[linux.git] / fs / exofs / dir.c

/*
 * Copyright (C) 2005, 2006
 * Avishay Traeger ([email protected])
 * Copyright (C) 2008, 2009
 * Boaz Harrosh <[email protected]>
 *
 * Copyrights for code taken from ext2:
 *     Copyright (C) 1992, 1993, 1994, 1995
 *     Remy Card ([email protected])
 *     Laboratoire MASI - Institut Blaise Pascal
 *     Universite Pierre et Marie Curie (Paris VI)
 *     from
 *     linux/fs/minix/inode.c
 *     Copyright (C) 1991, 1992  Linus Torvalds
 *
 * This file is part of exofs.
 *
 * exofs is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation.  Since it is based on ext2, and the only
 * valid version of GPL for the Linux kernel is version 2, the only valid
 * version of GPL for exofs is version 2.
 *
 * exofs is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with exofs; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */

#include "exofs.h"

static inline unsigned exofs_chunk_size(struct inode *inode)
{
	return inode->i_sb->s_blocksize;
}

static inline void exofs_put_page(struct page *page)
{
	kunmap(page);
	page_cache_release(page);
}

/* Accesses dir's inode->i_size must be called under inode lock */
static inline unsigned long dir_pages(struct inode *inode)
{
	return (inode->i_size + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
}

static unsigned exofs_last_byte(struct inode *inode, unsigned long page_nr)
{
	loff_t last_byte = inode->i_size;

	last_byte -= page_nr << PAGE_CACHE_SHIFT;
	if (last_byte > PAGE_CACHE_SIZE)
		last_byte = PAGE_CACHE_SIZE;
	return last_byte;
}

static int exofs_commit_chunk(struct page *page, loff_t pos, unsigned len)
{
	struct address_space *mapping = page->mapping;
	struct inode *dir = mapping->host;
	int err = 0;

	dir->i_version++;

	if (!PageUptodate(page))
		SetPageUptodate(page);

	if (pos+len > dir->i_size) {
		i_size_write(dir, pos+len);
		mark_inode_dirty(dir);
	}
	set_page_dirty(page);

	if (IS_DIRSYNC(dir))
		err = write_one_page(page, 1);
	else
		unlock_page(page);

	return err;
}

static void exofs_check_page(struct page *page)
{
	struct inode *dir = page->mapping->host;
	unsigned chunk_size = exofs_chunk_size(dir);
	char *kaddr = page_address(page);
	unsigned offs, rec_len;
	unsigned limit = PAGE_CACHE_SIZE;
	struct exofs_dir_entry *p;
	char *error;

	/* if the page is the last one in the directory */
	if ((dir->i_size >> PAGE_CACHE_SHIFT) == page->index) {
		limit = dir->i_size & ~PAGE_CACHE_MASK;
		if (limit & (chunk_size - 1))
			goto Ebadsize;
		if (!limit)
			goto out;
	}
	for (offs = 0; offs <= limit - EXOFS_DIR_REC_LEN(1); offs += rec_len) {
		p = (struct exofs_dir_entry *)(kaddr + offs);
		rec_len = le16_to_cpu(p->rec_len);

		if (rec_len < EXOFS_DIR_REC_LEN(1))
			goto Eshort;
		if (rec_len & 3)
			goto Ealign;
		if (rec_len < EXOFS_DIR_REC_LEN(p->name_len))
			goto Enamelen;
		if (((offs + rec_len - 1) ^ offs) & ~(chunk_size-1))
			goto Espan;
	}
	if (offs != limit)
		goto Eend;
out:
	SetPageChecked(page);
	return;

Ebadsize:
	EXOFS_ERR("ERROR [exofs_check_page]: "
		"size of directory(0x%lx) is not a multiple of chunk size\n",
		dir->i_ino
	);
	goto fail;
Eshort:
	error = "rec_len is smaller than minimal";
	goto bad_entry;
Ealign:
	error = "unaligned directory entry";
	goto bad_entry;
Enamelen:
	error = "rec_len is too small for name_len";
	goto bad_entry;
Espan:
	error = "directory entry across blocks";
	goto bad_entry;
bad_entry:
	EXOFS_ERR(
		"ERROR [exofs_check_page]: bad entry in directory(0x%lx): %s - "
		"offset=%lu, inode=0x%llu, rec_len=%d, name_len=%d\n",
		dir->i_ino, error, (page->index<<PAGE_CACHE_SHIFT)+offs,
		_LLU(le64_to_cpu(p->inode_no)),
		rec_len, p->name_len);
	goto fail;
Eend:
	p = (struct exofs_dir_entry *)(kaddr + offs);
	EXOFS_ERR("ERROR [exofs_check_page]: "
		"entry in directory(0x%lx) spans the page boundary"
		"offset=%lu, inode=0x%llx\n",
		dir->i_ino, (page->index<<PAGE_CACHE_SHIFT)+offs,
		_LLU(le64_to_cpu(p->inode_no)));
fail:
	SetPageChecked(page);
	SetPageError(page);
}

static struct page *exofs_get_page(struct inode *dir, unsigned long n)
{
	struct address_space *mapping = dir->i_mapping;
	struct page *page = read_mapping_page(mapping, n, NULL);

	if (!IS_ERR(page)) {
		kmap(page);
		if (!PageChecked(page))
			exofs_check_page(page);
		if (PageError(page))
			goto fail;
	}
	return page;

fail:
	exofs_put_page(page);
	return ERR_PTR(-EIO);
}

static inline int exofs_match(int len, const unsigned char *name,
					struct exofs_dir_entry *de)
{
	if (len != de->name_len)
		return 0;
	if (!de->inode_no)
		return 0;
	return !memcmp(name, de->name, len);
}

static inline
struct exofs_dir_entry *exofs_next_entry(struct exofs_dir_entry *p)
{
	return (struct exofs_dir_entry *)((char *)p + le16_to_cpu(p->rec_len));
}

static inline unsigned
exofs_validate_entry(char *base, unsigned offset, unsigned mask)
{
	struct exofs_dir_entry *de = (struct exofs_dir_entry *)(base + offset);
	struct exofs_dir_entry *p =
			(struct exofs_dir_entry *)(base + (offset&mask));
	while ((char *)p < (char *)de) {
		if (p->rec_len == 0)
			break;
		p = exofs_next_entry(p);
	}
	return (char *)p - base;
}

static unsigned char exofs_filetype_table[EXOFS_FT_MAX] = {
	[EXOFS_FT_UNKNOWN]	= DT_UNKNOWN,
	[EXOFS_FT_REG_FILE]	= DT_REG,
	[EXOFS_FT_DIR]		= DT_DIR,
	[EXOFS_FT_CHRDEV]	= DT_CHR,
	[EXOFS_FT_BLKDEV]	= DT_BLK,
	[EXOFS_FT_FIFO]		= DT_FIFO,
	[EXOFS_FT_SOCK]		= DT_SOCK,
	[EXOFS_FT_SYMLINK]	= DT_LNK,
};

#define S_SHIFT 12
static unsigned char exofs_type_by_mode[S_IFMT >> S_SHIFT] = {
	[S_IFREG >> S_SHIFT]	= EXOFS_FT_REG_FILE,
	[S_IFDIR >> S_SHIFT]	= EXOFS_FT_DIR,
	[S_IFCHR >> S_SHIFT]	= EXOFS_FT_CHRDEV,
	[S_IFBLK >> S_SHIFT]	= EXOFS_FT_BLKDEV,
	[S_IFIFO >> S_SHIFT]	= EXOFS_FT_FIFO,
	[S_IFSOCK >> S_SHIFT]	= EXOFS_FT_SOCK,
	[S_IFLNK >> S_SHIFT]	= EXOFS_FT_SYMLINK,
};

static inline
void exofs_set_de_type(struct exofs_dir_entry *de, struct inode *inode)
{
	umode_t mode = inode->i_mode;
	de->file_type = exofs_type_by_mode[(mode & S_IFMT) >> S_SHIFT];
}

static int
exofs_readdir(struct file *filp, void *dirent, filldir_t filldir)
{
	loff_t pos = filp->f_pos;
	struct inode *inode = file_inode(filp);
	unsigned int offset = pos & ~PAGE_CACHE_MASK;
	unsigned long n = pos >> PAGE_CACHE_SHIFT;
	unsigned long npages = dir_pages(inode);
	unsigned chunk_mask = ~(exofs_chunk_size(inode)-1);
	unsigned char *types = NULL;
	int need_revalidate = (filp->f_version != inode->i_version);

	if (pos > inode->i_size - EXOFS_DIR_REC_LEN(1))
		return 0;

	types = exofs_filetype_table;

	for ( ; n < npages; n++, offset = 0) {
		char *kaddr, *limit;
		struct exofs_dir_entry *de;
		struct page *page = exofs_get_page(inode, n);

		if (IS_ERR(page)) {
			EXOFS_ERR("ERROR: bad page in directory(0x%lx)\n",
				  inode->i_ino);
			filp->f_pos += PAGE_CACHE_SIZE - offset;
			return PTR_ERR(page);
		}
		kaddr = page_address(page);
		if (unlikely(need_revalidate)) {
			if (offset) {
				offset = exofs_validate_entry(kaddr, offset,
								chunk_mask);
				filp->f_pos = (n<<PAGE_CACHE_SHIFT) + offset;
			}
			filp->f_version = inode->i_version;
			need_revalidate = 0;
		}
		de = (struct exofs_dir_entry *)(kaddr + offset);
		limit = kaddr + exofs_last_byte(inode, n) -
							EXOFS_DIR_REC_LEN(1);
		for (; (char *)de <= limit; de = exofs_next_entry(de)) {
			if (de->rec_len == 0) {
				EXOFS_ERR("ERROR: "
				     "zero-length entry in directory(0x%lx)\n",
				     inode->i_ino);
				exofs_put_page(page);
				return -EIO;
			}
			if (de->inode_no) {
				int over;
				unsigned char d_type = DT_UNKNOWN;

				if (types && de->file_type < EXOFS_FT_MAX)
					d_type = types[de->file_type];

				offset = (char *)de - kaddr;
				over = filldir(dirent, de->name, de->name_len,
						(n<<PAGE_CACHE_SHIFT) | offset,
						le64_to_cpu(de->inode_no),
						d_type);
				if (over) {
					exofs_put_page(page);
					return 0;
				}
			}
			filp->f_pos += le16_to_cpu(de->rec_len);
		}
		exofs_put_page(page);
	}

	return 0;
}

struct exofs_dir_entry *exofs_find_entry(struct inode *dir,
			struct dentry *dentry, struct page **res_page)
{
	const unsigned char *name = dentry->d_name.name;
	int namelen = dentry->d_name.len;
	unsigned reclen = EXOFS_DIR_REC_LEN(namelen);
	unsigned long start, n;
	unsigned long npages = dir_pages(dir);
	struct page *page = NULL;
	struct exofs_i_info *oi = exofs_i(dir);
	struct exofs_dir_entry *de;

	if (npages == 0)
		goto out;

	*res_page = NULL;

	start = oi->i_dir_start_lookup;
	if (start >= npages)
		start = 0;
	n = start;
	do {
		char *kaddr;
		page = exofs_get_page(dir, n);
		if (!IS_ERR(page)) {
			kaddr = page_address(page);
			de = (struct exofs_dir_entry *) kaddr;
			kaddr += exofs_last_byte(dir, n) - reclen;
			while ((char *) de <= kaddr) {
				if (de->rec_len == 0) {
					EXOFS_ERR("ERROR: zero-length entry in "
						  "directory(0x%lx)\n",
						  dir->i_ino);
					exofs_put_page(page);
					goto out;
				}
				if (exofs_match(namelen, name, de))
					goto found;
				de = exofs_next_entry(de);
			}
			exofs_put_page(page);
		}
		if (++n >= npages)
			n = 0;
	} while (n != start);
out:
	return NULL;

found:
	*res_page = page;
	oi->i_dir_start_lookup = n;
	return de;
}

struct exofs_dir_entry *exofs_dotdot(struct inode *dir, struct page **p)
{
	struct page *page = exofs_get_page(dir, 0);
	struct exofs_dir_entry *de = NULL;

	if (!IS_ERR(page)) {
		de = exofs_next_entry(
				(struct exofs_dir_entry *)page_address(page));
		*p = page;
	}
	return de;
}

ino_t exofs_parent_ino(struct dentry *child)
{
	struct page *page;
	struct exofs_dir_entry *de;
	ino_t ino;

	de = exofs_dotdot(child->d_inode, &page);
	if (!de)
		return 0;

	ino = le64_to_cpu(de->inode_no);
	exofs_put_page(page);
	return ino;
}

ino_t exofs_inode_by_name(struct inode *dir, struct dentry *dentry)
{
	ino_t res = 0;
	struct exofs_dir_entry *de;
	struct page *page;

	de = exofs_find_entry(dir, dentry, &page);
	if (de) {
		res = le64_to_cpu(de->inode_no);
		exofs_put_page(page);
	}
	return res;
}

int exofs_set_link(struct inode *dir, struct exofs_dir_entry *de,
			struct page *page, struct inode *inode)
{
	loff_t pos = page_offset(page) +
			(char *) de - (char *) page_address(page);
	unsigned len = le16_to_cpu(de->rec_len);
	int err;

	lock_page(page);
	err = exofs_write_begin(NULL, page->mapping, pos, len,
				AOP_FLAG_UNINTERRUPTIBLE, &page, NULL);
	if (err)
		EXOFS_ERR("exofs_set_link: exofs_write_begin FAILED => %d\n",
			  err);

	de->inode_no = cpu_to_le64(inode->i_ino);
	exofs_set_de_type(de, inode);
	if (likely(!err))
		err = exofs_commit_chunk(page, pos, len);
	exofs_put_page(page);
	dir->i_mtime = dir->i_ctime = CURRENT_TIME;
	mark_inode_dirty(dir);
	return err;
}

int exofs_add_link(struct dentry *dentry, struct inode *inode)
{
	struct inode *dir = dentry->d_parent->d_inode;
	const unsigned char *name = dentry->d_name.name;
	int namelen = dentry->d_name.len;
	unsigned chunk_size = exofs_chunk_size(dir);
	unsigned reclen = EXOFS_DIR_REC_LEN(namelen);
	unsigned short rec_len, name_len;
	struct page *page = NULL;
	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
	struct exofs_dir_entry *de;
	unsigned long npages = dir_pages(dir);
	unsigned long n;
	char *kaddr;
	loff_t pos;
	int err;

	for (n = 0; n <= npages; n++) {
		char *dir_end;

		page = exofs_get_page(dir, n);
		err = PTR_ERR(page);
		if (IS_ERR(page))
			goto out;
		lock_page(page);
		kaddr = page_address(page);
		dir_end = kaddr + exofs_last_byte(dir, n);
		de = (struct exofs_dir_entry *)kaddr;
		kaddr += PAGE_CACHE_SIZE - reclen;
		while ((char *)de <= kaddr) {
			if ((char *)de == dir_end) {
				name_len = 0;
				rec_len = chunk_size;
				de->rec_len = cpu_to_le16(chunk_size);
				de->inode_no = 0;
				goto got_it;
			}
			if (de->rec_len == 0) {
				EXOFS_ERR("ERROR: exofs_add_link: "
				      "zero-length entry in directory(0x%lx)\n",
				      inode->i_ino);
				err = -EIO;
				goto out_unlock;
			}
			err = -EEXIST;
			if (exofs_match(namelen, name, de))
				goto out_unlock;
			name_len = EXOFS_DIR_REC_LEN(de->name_len);
			rec_len = le16_to_cpu(de->rec_len);
			if (!de->inode_no && rec_len >= reclen)
				goto got_it;
			if (rec_len >= name_len + reclen)
				goto got_it;
			de = (struct exofs_dir_entry *) ((char *) de + rec_len);
		}
		unlock_page(page);
		exofs_put_page(page);
	}

	EXOFS_ERR("exofs_add_link: BAD dentry=%p or inode=0x%lx\n",
		  dentry, inode->i_ino);
	return -EINVAL;

got_it:
	pos = page_offset(page) +
		(char *)de - (char *)page_address(page);
	err = exofs_write_begin(NULL, page->mapping, pos, rec_len, 0,
							&page, NULL);
	if (err)
		goto out_unlock;
	if (de->inode_no) {
		struct exofs_dir_entry *de1 =
			(struct exofs_dir_entry *)((char *)de + name_len);
		de1->rec_len = cpu_to_le16(rec_len - name_len);
		de->rec_len = cpu_to_le16(name_len);
		de = de1;
	}
	de->name_len = namelen;
	memcpy(de->name, name, namelen);
	de->inode_no = cpu_to_le64(inode->i_ino);
	exofs_set_de_type(de, inode);
	err = exofs_commit_chunk(page, pos, rec_len);
	dir->i_mtime = dir->i_ctime = CURRENT_TIME;
	mark_inode_dirty(dir);
	sbi->s_numfiles++;

out_put:
	exofs_put_page(page);
out:
	return err;
out_unlock:
	unlock_page(page);
	goto out_put;
}

int exofs_delete_entry(struct exofs_dir_entry *dir, struct page *page)
{
	struct address_space *mapping = page->mapping;
	struct inode *inode = mapping->host;
	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
	char *kaddr = page_address(page);
	unsigned from = ((char *)dir - kaddr) & ~(exofs_chunk_size(inode)-1);
	unsigned to = ((char *)dir - kaddr) + le16_to_cpu(dir->rec_len);
	loff_t pos;
	struct exofs_dir_entry *pde = NULL;
	struct exofs_dir_entry *de = (struct exofs_dir_entry *) (kaddr + from);
	int err;

	while (de < dir) {
		if (de->rec_len == 0) {
			EXOFS_ERR("ERROR: exofs_delete_entry:"
				  "zero-length entry in directory(0x%lx)\n",
				  inode->i_ino);
			err = -EIO;
			goto out;
		}
		pde = de;
		de = exofs_next_entry(de);
	}
	if (pde)
		from = (char *)pde - (char *)page_address(page);
	pos = page_offset(page) + from;
	lock_page(page);
	err = exofs_write_begin(NULL, page->mapping, pos, to - from, 0,
							&page, NULL);
	if (err)
		EXOFS_ERR("exofs_delete_entry: exofs_write_begin FAILED => %d\n",
			  err);
	if (pde)
		pde->rec_len = cpu_to_le16(to - from);
	dir->inode_no = 0;
	if (likely(!err))
		err = exofs_commit_chunk(page, pos, to - from);
	inode->i_ctime = inode->i_mtime = CURRENT_TIME;
	mark_inode_dirty(inode);
	sbi->s_numfiles--;
out:
	exofs_put_page(page);
	return err;
}

/* kept aligned on 4 bytes */
#define THIS_DIR ".\0\0"
#define PARENT_DIR "..\0"

int exofs_make_empty(struct inode *inode, struct inode *parent)
{
	struct address_space *mapping = inode->i_mapping;
	struct page *page = grab_cache_page(mapping, 0);
	unsigned chunk_size = exofs_chunk_size(inode);
	struct exofs_dir_entry *de;
	int err;
	void *kaddr;

	if (!page)
		return -ENOMEM;

	err = exofs_write_begin(NULL, page->mapping, 0, chunk_size, 0,
							&page, NULL);
	if (err) {
		unlock_page(page);
		goto fail;
	}

	kaddr = kmap_atomic(page);
	de = (struct exofs_dir_entry *)kaddr;
	de->name_len = 1;
	de->rec_len = cpu_to_le16(EXOFS_DIR_REC_LEN(1));
	memcpy(de->name, THIS_DIR, sizeof(THIS_DIR));
	de->inode_no = cpu_to_le64(inode->i_ino);
	exofs_set_de_type(de, inode);

	de = (struct exofs_dir_entry *)(kaddr + EXOFS_DIR_REC_LEN(1));
	de->name_len = 2;
	de->rec_len = cpu_to_le16(chunk_size - EXOFS_DIR_REC_LEN(1));
	de->inode_no = cpu_to_le64(parent->i_ino);
	memcpy(de->name, PARENT_DIR, sizeof(PARENT_DIR));
	exofs_set_de_type(de, inode);
	kunmap_atomic(kaddr);
	err = exofs_commit_chunk(page, 0, chunk_size);
fail:
	page_cache_release(page);
	return err;
}

int exofs_empty_dir(struct inode *inode)
{
	struct page *page = NULL;
	unsigned long i, npages = dir_pages(inode);

	for (i = 0; i < npages; i++) {
		char *kaddr;
		struct exofs_dir_entry *de;
		page = exofs_get_page(inode, i);

		if (IS_ERR(page))
			continue;

		kaddr = page_address(page);
		de = (struct exofs_dir_entry *)kaddr;
		kaddr += exofs_last_byte(inode, i) - EXOFS_DIR_REC_LEN(1);

		while ((char *)de <= kaddr) {
			if (de->rec_len == 0) {
				EXOFS_ERR("ERROR: exofs_empty_dir: "
					  "zero-length directory entry"
					  "kaddr=%p, de=%p\n", kaddr, de);
				goto not_empty;
			}
			if (de->inode_no != 0) {
				/* check for . and .. */
				if (de->name[0] != '.')
					goto not_empty;
				if (de->name_len > 2)
					goto not_empty;
				if (de->name_len < 2) {
					if (le64_to_cpu(de->inode_no) !=
					    inode->i_ino)
						goto not_empty;
				} else if (de->name[1] != '.')
					goto not_empty;
			}
			de = exofs_next_entry(de);
		}
		exofs_put_page(page);
	}
	return 1;

not_empty:
	exofs_put_page(page);
	return 0;
}

const struct file_operations exofs_dir_operations = {
	.llseek		= generic_file_llseek,
	.read		= generic_read_dir,
	.readdir	= exofs_readdir,
};
Commit	Line	Data
e6af00f1 BH	1	/*
e6af00f1 BH	2	* Copyright (C) 2005, 2006
27d2e149	3	* Avishay Traeger ([email protected])
e6af00f1 BH	4	* Copyright (C) 2008, 2009
	5	* Boaz Harrosh <[email protected]>
	6	*
	7	* Copyrights for code taken from ext2:
	8	* Copyright (C) 1992, 1993, 1994, 1995
	9	* Remy Card ([email protected])
	10	* Laboratoire MASI - Institut Blaise Pascal
	11	* Universite Pierre et Marie Curie (Paris VI)
	12	* from
	13	* linux/fs/minix/inode.c
	14	* Copyright (C) 1991, 1992 Linus Torvalds
	15	*
	16	* This file is part of exofs.
	17	*
	18	* exofs is free software; you can redistribute it and/or modify
	19	* it under the terms of the GNU General Public License as published by
	20	* the Free Software Foundation. Since it is based on ext2, and the only
	21	* valid version of GPL for the Linux kernel is version 2, the only valid
	22	* version of GPL for exofs is version 2.
	23	*
	24	* exofs is distributed in the hope that it will be useful,
	25	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	26	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	27	* GNU General Public License for more details.
	28	*
	29	* You should have received a copy of the GNU General Public License
	30	* along with exofs; if not, write to the Free Software
	31	* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
	32	*/
	33
	34	#include "exofs.h"
	35
	36	static inline unsigned exofs_chunk_size(struct inode *inode)
	37	{
	38	return inode->i_sb->s_blocksize;
	39	}
	40
	41	static inline void exofs_put_page(struct page *page)
	42	{
	43	kunmap(page);
	44	page_cache_release(page);
	45	}
	46
	47	/* Accesses dir's inode->i_size must be called under inode lock */
	48	static inline unsigned long dir_pages(struct inode *inode)
	49	{
	50	return (inode->i_size + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
	51	}
	52
	53	static unsigned exofs_last_byte(struct inode *inode, unsigned long page_nr)
	54	{
	55	loff_t last_byte = inode->i_size;
	56
	57	last_byte -= page_nr << PAGE_CACHE_SHIFT;
	58	if (last_byte > PAGE_CACHE_SIZE)
	59	last_byte = PAGE_CACHE_SIZE;
	60	return last_byte;
	61	}
	62
	63	static int exofs_commit_chunk(struct page *page, loff_t pos, unsigned len)
	64	{
	65	struct address_space *mapping = page->mapping;
	66	struct inode *dir = mapping->host;
	67	int err = 0;
68
69	dir->i_version++;
70
71	if (!PageUptodate(page))
72	SetPageUptodate(page);
73
74	if (pos+len > dir->i_size) {
75	i_size_write(dir, pos+len);
76	mark_inode_dirty(dir);
77	}
78	set_page_dirty(page);
79
80	if (IS_DIRSYNC(dir))
81	err = write_one_page(page, 1);
82	else
83	unlock_page(page);
84
85	return err;
86	}
87
88	static void exofs_check_page(struct page *page)
89	{
90	struct inode *dir = page->mapping->host;
91	unsigned chunk_size = exofs_chunk_size(dir);
92	char *kaddr = page_address(page);
93	unsigned offs, rec_len;
94	unsigned limit = PAGE_CACHE_SIZE;
95	struct exofs_dir_entry *p;
96	char *error;
97
98	/* if the page is the last one in the directory */
99	if ((dir->i_size >> PAGE_CACHE_SHIFT) == page->index) {
100	limit = dir->i_size & ~PAGE_CACHE_MASK;
101	if (limit & (chunk_size - 1))
102	goto Ebadsize;
103	if (!limit)
104	goto out;
105	}
106	for (offs = 0; offs <= limit - EXOFS_DIR_REC_LEN(1); offs += rec_len) {
107	p = (struct exofs_dir_entry *)(kaddr + offs);
108	rec_len = le16_to_cpu(p->rec_len);
109
110	if (rec_len < EXOFS_DIR_REC_LEN(1))
111	goto Eshort;
112	if (rec_len & 3)
113	goto Ealign;
114	if (rec_len < EXOFS_DIR_REC_LEN(p->name_len))
115	goto Enamelen;
116	if (((offs + rec_len - 1) ^ offs) & ~(chunk_size-1))
117	goto Espan;
118	}
119	if (offs != limit)
120	goto Eend;
121	out:
122	SetPageChecked(page);
123	return;
124
125	Ebadsize:
126	EXOFS_ERR("ERROR [exofs_check_page]: "
0a935519	127	"size of directory(0x%lx) is not a multiple of chunk size\n",
e6af00f1 BH	128	dir->i_ino
	129	);
	130	goto fail;
	131	Eshort:
	132	error = "rec_len is smaller than minimal";
	133	goto bad_entry;
	134	Ealign:
	135	error = "unaligned directory entry";
	136	goto bad_entry;
	137	Enamelen:
	138	error = "rec_len is too small for name_len";
	139	goto bad_entry;
	140	Espan:
	141	error = "directory entry across blocks";
	142	goto bad_entry;
	143	bad_entry:
	144	EXOFS_ERR(
0a935519 BH	145	"ERROR [exofs_check_page]: bad entry in directory(0x%lx): %s - "
0a935519 BH	146	"offset=%lu, inode=0x%llu, rec_len=%d, name_len=%d\n",
e6af00f1 BH	147	dir->i_ino, error, (page->index<<PAGE_CACHE_SHIFT)+offs,
	148	_LLU(le64_to_cpu(p->inode_no)),
	149	rec_len, p->name_len);
	150	goto fail;
	151	Eend:
	152	p = (struct exofs_dir_entry *)(kaddr + offs);
	153	EXOFS_ERR("ERROR [exofs_check_page]: "
0a935519 BH	154	"entry in directory(0x%lx) spans the page boundary"
0a935519 BH	155	"offset=%lu, inode=0x%llx\n",
e6af00f1 BH	156	dir->i_ino, (page->index<<PAGE_CACHE_SHIFT)+offs,
	157	_LLU(le64_to_cpu(p->inode_no)));
	158	fail:
	159	SetPageChecked(page);
	160	SetPageError(page);
	161	}
	162
	163	static struct page exofs_get_page(struct inode dir, unsigned long n)
	164	{
	165	struct address_space *mapping = dir->i_mapping;
	166	struct page *page = read_mapping_page(mapping, n, NULL);
	167
	168	if (!IS_ERR(page)) {
	169	kmap(page);
	170	if (!PageChecked(page))
	171	exofs_check_page(page);
	172	if (PageError(page))
	173	goto fail;
	174	}
	175	return page;
	176
	177	fail:
	178	exofs_put_page(page);
	179	return ERR_PTR(-EIO);
	180	}
	181
	182	static inline int exofs_match(int len, const unsigned char *name,
	183	struct exofs_dir_entry *de)
	184	{
	185	if (len != de->name_len)
	186	return 0;
	187	if (!de->inode_no)
	188	return 0;
	189	return !memcmp(name, de->name, len);
	190	}
	191
	192	static inline
	193	struct exofs_dir_entry exofs_next_entry(struct exofs_dir_entry p)
	194	{
	195	return (struct exofs_dir_entry )((char )p + le16_to_cpu(p->rec_len));
	196	}
	197
	198	static inline unsigned
	199	exofs_validate_entry(char *base, unsigned offset, unsigned mask)
	200	{
	201	struct exofs_dir_entry de = (struct exofs_dir_entry )(base + offset);
	202	struct exofs_dir_entry *p =
	203	(struct exofs_dir_entry *)(base + (offset&mask));
	204	while ((char )p < (char )de) {
	205	if (p->rec_len == 0)
	206	break;
	207	p = exofs_next_entry(p);
	208	}
	209	return (char *)p - base;
	210	}
	211
	212	static unsigned char exofs_filetype_table[EXOFS_FT_MAX] = {
	213	[EXOFS_FT_UNKNOWN] = DT_UNKNOWN,
	214	[EXOFS_FT_REG_FILE] = DT_REG,
	215	[EXOFS_FT_DIR] = DT_DIR,
	216	[EXOFS_FT_CHRDEV] = DT_CHR,
	217	[EXOFS_FT_BLKDEV] = DT_BLK,
	218	[EXOFS_FT_FIFO] = DT_FIFO,
	219	[EXOFS_FT_SOCK] = DT_SOCK,
220	[EXOFS_FT_SYMLINK] = DT_LNK,
221	};
222
223	#define S_SHIFT 12
224	static unsigned char exofs_type_by_mode[S_IFMT >> S_SHIFT] = {
225	[S_IFREG >> S_SHIFT] = EXOFS_FT_REG_FILE,
226	[S_IFDIR >> S_SHIFT] = EXOFS_FT_DIR,
227	[S_IFCHR >> S_SHIFT] = EXOFS_FT_CHRDEV,
228	[S_IFBLK >> S_SHIFT] = EXOFS_FT_BLKDEV,
229	[S_IFIFO >> S_SHIFT] = EXOFS_FT_FIFO,
230	[S_IFSOCK >> S_SHIFT] = EXOFS_FT_SOCK,
231	[S_IFLNK >> S_SHIFT] = EXOFS_FT_SYMLINK,
232	};
233
234	static inline
235	void exofs_set_de_type(struct exofs_dir_entry de, struct inode inode)
236	{
bef41c26	237	umode_t mode = inode->i_mode;
e6af00f1 BH	238	de->file_type = exofs_type_by_mode[(mode & S_IFMT) >> S_SHIFT];
	239	}
	240
	241	static int
	242	exofs_readdir(struct file filp, void dirent, filldir_t filldir)
	243	{
	244	loff_t pos = filp->f_pos;
496ad9aa	245	struct inode *inode = file_inode(filp);
e6af00f1 BH	246	unsigned int offset = pos & ~PAGE_CACHE_MASK;
	247	unsigned long n = pos >> PAGE_CACHE_SHIFT;
	248	unsigned long npages = dir_pages(inode);
	249	unsigned chunk_mask = ~(exofs_chunk_size(inode)-1);
	250	unsigned char *types = NULL;
	251	int need_revalidate = (filp->f_version != inode->i_version);
	252
	253	if (pos > inode->i_size - EXOFS_DIR_REC_LEN(1))
	254	return 0;
	255
	256	types = exofs_filetype_table;
	257
	258	for ( ; n < npages; n++, offset = 0) {
	259	char kaddr, limit;
	260	struct exofs_dir_entry *de;
	261	struct page *page = exofs_get_page(inode, n);
	262
	263	if (IS_ERR(page)) {
0a935519 BH	264	EXOFS_ERR("ERROR: bad page in directory(0x%lx)\n",
0a935519 BH	265	inode->i_ino);
e6af00f1 BH	266	filp->f_pos += PAGE_CACHE_SIZE - offset;
	267	return PTR_ERR(page);
	268	}
	269	kaddr = page_address(page);
	270	if (unlikely(need_revalidate)) {
	271	if (offset) {
	272	offset = exofs_validate_entry(kaddr, offset,
	273	chunk_mask);
	274	filp->f_pos = (n<<PAGE_CACHE_SHIFT) + offset;
	275	}
	276	filp->f_version = inode->i_version;
	277	need_revalidate = 0;
	278	}
	279	de = (struct exofs_dir_entry *)(kaddr + offset);
	280	limit = kaddr + exofs_last_byte(inode, n) -
	281	EXOFS_DIR_REC_LEN(1);
	282	for (; (char *)de <= limit; de = exofs_next_entry(de)) {
	283	if (de->rec_len == 0) {
	284	EXOFS_ERR("ERROR: "
0a935519 BH	285	"zero-length entry in directory(0x%lx)\n",
0a935519 BH	286	inode->i_ino);
e6af00f1 BH	287	exofs_put_page(page);
	288	return -EIO;
	289	}
	290	if (de->inode_no) {
	291	int over;
	292	unsigned char d_type = DT_UNKNOWN;
	293
	294	if (types && de->file_type < EXOFS_FT_MAX)
	295	d_type = types[de->file_type];
	296
	297	offset = (char *)de - kaddr;
	298	over = filldir(dirent, de->name, de->name_len,
	299	(n<<PAGE_CACHE_SHIFT) \| offset,
	300	le64_to_cpu(de->inode_no),
	301	d_type);
	302	if (over) {
	303	exofs_put_page(page);
	304	return 0;
	305	}
	306	}
	307	filp->f_pos += le16_to_cpu(de->rec_len);
	308	}
	309	exofs_put_page(page);
	310	}
	311
	312	return 0;
	313	}
	314
	315	struct exofs_dir_entry exofs_find_entry(struct inode dir,
	316	struct dentry dentry, struct page *res_page)
	317	{
	318	const unsigned char *name = dentry->d_name.name;
	319	int namelen = dentry->d_name.len;
	320	unsigned reclen = EXOFS_DIR_REC_LEN(namelen);
	321	unsigned long start, n;
	322	unsigned long npages = dir_pages(dir);
	323	struct page *page = NULL;
	324	struct exofs_i_info *oi = exofs_i(dir);
	325	struct exofs_dir_entry *de;
	326
	327	if (npages == 0)
	328	goto out;
	329
	330	*res_page = NULL;
	331
	332	start = oi->i_dir_start_lookup;
	333	if (start >= npages)
	334	start = 0;
	335	n = start;
	336	do {
	337	char *kaddr;
	338	page = exofs_get_page(dir, n);
	339	if (!IS_ERR(page)) {
	340	kaddr = page_address(page);
	341	de = (struct exofs_dir_entry *) kaddr;
	342	kaddr += exofs_last_byte(dir, n) - reclen;
	343	while ((char *) de <= kaddr) {
	344	if (de->rec_len == 0) {
0a935519 BH	345	EXOFS_ERR("ERROR: zero-length entry in "
	346	"directory(0x%lx)\n",
	347	dir->i_ino);
e6af00f1 BH	348	exofs_put_page(page);
	349	goto out;
	350	}
	351	if (exofs_match(namelen, name, de))
	352	goto found;
	353	de = exofs_next_entry(de);
	354	}
	355	exofs_put_page(page);
	356	}
	357	if (++n >= npages)
	358	n = 0;
	359	} while (n != start);
	360	out:
	361	return NULL;
	362
	363	found:
	364	*res_page = page;
	365	oi->i_dir_start_lookup = n;
	366	return de;
	367	}
	368
	369	struct exofs_dir_entry exofs_dotdot(struct inode dir, struct page **p)
	370	{
	371	struct page *page = exofs_get_page(dir, 0);
	372	struct exofs_dir_entry *de = NULL;
	373
	374	if (!IS_ERR(page)) {
	375	de = exofs_next_entry(
	376	(struct exofs_dir_entry *)page_address(page));
	377	*p = page;
	378	}
	379	return de;
	380	}
	381
8cf74b39 BH	382	ino_t exofs_parent_ino(struct dentry *child)
	383	{
	384	struct page *page;
	385	struct exofs_dir_entry *de;
	386	ino_t ino;
	387
	388	de = exofs_dotdot(child->d_inode, &page);
	389	if (!de)
	390	return 0;
	391
	392	ino = le64_to_cpu(de->inode_no);
	393	exofs_put_page(page);
	394	return ino;
	395	}
	396
e6af00f1 BH	397	ino_t exofs_inode_by_name(struct inode dir, struct dentry dentry)
	398	{
	399	ino_t res = 0;
	400	struct exofs_dir_entry *de;
	401	struct page *page;
	402
	403	de = exofs_find_entry(dir, dentry, &page);
	404	if (de) {
	405	res = le64_to_cpu(de->inode_no);
	406	exofs_put_page(page);
	407	}
	408	return res;
	409	}
	410
	411	int exofs_set_link(struct inode dir, struct exofs_dir_entry de,
	412	struct page page, struct inode inode)
	413	{
	414	loff_t pos = page_offset(page) +
	415	(char ) de - (char ) page_address(page);
	416	unsigned len = le16_to_cpu(de->rec_len);
	417	int err;
	418
	419	lock_page(page);
	420	err = exofs_write_begin(NULL, page->mapping, pos, len,
	421	AOP_FLAG_UNINTERRUPTIBLE, &page, NULL);
	422	if (err)
571f7f46	423	EXOFS_ERR("exofs_set_link: exofs_write_begin FAILED => %d\n",
e6af00f1 BH	424	err);
	425
	426	de->inode_no = cpu_to_le64(inode->i_ino);
	427	exofs_set_de_type(de, inode);
	428	if (likely(!err))
	429	err = exofs_commit_chunk(page, pos, len);
	430	exofs_put_page(page);
	431	dir->i_mtime = dir->i_ctime = CURRENT_TIME;
	432	mark_inode_dirty(dir);
	433	return err;
	434	}
	435
	436	int exofs_add_link(struct dentry dentry, struct inode inode)
	437	{
	438	struct inode *dir = dentry->d_parent->d_inode;
	439	const unsigned char *name = dentry->d_name.name;
	440	int namelen = dentry->d_name.len;
	441	unsigned chunk_size = exofs_chunk_size(dir);
	442	unsigned reclen = EXOFS_DIR_REC_LEN(namelen);
	443	unsigned short rec_len, name_len;
	444	struct page *page = NULL;
	445	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
	446	struct exofs_dir_entry *de;
	447	unsigned long npages = dir_pages(dir);
	448	unsigned long n;
	449	char *kaddr;
	450	loff_t pos;
	451	int err;
	452
	453	for (n = 0; n <= npages; n++) {
	454	char *dir_end;
	455
	456	page = exofs_get_page(dir, n);
	457	err = PTR_ERR(page);
	458	if (IS_ERR(page))
	459	goto out;
	460	lock_page(page);
	461	kaddr = page_address(page);
	462	dir_end = kaddr + exofs_last_byte(dir, n);
	463	de = (struct exofs_dir_entry *)kaddr;
	464	kaddr += PAGE_CACHE_SIZE - reclen;
	465	while ((char *)de <= kaddr) {
	466	if ((char *)de == dir_end) {
	467	name_len = 0;
	468	rec_len = chunk_size;
	469	de->rec_len = cpu_to_le16(chunk_size);
	470	de->inode_no = 0;
	471	goto got_it;
	472	}
	473	if (de->rec_len == 0) {
	474	EXOFS_ERR("ERROR: exofs_add_link: "
0a935519 BH	475	"zero-length entry in directory(0x%lx)\n",
0a935519 BH	476	inode->i_ino);
e6af00f1 BH	477	err = -EIO;
	478	goto out_unlock;
	479	}
	480	err = -EEXIST;
	481	if (exofs_match(namelen, name, de))
	482	goto out_unlock;
	483	name_len = EXOFS_DIR_REC_LEN(de->name_len);
	484	rec_len = le16_to_cpu(de->rec_len);
	485	if (!de->inode_no && rec_len >= reclen)
	486	goto got_it;
	487	if (rec_len >= name_len + reclen)
	488	goto got_it;
	489	de = (struct exofs_dir_entry ) ((char ) de + rec_len);
	490	}
	491	unlock_page(page);
	492	exofs_put_page(page);
	493	}
	494
0a935519 BH	495	EXOFS_ERR("exofs_add_link: BAD dentry=%p or inode=0x%lx\n",
0a935519 BH	496	dentry, inode->i_ino);
e6af00f1 BH	497	return -EINVAL;
	498
	499	got_it:
	500	pos = page_offset(page) +
	501	(char )de - (char )page_address(page);
	502	err = exofs_write_begin(NULL, page->mapping, pos, rec_len, 0,
	503	&page, NULL);
	504	if (err)
	505	goto out_unlock;
	506	if (de->inode_no) {
	507	struct exofs_dir_entry *de1 =
	508	(struct exofs_dir_entry )((char )de + name_len);
	509	de1->rec_len = cpu_to_le16(rec_len - name_len);
	510	de->rec_len = cpu_to_le16(name_len);
	511	de = de1;
	512	}
	513	de->name_len = namelen;
	514	memcpy(de->name, name, namelen);
	515	de->inode_no = cpu_to_le64(inode->i_ino);
	516	exofs_set_de_type(de, inode);
	517	err = exofs_commit_chunk(page, pos, rec_len);
	518	dir->i_mtime = dir->i_ctime = CURRENT_TIME;
	519	mark_inode_dirty(dir);
	520	sbi->s_numfiles++;
	521
	522	out_put:
	523	exofs_put_page(page);
	524	out:
	525	return err;
	526	out_unlock:
	527	unlock_page(page);
	528	goto out_put;
	529	}
	530
	531	int exofs_delete_entry(struct exofs_dir_entry dir, struct page page)
	532	{
	533	struct address_space *mapping = page->mapping;
	534	struct inode *inode = mapping->host;
	535	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
	536	char *kaddr = page_address(page);
	537	unsigned from = ((char *)dir - kaddr) & ~(exofs_chunk_size(inode)-1);
	538	unsigned to = ((char *)dir - kaddr) + le16_to_cpu(dir->rec_len);
	539	loff_t pos;
	540	struct exofs_dir_entry *pde = NULL;
	541	struct exofs_dir_entry de = (struct exofs_dir_entry ) (kaddr + from);
	542	int err;
	543
	544	while (de < dir) {
	545	if (de->rec_len == 0) {
	546	EXOFS_ERR("ERROR: exofs_delete_entry:"
0a935519 BH	547	"zero-length entry in directory(0x%lx)\n",
0a935519 BH	548	inode->i_ino);
e6af00f1 BH	549	err = -EIO;
	550	goto out;
	551	}
	552	pde = de;
	553	de = exofs_next_entry(de);
	554	}
	555	if (pde)
	556	from = (char )pde - (char )page_address(page);
	557	pos = page_offset(page) + from;
	558	lock_page(page);
	559	err = exofs_write_begin(NULL, page->mapping, pos, to - from, 0,
	560	&page, NULL);
	561	if (err)
571f7f46	562	EXOFS_ERR("exofs_delete_entry: exofs_write_begin FAILED => %d\n",
e6af00f1 BH	563	err);
	564	if (pde)
	565	pde->rec_len = cpu_to_le16(to - from);
	566	dir->inode_no = 0;
	567	if (likely(!err))
	568	err = exofs_commit_chunk(page, pos, to - from);
	569	inode->i_ctime = inode->i_mtime = CURRENT_TIME;
	570	mark_inode_dirty(inode);
	571	sbi->s_numfiles--;
	572	out:
	573	exofs_put_page(page);
	574	return err;
	575	}
	576
	577	/* kept aligned on 4 bytes */
	578	#define THIS_DIR ".\0\0"
	579	#define PARENT_DIR "..\0"
	580
	581	int exofs_make_empty(struct inode inode, struct inode parent)
	582	{
	583	struct address_space *mapping = inode->i_mapping;
	584	struct page *page = grab_cache_page(mapping, 0);
	585	unsigned chunk_size = exofs_chunk_size(inode);
	586	struct exofs_dir_entry *de;
	587	int err;
	588	void *kaddr;
	589
	590	if (!page)
	591	return -ENOMEM;
	592
	593	err = exofs_write_begin(NULL, page->mapping, 0, chunk_size, 0,
	594	&page, NULL);
	595	if (err) {
	596	unlock_page(page);
	597	goto fail;
	598	}
	599
bf7014b6	600	kaddr = kmap_atomic(page);
e6af00f1 BH	601	de = (struct exofs_dir_entry *)kaddr;
	602	de->name_len = 1;
	603	de->rec_len = cpu_to_le16(EXOFS_DIR_REC_LEN(1));
	604	memcpy(de->name, THIS_DIR, sizeof(THIS_DIR));
	605	de->inode_no = cpu_to_le64(inode->i_ino);
	606	exofs_set_de_type(de, inode);
	607
	608	de = (struct exofs_dir_entry *)(kaddr + EXOFS_DIR_REC_LEN(1));
	609	de->name_len = 2;
	610	de->rec_len = cpu_to_le16(chunk_size - EXOFS_DIR_REC_LEN(1));
	611	de->inode_no = cpu_to_le64(parent->i_ino);
	612	memcpy(de->name, PARENT_DIR, sizeof(PARENT_DIR));
	613	exofs_set_de_type(de, inode);
bf7014b6	614	kunmap_atomic(kaddr);
e6af00f1 BH	615	err = exofs_commit_chunk(page, 0, chunk_size);
	616	fail:
	617	page_cache_release(page);
	618	return err;
	619	}
	620
	621	int exofs_empty_dir(struct inode *inode)
	622	{
	623	struct page *page = NULL;
	624	unsigned long i, npages = dir_pages(inode);
	625
	626	for (i = 0; i < npages; i++) {
	627	char *kaddr;
	628	struct exofs_dir_entry *de;
	629	page = exofs_get_page(inode, i);
	630
	631	if (IS_ERR(page))
	632	continue;
	633
	634	kaddr = page_address(page);
	635	de = (struct exofs_dir_entry *)kaddr;
	636	kaddr += exofs_last_byte(inode, i) - EXOFS_DIR_REC_LEN(1);
	637
	638	while ((char *)de <= kaddr) {
	639	if (de->rec_len == 0) {
	640	EXOFS_ERR("ERROR: exofs_empty_dir: "
	641	"zero-length directory entry"
	642	"kaddr=%p, de=%p\n", kaddr, de);
	643	goto not_empty;
	644	}
	645	if (de->inode_no != 0) {
	646	/* check for . and .. */
	647	if (de->name[0] != '.')
	648	goto not_empty;
	649	if (de->name_len > 2)
	650	goto not_empty;
	651	if (de->name_len < 2) {
	652	if (le64_to_cpu(de->inode_no) !=
	653	inode->i_ino)
	654	goto not_empty;
	655	} else if (de->name[1] != '.')
	656	goto not_empty;
	657	}
	658	de = exofs_next_entry(de);
	659	}
	660	exofs_put_page(page);
	661	}
	662	return 1;
	663
	664	not_empty:
	665	exofs_put_page(page);
	666	return 0;
	667	}
	668
	669	const struct file_operations exofs_dir_operations = {
	670	.llseek = generic_file_llseek,
	671	.read = generic_read_dir,
	672	.readdir = exofs_readdir,
	673	};