[linux-2.6-block.git] / fs / logfs / file.c

/*
 * fs/logfs/file.c	- prepare_write, commit_write and friends
 *
 * As should be obvious for Linux kernel code, license is GPLv2
 *
 * Copyright (c) 2005-2008 Joern Engel <joern@logfs.org>
 */
#include "logfs.h"
#include <linux/sched.h>
#include <linux/writeback.h>

static int logfs_write_begin(struct file *file, struct address_space *mapping,
		loff_t pos, unsigned len, unsigned flags,
		struct page **pagep, void **fsdata)
{
	struct inode *inode = mapping->host;
	struct page *page;
	pgoff_t index = pos >> PAGE_SHIFT;

	page = grab_cache_page_write_begin(mapping, index, flags);
	if (!page)
		return -ENOMEM;
	*pagep = page;

	if ((len == PAGE_SIZE) || PageUptodate(page))
		return 0;
	if ((pos & PAGE_MASK) >= i_size_read(inode)) {
		unsigned start = pos & (PAGE_SIZE - 1);
		unsigned end = start + len;

		/* Reading beyond i_size is simple: memset to zero */
		zero_user_segments(page, 0, start, end, PAGE_SIZE);
		return 0;
	}
	return logfs_readpage_nolock(page);
}

static int logfs_write_end(struct file *file, struct address_space *mapping,
		loff_t pos, unsigned len, unsigned copied, struct page *page,
		void *fsdata)
{
	struct inode *inode = mapping->host;
	pgoff_t index = page->index;
	unsigned start = pos & (PAGE_SIZE - 1);
	unsigned end = start + copied;
	int ret = 0;

	BUG_ON(PAGE_SIZE != inode->i_sb->s_blocksize);
	BUG_ON(page->index > I3_BLOCKS);

	if (copied < len) {
		/*
		 * Short write of a non-initialized paged.  Just tell userspace
		 * to retry the entire page.
		 */
		if (!PageUptodate(page)) {
			copied = 0;
			goto out;
		}
	}
	if (copied == 0)
		goto out; /* FIXME: do we need to update inode? */

	if (i_size_read(inode) < (index << PAGE_SHIFT) + end) {
		i_size_write(inode, (index << PAGE_SHIFT) + end);
		mark_inode_dirty_sync(inode);
	}

	SetPageUptodate(page);
	if (!PageDirty(page)) {
		if (!get_page_reserve(inode, page))
			__set_page_dirty_nobuffers(page);
		else
			ret = logfs_write_buf(inode, page, WF_LOCK);
	}
out:
	unlock_page(page);
	put_page(page);
	return ret ? ret : copied;
}

int logfs_readpage(struct file *file, struct page *page)
{
	int ret;

	ret = logfs_readpage_nolock(page);
	unlock_page(page);
	return ret;
}

/* Clear the page's dirty flag in the radix tree. */
/* TODO: mucking with PageWriteback is silly.  Add a generic function to clear
 * the dirty bit from the radix tree for filesystems that don't have to wait
 * for page writeback to finish (i.e. any compressing filesystem).
 */
static void clear_radix_tree_dirty(struct page *page)
{
	BUG_ON(PagePrivate(page) || page->private);
	set_page_writeback(page);
	end_page_writeback(page);
}

static int __logfs_writepage(struct page *page)
{
	struct inode *inode = page->mapping->host;
	int err;

	err = logfs_write_buf(inode, page, WF_LOCK);
	if (err)
		set_page_dirty(page);
	else
		clear_radix_tree_dirty(page);
	unlock_page(page);
	return err;
}

static int logfs_writepage(struct page *page, struct writeback_control *wbc)
{
	struct inode *inode = page->mapping->host;
	loff_t i_size = i_size_read(inode);
	pgoff_t end_index = i_size >> PAGE_SHIFT;
	unsigned offset;
	u64 bix;
	level_t level;

	log_file("logfs_writepage(%lx, %lx, %p)\n", inode->i_ino, page->index,
			page);

	logfs_unpack_index(page->index, &bix, &level);

	/* Indirect blocks are never truncated */
	if (level != 0)
		return __logfs_writepage(page);

	/*
	 * TODO: everything below is a near-verbatim copy of nobh_writepage().
	 * The relevant bits should be factored out after logfs is merged.
	 */

	/* Is the page fully inside i_size? */
	if (bix < end_index)
		return __logfs_writepage(page);

	 /* Is the page fully outside i_size? (truncate in progress) */
	offset = i_size & (PAGE_SIZE-1);
	if (bix > end_index || offset == 0) {
		unlock_page(page);
		return 0; /* don't care */
	}

	/*
	 * The page straddles i_size.  It must be zeroed out on each and every
	 * writepage invokation because it may be mmapped.  "A file is mapped
	 * in multiples of the page size.  For a file that is not a multiple of
	 * the  page size, the remaining memory is zeroed when mapped, and
	 * writes to that region are not written out to the file."
	 */
	zero_user_segment(page, offset, PAGE_SIZE);
	return __logfs_writepage(page);
}

static void logfs_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
{
	struct logfs_block *block = logfs_block(page);

	if (block->reserved_bytes) {
		struct super_block *sb = page->mapping->host->i_sb;
		struct logfs_super *super = logfs_super(sb);

		super->s_dirty_pages -= block->reserved_bytes;
		block->ops->free_block(sb, block);
		BUG_ON(bitmap_weight(block->alias_map, LOGFS_BLOCK_FACTOR));
	} else
		move_page_to_btree(page);
	BUG_ON(PagePrivate(page) || page->private);
}

static int logfs_releasepage(struct page *page, gfp_t only_xfs_uses_this)
{
	return 0; /* None of these are easy to release */
}


long logfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
	struct inode *inode = file_inode(file);
	struct logfs_inode *li = logfs_inode(inode);
	unsigned int oldflags, flags;
	int err;

	switch (cmd) {
	case FS_IOC_GETFLAGS:
		flags = li->li_flags & LOGFS_FL_USER_VISIBLE;
		return put_user(flags, (int __user *)arg);
	case FS_IOC_SETFLAGS:
		if (IS_RDONLY(inode))
			return -EROFS;

		if (!inode_owner_or_capable(inode))
			return -EACCES;

		err = get_user(flags, (int __user *)arg);
		if (err)
			return err;

		inode_lock(inode);
		oldflags = li->li_flags;
		flags &= LOGFS_FL_USER_MODIFIABLE;
		flags |= oldflags & ~LOGFS_FL_USER_MODIFIABLE;
		li->li_flags = flags;
		inode_unlock(inode);

		inode->i_ctime = CURRENT_TIME;
		mark_inode_dirty_sync(inode);
		return 0;

	default:
		return -ENOTTY;
	}
}

int logfs_fsync(struct file *file, loff_t start, loff_t end, int datasync)
{
	struct super_block *sb = file->f_mapping->host->i_sb;
	struct inode *inode = file->f_mapping->host;
	int ret;

	ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
	if (ret)
		return ret;

	inode_lock(inode);
	logfs_get_wblocks(sb, NULL, WF_LOCK);
	logfs_write_anchor(sb);
	logfs_put_wblocks(sb, NULL, WF_LOCK);
	inode_unlock(inode);

	return 0;
}

static int logfs_setattr(struct dentry *dentry, struct iattr *attr)
{
	struct inode *inode = d_inode(dentry);
	int err = 0;

	err = inode_change_ok(inode, attr);
	if (err)
		return err;

	if (attr->ia_valid & ATTR_SIZE) {
		err = logfs_truncate(inode, attr->ia_size);
		if (err)
			return err;
	}

	setattr_copy(inode, attr);
	mark_inode_dirty(inode);
	return 0;
}

const struct inode_operations logfs_reg_iops = {
	.setattr	= logfs_setattr,
};

const struct file_operations logfs_reg_fops = {
	.read_iter	= generic_file_read_iter,
	.write_iter	= generic_file_write_iter,
	.fsync		= logfs_fsync,
	.unlocked_ioctl	= logfs_ioctl,
	.llseek		= generic_file_llseek,
	.mmap		= generic_file_readonly_mmap,
	.open		= generic_file_open,
};

const struct address_space_operations logfs_reg_aops = {
	.invalidatepage	= logfs_invalidatepage,
	.readpage	= logfs_readpage,
	.releasepage	= logfs_releasepage,
	.set_page_dirty	= __set_page_dirty_nobuffers,
	.writepage	= logfs_writepage,
	.writepages	= generic_writepages,
	.write_begin	= logfs_write_begin,
	.write_end	= logfs_write_end,
};
Commit	Line	Data
5db53f3e JE	1	/*
	2	* fs/logfs/file.c - prepare_write, commit_write and friends
	3	*
	4	* As should be obvious for Linux kernel code, license is GPLv2
	5	*
	6	* Copyright (c) 2005-2008 Joern Engel <joern@logfs.org>
	7	*/
	8	#include "logfs.h"
	9	#include <linux/sched.h>
	10	#include <linux/writeback.h>
	11
	12	static int logfs_write_begin(struct file file, struct address_space mapping,
	13	loff_t pos, unsigned len, unsigned flags,
	14	struct page pagep, void fsdata)
	15	{
	16	struct inode *inode = mapping->host;
	17	struct page *page;
09cbfeaf	18	pgoff_t index = pos >> PAGE_SHIFT;
5db53f3e JE	19
	20	page = grab_cache_page_write_begin(mapping, index, flags);
	21	if (!page)
	22	return -ENOMEM;
	23	*pagep = page;
	24
09cbfeaf	25	if ((len == PAGE_SIZE) \|\| PageUptodate(page))
5db53f3e	26	return 0;
09cbfeaf KS	27	if ((pos & PAGE_MASK) >= i_size_read(inode)) {
09cbfeaf KS	28	unsigned start = pos & (PAGE_SIZE - 1);
5db53f3e JE	29	unsigned end = start + len;
	30
	31	/* Reading beyond i_size is simple: memset to zero */
09cbfeaf	32	zero_user_segments(page, 0, start, end, PAGE_SIZE);
5db53f3e JE	33	return 0;
	34	}
	35	return logfs_readpage_nolock(page);
	36	}
	37
	38	static int logfs_write_end(struct file file, struct address_space mapping,
	39	loff_t pos, unsigned len, unsigned copied, struct page *page,
	40	void *fsdata)
	41	{
	42	struct inode *inode = mapping->host;
	43	pgoff_t index = page->index;
09cbfeaf	44	unsigned start = pos & (PAGE_SIZE - 1);
5db53f3e JE	45	unsigned end = start + copied;
	46	int ret = 0;
	47
09cbfeaf	48	BUG_ON(PAGE_SIZE != inode->i_sb->s_blocksize);
5db53f3e JE	49	BUG_ON(page->index > I3_BLOCKS);
	50
	51	if (copied < len) {
	52	/*
	53	* Short write of a non-initialized paged. Just tell userspace
	54	* to retry the entire page.
	55	*/
	56	if (!PageUptodate(page)) {
	57	copied = 0;
	58	goto out;
	59	}
	60	}
	61	if (copied == 0)
	62	goto out; /* FIXME: do we need to update inode? */
	63
09cbfeaf KS	64	if (i_size_read(inode) < (index << PAGE_SHIFT) + end) {
09cbfeaf KS	65	i_size_write(inode, (index << PAGE_SHIFT) + end);
5db53f3e JE	66	mark_inode_dirty_sync(inode);
	67	}
	68
	69	SetPageUptodate(page);
	70	if (!PageDirty(page)) {
	71	if (!get_page_reserve(inode, page))
	72	__set_page_dirty_nobuffers(page);
	73	else
	74	ret = logfs_write_buf(inode, page, WF_LOCK);
	75	}
	76	out:
	77	unlock_page(page);
09cbfeaf	78	put_page(page);
5db53f3e JE	79	return ret ? ret : copied;
	80	}
	81
	82	int logfs_readpage(struct file file, struct page page)
	83	{
	84	int ret;
	85
	86	ret = logfs_readpage_nolock(page);
	87	unlock_page(page);
	88	return ret;
	89	}
	90
	91	/* Clear the page's dirty flag in the radix tree. */
	92	/* TODO: mucking with PageWriteback is silly. Add a generic function to clear
	93	* the dirty bit from the radix tree for filesystems that don't have to wait
	94	* for page writeback to finish (i.e. any compressing filesystem).
	95	*/
	96	static void clear_radix_tree_dirty(struct page *page)
	97	{
	98	BUG_ON(PagePrivate(page) \|\| page->private);
	99	set_page_writeback(page);
	100	end_page_writeback(page);
	101	}
	102
	103	static int __logfs_writepage(struct page *page)
	104	{
	105	struct inode *inode = page->mapping->host;
	106	int err;
	107
	108	err = logfs_write_buf(inode, page, WF_LOCK);
	109	if (err)
	110	set_page_dirty(page);
	111	else
	112	clear_radix_tree_dirty(page);
	113	unlock_page(page);
	114	return err;
	115	}
	116
	117	static int logfs_writepage(struct page page, struct writeback_control wbc)
	118	{
	119	struct inode *inode = page->mapping->host;
	120	loff_t i_size = i_size_read(inode);
09cbfeaf	121	pgoff_t end_index = i_size >> PAGE_SHIFT;
5db53f3e JE	122	unsigned offset;
	123	u64 bix;
	124	level_t level;
	125
	126	log_file("logfs_writepage(%lx, %lx, %p)\n", inode->i_ino, page->index,
	127	page);
	128
	129	logfs_unpack_index(page->index, &bix, &level);
	130
	131	/* Indirect blocks are never truncated */
	132	if (level != 0)
	133	return __logfs_writepage(page);
	134
	135	/*
	136	* TODO: everything below is a near-verbatim copy of nobh_writepage().
	137	* The relevant bits should be factored out after logfs is merged.
	138	*/
	139
	140	/* Is the page fully inside i_size? */
	141	if (bix < end_index)
	142	return __logfs_writepage(page);
	143
	144	/* Is the page fully outside i_size? (truncate in progress) */
09cbfeaf	145	offset = i_size & (PAGE_SIZE-1);
5db53f3e JE	146	if (bix > end_index \|\| offset == 0) {
	147	unlock_page(page);
	148	return 0; /* don't care */
	149	}
	150
	151	/*
	152	* The page straddles i_size. It must be zeroed out on each and every
	153	* writepage invokation because it may be mmapped. "A file is mapped
	154	* in multiples of the page size. For a file that is not a multiple of
	155	* the page size, the remaining memory is zeroed when mapped, and
	156	* writes to that region are not written out to the file."
	157	*/
09cbfeaf	158	zero_user_segment(page, offset, PAGE_SIZE);
5db53f3e JE	159	return __logfs_writepage(page);
	160	}
	161
d47992f8 LC	162	static void logfs_invalidatepage(struct page *page, unsigned int offset,
d47992f8 LC	163	unsigned int length)
5db53f3e	164	{
05ebad85 JE	165	struct logfs_block *block = logfs_block(page);
	166
	167	if (block->reserved_bytes) {
	168	struct super_block *sb = page->mapping->host->i_sb;
	169	struct logfs_super *super = logfs_super(sb);
	170
	171	super->s_dirty_pages -= block->reserved_bytes;
	172	block->ops->free_block(sb, block);
	173	BUG_ON(bitmap_weight(block->alias_map, LOGFS_BLOCK_FACTOR));
	174	} else
	175	move_page_to_btree(page);
5db53f3e JE	176	BUG_ON(PagePrivate(page) \|\| page->private);
	177	}
	178
	179	static int logfs_releasepage(struct page *page, gfp_t only_xfs_uses_this)
	180	{
	181	return 0; /* None of these are easy to release */
	182	}
	183
	184
02d6d685	185	long logfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
5db53f3e	186	{
496ad9aa	187	struct inode *inode = file_inode(file);
5db53f3e JE	188	struct logfs_inode *li = logfs_inode(inode);
	189	unsigned int oldflags, flags;
	190	int err;
	191
	192	switch (cmd) {
	193	case FS_IOC_GETFLAGS:
	194	flags = li->li_flags & LOGFS_FL_USER_VISIBLE;
	195	return put_user(flags, (int __user *)arg);
	196	case FS_IOC_SETFLAGS:
	197	if (IS_RDONLY(inode))
	198	return -EROFS;
	199
2e149670	200	if (!inode_owner_or_capable(inode))
5db53f3e JE	201	return -EACCES;
	202
	203	err = get_user(flags, (int __user *)arg);
	204	if (err)
	205	return err;
	206
5955102c	207	inode_lock(inode);
5db53f3e JE	208	oldflags = li->li_flags;
	209	flags &= LOGFS_FL_USER_MODIFIABLE;
	210	flags \|= oldflags & ~LOGFS_FL_USER_MODIFIABLE;
	211	li->li_flags = flags;
5955102c	212	inode_unlock(inode);
5db53f3e JE	213
	214	inode->i_ctime = CURRENT_TIME;
	215	mark_inode_dirty_sync(inode);
	216	return 0;
	217
	218	default:
	219	return -ENOTTY;
	220	}
	221	}
	222
02c24a82	223	int logfs_fsync(struct file *file, loff_t start, loff_t end, int datasync)
5db53f3e	224	{
7ea80859	225	struct super_block *sb = file->f_mapping->host->i_sb;
02c24a82 JB	226	struct inode *inode = file->f_mapping->host;
	227	int ret;
	228
	229	ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
	230	if (ret)
	231	return ret;
5db53f3e	232
5955102c	233	inode_lock(inode);
13ced29c	234	logfs_get_wblocks(sb, NULL, WF_LOCK);
c0c79c31	235	logfs_write_anchor(sb);
13ced29c	236	logfs_put_wblocks(sb, NULL, WF_LOCK);
5955102c	237	inode_unlock(inode);
02c24a82	238
5db53f3e JE	239	return 0;
	240	}
	241
	242	static int logfs_setattr(struct dentry dentry, struct iattr attr)
	243	{
2b0143b5	244	struct inode *inode = d_inode(dentry);
5db53f3e JE	245	int err = 0;
5db53f3e JE	246
db78b877 CH	247	err = inode_change_ok(inode, attr);
	248	if (err)
	249	return err;
	250
1025774c	251	if (attr->ia_valid & ATTR_SIZE) {
5db53f3e	252	err = logfs_truncate(inode, attr->ia_size);
1025774c CH	253	if (err)
	254	return err;
	255	}
5db53f3e	256
1025774c CH	257	setattr_copy(inode, attr);
	258	mark_inode_dirty(inode);
	259	return 0;
5db53f3e JE	260	}
	261
	262	const struct inode_operations logfs_reg_iops = {
	263	.setattr = logfs_setattr,
	264	};
	265
	266	const struct file_operations logfs_reg_fops = {
aad4f8bb	267	.read_iter = generic_file_read_iter,
8174202b	268	.write_iter = generic_file_write_iter,
5db53f3e	269	.fsync = logfs_fsync,
02d6d685	270	.unlocked_ioctl = logfs_ioctl,
5db53f3e JE	271	.llseek = generic_file_llseek,
	272	.mmap = generic_file_readonly_mmap,
	273	.open = generic_file_open,
5db53f3e JE	274	};
	275
	276	const struct address_space_operations logfs_reg_aops = {
	277	.invalidatepage = logfs_invalidatepage,
	278	.readpage = logfs_readpage,
	279	.releasepage = logfs_releasepage,
	280	.set_page_dirty = __set_page_dirty_nobuffers,
	281	.writepage = logfs_writepage,
	282	.writepages = generic_writepages,
	283	.write_begin = logfs_write_begin,
	284	.write_end = logfs_write_end,
	285	};