cache_entry ovl_cache_entry_new(const char name, int len,
	83	u64 ino, unsigned int d_type)
	84	{
	85	struct ovl_cache_entry *p;
68bf8611	86	size_t size = offsetof(struct ovl_cache_entry, name[len + 1]);
e9be9d5e	87
68bf8611	88	p = kmalloc(size, GFP_KERNEL);
e9be9d5e	89	if (p) {
68bf8611 AV	90	memcpy(p->name, name, len);
68bf8611 AV	91	p->name[len] = '\0';
e9be9d5e MS	92	p->len = len;
	93	p->type = d_type;
	94	p->ino = ino;
	95	p->is_whiteout = false;
	96	}
	97
	98	return p;
	99	}
	100
	101	static int ovl_cache_entry_add_rb(struct ovl_readdir_data *rdd,
	102	const char *name, int len, u64 ino,
	103	unsigned int d_type)
	104	{
49be4fb9	105	struct rb_node **newp = &rdd->root.rb_node;
e9be9d5e MS	106	struct rb_node *parent = NULL;
	107	struct ovl_cache_entry *p;
	108
	109	while (*newp) {
	110	int cmp;
	111	struct ovl_cache_entry *tmp;
	112
	113	parent = *newp;
	114	tmp = ovl_cache_entry_from_node(*newp);
	115	cmp = strncmp(name, tmp->name, len);
	116	if (cmp > 0)
	117	newp = &tmp->node.rb_right;
	118	else if (cmp < 0 \|\| len < tmp->len)
	119	newp = &tmp->node.rb_left;
	120	else
	121	return 0;
	122	}
	123
	124	p = ovl_cache_entry_new(name, len, ino, d_type);
	125	if (p == NULL)
	126	return -ENOMEM;
	127
	128	list_add_tail(&p->l_node, rdd->list);
	129	rb_link_node(&p->node, parent, newp);
49be4fb9	130	rb_insert_color(&p->node, &rdd->root);
e9be9d5e MS	131
	132	return 0;
	133	}
	134
	135	static int ovl_fill_lower(struct ovl_readdir_data *rdd,
	136	const char *name, int namelen,
	137	loff_t offset, u64 ino, unsigned int d_type)
	138	{
	139	struct ovl_cache_entry *p;
	140
49be4fb9	141	p = ovl_cache_entry_find(&rdd->root, name, namelen);
e9be9d5e	142	if (p) {
db6ec212	143	list_move_tail(&p->l_node, &rdd->middle);
e9be9d5e MS	144	} else {
	145	p = ovl_cache_entry_new(name, namelen, ino, d_type);
	146	if (p == NULL)
	147	rdd->err = -ENOMEM;
	148	else
db6ec212	149	list_add_tail(&p->l_node, &rdd->middle);
e9be9d5e MS	150	}
	151
	152	return rdd->err;
	153	}
	154
	155	void ovl_cache_free(struct list_head *list)
	156	{
	157	struct ovl_cache_entry *p;
	158	struct ovl_cache_entry *n;
	159
	160	list_for_each_entry_safe(p, n, list, l_node)
	161	kfree(p);
	162
	163	INIT_LIST_HEAD(list);
	164	}
	165
	166	static void ovl_cache_put(struct ovl_dir_file od, struct dentry dentry)
	167	{
	168	struct ovl_dir_cache *cache = od->cache;
	169
	170	list_del(&od->cursor.l_node);
	171	WARN_ON(cache->refcount <= 0);
	172	cache->refcount--;
	173	if (!cache->refcount) {
	174	if (ovl_dir_cache(dentry) == cache)
	175	ovl_set_dir_cache(dentry, NULL);
	176
	177	ovl_cache_free(&cache->entries);
	178	kfree(cache);
	179	}
	180	}
	181
	182	static int ovl_fill_merge(void buf, const char name, int namelen,
	183	loff_t offset, u64 ino, unsigned int d_type)
	184	{
	185	struct ovl_readdir_data *rdd = buf;
	186
	187	rdd->count++;
	188	if (!rdd->is_merge)
	189	return ovl_cache_entry_add_rb(rdd, name, namelen, ino, d_type);
	190	else
	191	return ovl_fill_lower(rdd, name, namelen, offset, ino, d_type);
	192	}
	193
	194	static inline int ovl_dir_read(struct path *realpath,
	195	struct ovl_readdir_data *rdd)
	196	{
	197	struct file *realfile;
	198	int err;
	199
	200	realfile = ovl_path_open(realpath, O_RDONLY \| O_DIRECTORY);
	201	if (IS_ERR(realfile))
	202	return PTR_ERR(realfile);
	203
	204	rdd->ctx.pos = 0;
	205	do {
	206	rdd->count = 0;
	207	rdd->err = 0;
	208	err = iterate_dir(realfile, &rdd->ctx);
	209	if (err >= 0)
	210	err = rdd->err;
	211	} while (!err && rdd->count);
	212	fput(realfile);
	213
214	return err;
215	}
216
217	static void ovl_dir_reset(struct file *file)
218	{
219	struct ovl_dir_file *od = file->private_data;
220	struct ovl_dir_cache *cache = od->cache;
221	struct dentry *dentry = file->f_path.dentry;
222	enum ovl_path_type type = ovl_path_type(dentry);
223
224	if (cache && ovl_dentry_version_get(dentry) != cache->version) {
225	ovl_cache_put(od, dentry);
226	od->cache = NULL;
227	}
228	WARN_ON(!od->is_real && type != OVL_PATH_MERGE);
229	if (od->is_real && type == OVL_PATH_MERGE)
230	od->is_real = false;
231	}
232
233	static int ovl_dir_mark_whiteouts(struct dentry *dir,
234	struct ovl_readdir_data *rdd)
235	{
236	struct ovl_cache_entry *p;
237	struct dentry *dentry;
238	const struct cred *old_cred;
239	struct cred *override_cred;
240
241	override_cred = prepare_creds();
242	if (!override_cred) {
243	ovl_cache_free(rdd->list);
244	return -ENOMEM;
245	}
246
247	/*
248	* CAP_DAC_OVERRIDE for lookup
249	*/
250	cap_raise(override_cred->cap_effective, CAP_DAC_OVERRIDE);
251	old_cred = override_creds(override_cred);
252
253	mutex_lock(&dir->d_inode->i_mutex);
254	list_for_each_entry(p, rdd->list, l_node) {
c2096537	255	if (p->is_cursor)
e9be9d5e MS	256	continue;
	257
	258	if (p->type != DT_CHR)
	259	continue;
	260
	261	dentry = lookup_one_len(p->name, dir, p->len);
	262	if (IS_ERR(dentry))
	263	continue;
	264
	265	p->is_whiteout = ovl_is_whiteout(dentry);
	266	dput(dentry);
	267	}
	268	mutex_unlock(&dir->d_inode->i_mutex);
	269
	270	revert_creds(old_cred);
	271	put_cred(override_cred);
	272
	273	return 0;
	274	}
	275
	276	static inline int ovl_dir_read_merged(struct path *upperpath,
	277	struct path *lowerpath,
	278	struct list_head *list)
	279	{
	280	int err;
e9be9d5e MS	281	struct ovl_readdir_data rdd = {
	282	.ctx.actor = ovl_fill_merge,
	283	.list = list,
49be4fb9	284	.root = RB_ROOT,
e9be9d5e MS	285	.is_merge = false,
	286	};
	287
	288	if (upperpath->dentry) {
	289	err = ovl_dir_read(upperpath, &rdd);
	290	if (err)
	291	goto out;
	292
	293	if (lowerpath->dentry) {
	294	err = ovl_dir_mark_whiteouts(upperpath->dentry, &rdd);
	295	if (err)
	296	goto out;
	297	}
	298	}
	299	if (lowerpath->dentry) {
	300	/*
	301	* Insert lowerpath entries before upperpath ones, this allows
	302	* offsets to be reasonably constant
	303	*/
db6ec212	304	list_add(&rdd.middle, rdd.list);
e9be9d5e MS	305	rdd.is_merge = true;
e9be9d5e MS	306	err = ovl_dir_read(lowerpath, &rdd);
db6ec212	307	list_del(&rdd.middle);
e9be9d5e MS	308	}
	309	out:
	310	return err;
e9be9d5e MS	311	}
	312
	313	static void ovl_seek_cursor(struct ovl_dir_file *od, loff_t pos)
	314	{
	315	struct ovl_cache_entry *p;
	316	loff_t off = 0;
	317
	318	list_for_each_entry(p, &od->cache->entries, l_node) {
c2096537	319	if (p->is_cursor)
e9be9d5e MS	320	continue;
	321	if (off >= pos)
	322	break;
	323	off++;
	324	}
	325	list_move_tail(&od->cursor.l_node, &p->l_node);
	326	}
	327
	328	static struct ovl_dir_cache ovl_cache_get(struct dentry dentry)
	329	{
	330	int res;
	331	struct path lowerpath;
	332	struct path upperpath;
	333	struct ovl_dir_cache *cache;
	334
	335	cache = ovl_dir_cache(dentry);
	336	if (cache && ovl_dentry_version_get(dentry) == cache->version) {
	337	cache->refcount++;
	338	return cache;
	339	}
	340	ovl_set_dir_cache(dentry, NULL);
	341
	342	cache = kzalloc(sizeof(struct ovl_dir_cache), GFP_KERNEL);
	343	if (!cache)
	344	return ERR_PTR(-ENOMEM);
	345
	346	cache->refcount = 1;
	347	INIT_LIST_HEAD(&cache->entries);
	348
	349	ovl_path_lower(dentry, &lowerpath);
	350	ovl_path_upper(dentry, &upperpath);
	351
	352	res = ovl_dir_read_merged(&upperpath, &lowerpath, &cache->entries);
	353	if (res) {
	354	ovl_cache_free(&cache->entries);
	355	kfree(cache);
	356	return ERR_PTR(res);
	357	}
	358
	359	cache->version = ovl_dentry_version_get(dentry);
	360	ovl_set_dir_cache(dentry, cache);
	361
	362	return cache;
	363	}
	364
	365	static int ovl_iterate(struct file file, struct dir_context ctx)
	366	{
	367	struct ovl_dir_file *od = file->private_data;
	368	struct dentry *dentry = file->f_path.dentry;
	369
	370	if (!ctx->pos)
	371	ovl_dir_reset(file);
	372
	373	if (od->is_real)
	374	return iterate_dir(od->realfile, ctx);
	375
	376	if (!od->cache) {
	377	struct ovl_dir_cache *cache;
	378
	379	cache = ovl_cache_get(dentry);
	380	if (IS_ERR(cache))
	381	return PTR_ERR(cache);
	382
	383	od->cache = cache;
384	ovl_seek_cursor(od, ctx->pos);
385	}
386
387	while (od->cursor.l_node.next != &od->cache->entries) {
388	struct ovl_cache_entry *p;
389
390	p = list_entry(od->cursor.l_node.next, struct ovl_cache_entry, l_node);
391	/* Skip cursors */
c2096537	392	if (!p->is_cursor) {
e9be9d5e MS	393	if (!p->is_whiteout) {
	394	if (!dir_emit(ctx, p->name, p->len, p->ino, p->type))
	395	break;
	396	}
	397	ctx->pos++;
	398	}
	399	list_move(&od->cursor.l_node, &p->l_node);
	400	}
	401	return 0;
	402	}
	403
	404	static loff_t ovl_dir_llseek(struct file *file, loff_t offset, int origin)
	405	{
	406	loff_t res;
	407	struct ovl_dir_file *od = file->private_data;
	408
	409	mutex_lock(&file_inode(file)->i_mutex);
	410	if (!file->f_pos)
	411	ovl_dir_reset(file);
	412
	413	if (od->is_real) {
	414	res = vfs_llseek(od->realfile, offset, origin);
	415	file->f_pos = od->realfile->f_pos;
	416	} else {
	417	res = -EINVAL;
	418
	419	switch (origin) {
	420	case SEEK_CUR:
	421	offset += file->f_pos;
	422	break;
	423	case SEEK_SET:
	424	break;
	425	default:
	426	goto out_unlock;
	427	}
	428	if (offset < 0)
	429	goto out_unlock;
	430
	431	if (offset != file->f_pos) {
	432	file->f_pos = offset;
	433	if (od->cache)
	434	ovl_seek_cursor(od, offset);
	435	}
	436	res = offset;
	437	}
	438	out_unlock:
	439	mutex_unlock(&file_inode(file)->i_mutex);
	440
	441	return res;
	442	}
	443
	444	static int ovl_dir_fsync(struct file *file, loff_t start, loff_t end,
	445	int datasync)
	446	{
	447	struct ovl_dir_file *od = file->private_data;
	448	struct dentry *dentry = file->f_path.dentry;
	449	struct file *realfile = od->realfile;
	450
	451	/*
	452	* Need to check if we started out being a lower dir, but got copied up
	453	*/
	454	if (!od->is_upper && ovl_path_type(dentry) == OVL_PATH_MERGE) {
	455	struct inode *inode = file_inode(file);
	456
d45f00ae	457	realfile =lockless_dereference(od->upperfile);
e9be9d5e MS	458	if (!realfile) {
	459	struct path upperpath;
	460
	461	ovl_path_upper(dentry, &upperpath);
	462	realfile = ovl_path_open(&upperpath, O_RDONLY);
d45f00ae	463	smp_mb__before_spinlock();
3d268c9b AV	464	mutex_lock(&inode->i_mutex);
	465	if (!od->upperfile) {
	466	if (IS_ERR(realfile)) {
	467	mutex_unlock(&inode->i_mutex);
	468	return PTR_ERR(realfile);
	469	}
	470	od->upperfile = realfile;
	471	} else {
	472	/* somebody has beaten us to it */
	473	if (!IS_ERR(realfile))
	474	fput(realfile);
	475	realfile = od->upperfile;
e9be9d5e	476	}
3d268c9b	477	mutex_unlock(&inode->i_mutex);
e9be9d5e	478	}
e9be9d5e MS	479	}
	480
	481	return vfs_fsync_range(realfile, start, end, datasync);
	482	}
	483
	484	static int ovl_dir_release(struct inode inode, struct file file)
	485	{
	486	struct ovl_dir_file *od = file->private_data;
	487
	488	if (od->cache) {
	489	mutex_lock(&inode->i_mutex);
	490	ovl_cache_put(od, file->f_path.dentry);
	491	mutex_unlock(&inode->i_mutex);
	492	}
	493	fput(od->realfile);
	494	if (od->upperfile)
	495	fput(od->upperfile);
	496	kfree(od);
	497
	498	return 0;
	499	}
	500
	501	static int ovl_dir_open(struct inode inode, struct file file)
	502	{
	503	struct path realpath;
	504	struct file *realfile;
	505	struct ovl_dir_file *od;
	506	enum ovl_path_type type;
	507
	508	od = kzalloc(sizeof(struct ovl_dir_file), GFP_KERNEL);
	509	if (!od)
	510	return -ENOMEM;
	511
	512	type = ovl_path_real(file->f_path.dentry, &realpath);
	513	realfile = ovl_path_open(&realpath, file->f_flags);
	514	if (IS_ERR(realfile)) {
	515	kfree(od);
	516	return PTR_ERR(realfile);
	517	}
	518	INIT_LIST_HEAD(&od->cursor.l_node);
	519	od->realfile = realfile;
	520	od->is_real = (type != OVL_PATH_MERGE);
	521	od->is_upper = (type != OVL_PATH_LOWER);
c2096537	522	od->cursor.is_cursor = true;
e9be9d5e MS	523	file->private_data = od;
	524
	525	return 0;
	526	}
	527
	528	const struct file_operations ovl_dir_operations = {
	529	.read = generic_read_dir,
	530	.open = ovl_dir_open,
	531	.iterate = ovl_iterate,
	532	.llseek = ovl_dir_llseek,
	533	.fsync = ovl_dir_fsync,
	534	.release = ovl_dir_release,
	535	};
	536
	537	int ovl_check_empty_dir(struct dentry dentry, struct list_head list)
	538	{
	539	int err;
	540	struct path lowerpath;
	541	struct path upperpath;
	542	struct ovl_cache_entry *p;
	543
	544	ovl_path_upper(dentry, &upperpath);
	545	ovl_path_lower(dentry, &lowerpath);
	546
	547	err = ovl_dir_read_merged(&upperpath, &lowerpath, list);
	548	if (err)
	549	return err;
	550
	551	err = 0;
	552
	553	list_for_each_entry(p, list, l_node) {
	554	if (p->is_whiteout)
	555	continue;
	556
	557	if (p->name[0] == '.') {
	558	if (p->len == 1)
	559	continue;
	560	if (p->len == 2 && p->name[1] == '.')
	561	continue;
	562	}
	563	err = -ENOTEMPTY;
	564	break;
	565	}
	566
	567	return err;
	568	}
	569
	570	void ovl_cleanup_whiteouts(struct dentry upper, struct list_head list)
	571	{
	572	struct ovl_cache_entry *p;
	573
	574	mutex_lock_nested(&upper->d_inode->i_mutex, I_MUTEX_PARENT);
	575	list_for_each_entry(p, list, l_node) {
	576	struct dentry *dentry;
	577
	578	if (!p->is_whiteout)
	579	continue;
	580
	581	dentry = lookup_one_len(p->name, upper, p->len);
	582	if (IS_ERR(dentry)) {
	583	pr_err("overlayfs: lookup '%s/%.*s' failed (%i)\n",
	584	upper->d_name.name, p->len, p->name,
	585	(int) PTR_ERR(dentry));
	586	continue;
587	}
588	ovl_cleanup(upper->d_inode, dentry);
589	dput(dentry);
590	}
591	mutex_unlock(&upper->d_inode->i_mutex);
592	}