[net-next-2.6.git] / fs / nilfs2 / gcinode.c

/*
 * gcinode.c - dummy inodes to buffer blocks for garbage collection
 *
 * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
 *
 * This program 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; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program 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 this program; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 *
 * Written by Seiji Kihara <kihara@osrg.net>, Amagai Yoshiji <amagai@osrg.net>,
 *            and Ryusuke Konishi <ryusuke@osrg.net>.
 * Revised by Ryusuke Konishi <ryusuke@osrg.net>.
 *
 */
/*
 * This file adds the cache of on-disk blocks to be moved in garbage
 * collection.  The disk blocks are held with dummy inodes (called
 * gcinodes), and this file provides lookup function of the dummy
 * inodes and their buffer read function.
 *
 * Since NILFS2 keeps up multiple checkpoints/snapshots across GC, it
 * has to treat blocks that belong to a same file but have different
 * checkpoint numbers.  To avoid interference among generations, dummy
 * inodes are managed separately from actual inodes, and their lookup
 * function (nilfs_gc_iget) is designed to be specified with a
 * checkpoint number argument as well as an inode number.
 *
 * Buffers and pages held by the dummy inodes will be released each
 * time after they are copied to a new log.  Dirty blocks made on the
 * current generation and the blocks to be moved by GC never overlap
 * because the dirty blocks make a new generation; they rather must be
 * written individually.
 */

#include <linux/buffer_head.h>
#include <linux/mpage.h>
#include <linux/hash.h>
#include <linux/slab.h>
#include <linux/swap.h>
#include "nilfs.h"
#include "btree.h"
#include "btnode.h"
#include "page.h"
#include "mdt.h"
#include "dat.h"
#include "ifile.h"

static const struct address_space_operations def_gcinode_aops = {
	.sync_page		= block_sync_page,
};

/*
 * nilfs_gccache_submit_read_data() - add data buffer and submit read request
 * @inode - gc inode
 * @blkoff - dummy offset treated as the key for the page cache
 * @pbn - physical block number of the block
 * @vbn - virtual block number of the block, 0 for non-virtual block
 * @out_bh - indirect pointer to a buffer_head struct to receive the results
 *
 * Description: nilfs_gccache_submit_read_data() registers the data buffer
 * specified by @pbn to the GC pagecache with the key @blkoff.
 * This function sets @vbn (@pbn if @vbn is zero) in b_blocknr of the buffer.
 *
 * Return Value: On success, 0 is returned. On Error, one of the following
 * negative error code is returned.
 *
 * %-EIO - I/O error.
 *
 * %-ENOMEM - Insufficient amount of memory available.
 *
 * %-ENOENT - The block specified with @pbn does not exist.
 */
int nilfs_gccache_submit_read_data(struct inode *inode, sector_t blkoff,
				   sector_t pbn, __u64 vbn,
				   struct buffer_head **out_bh)
{
	struct buffer_head *bh;
	int err;

	bh = nilfs_grab_buffer(inode, inode->i_mapping, blkoff, 0);
	if (unlikely(!bh))
		return -ENOMEM;

	if (buffer_uptodate(bh))
		goto out;

	if (pbn == 0) {
		struct inode *dat_inode = NILFS_I_NILFS(inode)->ns_dat;
					  /* use original dat, not gc dat. */
		err = nilfs_dat_translate(dat_inode, vbn, &pbn);
		if (unlikely(err)) { /* -EIO, -ENOMEM, -ENOENT */
			brelse(bh);
			goto failed;
		}
	}

	lock_buffer(bh);
	if (buffer_uptodate(bh)) {
		unlock_buffer(bh);
		goto out;
	}

	if (!buffer_mapped(bh)) {
		bh->b_bdev = NILFS_I_NILFS(inode)->ns_bdev;
		set_buffer_mapped(bh);
	}
	bh->b_blocknr = pbn;
	bh->b_end_io = end_buffer_read_sync;
	get_bh(bh);
	submit_bh(READ, bh);
	if (vbn)
		bh->b_blocknr = vbn;
 out:
	err = 0;
	*out_bh = bh;

 failed:
	unlock_page(bh->b_page);
	page_cache_release(bh->b_page);
	return err;
}

/*
 * nilfs_gccache_submit_read_node() - add node buffer and submit read request
 * @inode - gc inode
 * @pbn - physical block number for the block
 * @vbn - virtual block number for the block
 * @out_bh - indirect pointer to a buffer_head struct to receive the results
 *
 * Description: nilfs_gccache_submit_read_node() registers the node buffer
 * specified by @vbn to the GC pagecache.  @pbn can be supplied by the
 * caller to avoid translation of the disk block address.
 *
 * Return Value: On success, 0 is returned. On Error, one of the following
 * negative error code is returned.
 *
 * %-EIO - I/O error.
 *
 * %-ENOMEM - Insufficient amount of memory available.
 */
int nilfs_gccache_submit_read_node(struct inode *inode, sector_t pbn,
				   __u64 vbn, struct buffer_head **out_bh)
{
	int ret;

	ret = nilfs_btnode_submit_block(&NILFS_I(inode)->i_btnode_cache,
					vbn ? : pbn, pbn, READ, out_bh, &pbn);
	if (ret == -EEXIST) /* internal code (cache hit) */
		ret = 0;
	return ret;
}

int nilfs_gccache_wait_and_mark_dirty(struct buffer_head *bh)
{
	wait_on_buffer(bh);
	if (!buffer_uptodate(bh))
		return -EIO;
	if (buffer_dirty(bh))
		return -EEXIST;

	if (buffer_nilfs_node(bh)) {
		if (nilfs_btree_broken_node_block(bh)) {
			clear_buffer_uptodate(bh);
			return -EIO;
		}
		nilfs_btnode_mark_dirty(bh);
	} else {
		nilfs_mdt_mark_buffer_dirty(bh);
	}
	return 0;
}

/*
 * nilfs_init_gccache() - allocate and initialize gc_inode hash table
 * @nilfs - the_nilfs
 *
 * Return Value: On success, 0.
 * On error, a negative error code is returned.
 */
int nilfs_init_gccache(struct the_nilfs *nilfs)
{
	int loop;

	BUG_ON(nilfs->ns_gc_inodes_h);

	INIT_LIST_HEAD(&nilfs->ns_gc_inodes);

	nilfs->ns_gc_inodes_h =
		kmalloc(sizeof(struct hlist_head) * NILFS_GCINODE_HASH_SIZE,
			GFP_NOFS);
	if (nilfs->ns_gc_inodes_h == NULL)
		return -ENOMEM;

	for (loop = 0; loop < NILFS_GCINODE_HASH_SIZE; loop++)
		INIT_HLIST_HEAD(&nilfs->ns_gc_inodes_h[loop]);
	return 0;
}

/*
 * nilfs_destroy_gccache() - free gc_inode hash table
 * @nilfs - the nilfs
 */
void nilfs_destroy_gccache(struct the_nilfs *nilfs)
{
	if (nilfs->ns_gc_inodes_h) {
		nilfs_remove_all_gcinode(nilfs);
		kfree(nilfs->ns_gc_inodes_h);
		nilfs->ns_gc_inodes_h = NULL;
	}
}

static struct inode *alloc_gcinode(struct the_nilfs *nilfs, ino_t ino,
				   __u64 cno)
{
	struct inode *inode;
	struct nilfs_inode_info *ii;

	inode = nilfs_mdt_new_common(nilfs, NULL, ino, GFP_NOFS, 0);
	if (!inode)
		return NULL;

	inode->i_op = NULL;
	inode->i_fop = NULL;
	inode->i_mapping->a_ops = &def_gcinode_aops;

	ii = NILFS_I(inode);
	ii->i_cno = cno;
	ii->i_flags = 0;
	ii->i_state = 1 << NILFS_I_GCINODE;
	ii->i_bh = NULL;
	nilfs_bmap_init_gc(ii->i_bmap);

	return inode;
}

static unsigned long ihash(ino_t ino, __u64 cno)
{
	return hash_long((unsigned long)((ino << 2) + cno),
			 NILFS_GCINODE_HASH_BITS);
}

/*
 * nilfs_gc_iget() - find or create gc inode with specified (ino,cno)
 */
struct inode *nilfs_gc_iget(struct the_nilfs *nilfs, ino_t ino, __u64 cno)
{
	struct hlist_head *head = nilfs->ns_gc_inodes_h + ihash(ino, cno);
	struct hlist_node *node;
	struct inode *inode;

	hlist_for_each_entry(inode, node, head, i_hash) {
		if (inode->i_ino == ino && NILFS_I(inode)->i_cno == cno)
			return inode;
	}

	inode = alloc_gcinode(nilfs, ino, cno);
	if (likely(inode)) {
		hlist_add_head(&inode->i_hash, head);
		list_add(&NILFS_I(inode)->i_dirty, &nilfs->ns_gc_inodes);
	}
	return inode;
}

/*
 * nilfs_clear_gcinode() - clear and free a gc inode
 */
void nilfs_clear_gcinode(struct inode *inode)
{
	nilfs_mdt_destroy(inode);
}

/*
 * nilfs_remove_all_gcinode() - remove all inodes from the_nilfs
 */
void nilfs_remove_all_gcinode(struct the_nilfs *nilfs)
{
	struct hlist_head *head = nilfs->ns_gc_inodes_h;
	struct hlist_node *node, *n;
	struct inode *inode;
	int loop;

	for (loop = 0; loop < NILFS_GCINODE_HASH_SIZE; loop++, head++) {
		hlist_for_each_entry_safe(inode, node, n, head, i_hash) {
			hlist_del_init(&inode->i_hash);
			list_del_init(&NILFS_I(inode)->i_dirty);
			nilfs_clear_gcinode(inode); /* might sleep */
		}
	}
}
Commit	Line	Data
a3d93f70	1	/*
047180f2	2	* gcinode.c - dummy inodes to buffer blocks for garbage collection
a3d93f70 RK	3	*
	4	* Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
	5	*
	6	* This program is free software; you can redistribute it and/or modify
	7	* it under the terms of the GNU General Public License as published by
	8	* the Free Software Foundation; either version 2 of the License, or
	9	* (at your option) any later version.
	10	*
	11	* This program is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* along with this program; if not, write to the Free Software
	18	* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
	19	*
	20	* Written by Seiji Kihara <kihara@osrg.net>, Amagai Yoshiji <amagai@osrg.net>,
	21	* and Ryusuke Konishi <ryusuke@osrg.net>.
	22	* Revised by Ryusuke Konishi <ryusuke@osrg.net>.
	23	*
	24	*/
047180f2 RK	25	/*
	26	* This file adds the cache of on-disk blocks to be moved in garbage
	27	* collection. The disk blocks are held with dummy inodes (called
	28	* gcinodes), and this file provides lookup function of the dummy
	29	* inodes and their buffer read function.
	30	*
7a65004b	31	* Since NILFS2 keeps up multiple checkpoints/snapshots across GC, it
047180f2 RK	32	* has to treat blocks that belong to a same file but have different
047180f2 RK	33	* checkpoint numbers. To avoid interference among generations, dummy
7a65004b	34	* inodes are managed separately from actual inodes, and their lookup
047180f2 RK	35	* function (nilfs_gc_iget) is designed to be specified with a
	36	* checkpoint number argument as well as an inode number.
	37	*
	38	* Buffers and pages held by the dummy inodes will be released each
	39	* time after they are copied to a new log. Dirty blocks made on the
	40	* current generation and the blocks to be moved by GC never overlap
	41	* because the dirty blocks make a new generation; they rather must be
	42	* written individually.
	43	*/
a3d93f70 RK	44
	45	#include <linux/buffer_head.h>
	46	#include <linux/mpage.h>
	47	#include <linux/hash.h>
5a0e3ad6	48	#include <linux/slab.h>
a3d93f70 RK	49	#include <linux/swap.h>
a3d93f70 RK	50	#include "nilfs.h"
05d0e94b RK	51	#include "btree.h"
05d0e94b RK	52	#include "btnode.h"
a3d93f70 RK	53	#include "page.h"
	54	#include "mdt.h"
	55	#include "dat.h"
	56	#include "ifile.h"
	57
7f09410b	58	static const struct address_space_operations def_gcinode_aops = {
fa032744 RK	59	.sync_page = block_sync_page,
fa032744 RK	60	};
a3d93f70 RK	61
	62	/*
	63	* nilfs_gccache_submit_read_data() - add data buffer and submit read request
	64	* @inode - gc inode
	65	* @blkoff - dummy offset treated as the key for the page cache
	66	* @pbn - physical block number of the block
	67	* @vbn - virtual block number of the block, 0 for non-virtual block
	68	* @out_bh - indirect pointer to a buffer_head struct to receive the results
	69	*
	70	* Description: nilfs_gccache_submit_read_data() registers the data buffer
	71	* specified by @pbn to the GC pagecache with the key @blkoff.
	72	* This function sets @vbn (@pbn if @vbn is zero) in b_blocknr of the buffer.
	73	*
	74	* Return Value: On success, 0 is returned. On Error, one of the following
	75	* negative error code is returned.
	76	*
	77	* %-EIO - I/O error.
	78	*
	79	* %-ENOMEM - Insufficient amount of memory available.
	80	*
	81	* %-ENOENT - The block specified with @pbn does not exist.
	82	*/
	83	int nilfs_gccache_submit_read_data(struct inode *inode, sector_t blkoff,
	84	sector_t pbn, __u64 vbn,
	85	struct buffer_head **out_bh)
	86	{
	87	struct buffer_head *bh;
	88	int err;
	89
	90	bh = nilfs_grab_buffer(inode, inode->i_mapping, blkoff, 0);
	91	if (unlikely(!bh))
	92	return -ENOMEM;
	93
	94	if (buffer_uptodate(bh))
	95	goto out;
	96
	97	if (pbn == 0) {
	98	struct inode *dat_inode = NILFS_I_NILFS(inode)->ns_dat;
	99	/* use original dat, not gc dat. */
	100	err = nilfs_dat_translate(dat_inode, vbn, &pbn);
	101	if (unlikely(err)) { /* -EIO, -ENOMEM, -ENOENT */
	102	brelse(bh);
	103	goto failed;
	104	}
	105	}
	106
	107	lock_buffer(bh);
	108	if (buffer_uptodate(bh)) {
	109	unlock_buffer(bh);
	110	goto out;
	111	}
	112
	113	if (!buffer_mapped(bh)) {
	114	bh->b_bdev = NILFS_I_NILFS(inode)->ns_bdev;
	115	set_buffer_mapped(bh);
	116	}
	117	bh->b_blocknr = pbn;
	118	bh->b_end_io = end_buffer_read_sync;
	119	get_bh(bh);
	120	submit_bh(READ, bh);
	121	if (vbn)
	122	bh->b_blocknr = vbn;
	123	out:
	124	err = 0;
125	*out_bh = bh;
126
127	failed:
128	unlock_page(bh->b_page);
129	page_cache_release(bh->b_page);
130	return err;
131	}
132
133	/*
134	* nilfs_gccache_submit_read_node() - add node buffer and submit read request
135	* @inode - gc inode
136	* @pbn - physical block number for the block
137	* @vbn - virtual block number for the block
138	* @out_bh - indirect pointer to a buffer_head struct to receive the results
139	*
140	* Description: nilfs_gccache_submit_read_node() registers the node buffer
141	* specified by @vbn to the GC pagecache. @pbn can be supplied by the
142	* caller to avoid translation of the disk block address.
143	*
144	* Return Value: On success, 0 is returned. On Error, one of the following
145	* negative error code is returned.
146	*
147	* %-EIO - I/O error.
148	*
149	* %-ENOMEM - Insufficient amount of memory available.
150	*/
151	int nilfs_gccache_submit_read_node(struct inode *inode, sector_t pbn,
152	__u64 vbn, struct buffer_head **out_bh)
153	{
26dfdd8e RK	154	int ret;
	155
	156	ret = nilfs_btnode_submit_block(&NILFS_I(inode)->i_btnode_cache,
	157	vbn ? : pbn, pbn, READ, out_bh, &pbn);
a3d93f70 RK	158	if (ret == -EEXIST) /* internal code (cache hit) */
	159	ret = 0;
	160	return ret;
	161	}
	162
	163	int nilfs_gccache_wait_and_mark_dirty(struct buffer_head *bh)
	164	{
	165	wait_on_buffer(bh);
	166	if (!buffer_uptodate(bh))
	167	return -EIO;
	168	if (buffer_dirty(bh))
	169	return -EEXIST;
	170
1d5385b9 RK	171	if (buffer_nilfs_node(bh)) {
	172	if (nilfs_btree_broken_node_block(bh)) {
	173	clear_buffer_uptodate(bh);
	174	return -EIO;
	175	}
a3d93f70	176	nilfs_btnode_mark_dirty(bh);
1d5385b9	177	} else {
a3d93f70	178	nilfs_mdt_mark_buffer_dirty(bh);
1d5385b9	179	}
a3d93f70 RK	180	return 0;
	181	}
	182
	183	/*
	184	* nilfs_init_gccache() - allocate and initialize gc_inode hash table
	185	* @nilfs - the_nilfs
	186	*
	187	* Return Value: On success, 0.
	188	* On error, a negative error code is returned.
	189	*/
	190	int nilfs_init_gccache(struct the_nilfs *nilfs)
	191	{
	192	int loop;
	193
	194	BUG_ON(nilfs->ns_gc_inodes_h);
	195
	196	INIT_LIST_HEAD(&nilfs->ns_gc_inodes);
	197
	198	nilfs->ns_gc_inodes_h =
	199	kmalloc(sizeof(struct hlist_head) * NILFS_GCINODE_HASH_SIZE,
	200	GFP_NOFS);
	201	if (nilfs->ns_gc_inodes_h == NULL)
	202	return -ENOMEM;
	203
	204	for (loop = 0; loop < NILFS_GCINODE_HASH_SIZE; loop++)
	205	INIT_HLIST_HEAD(&nilfs->ns_gc_inodes_h[loop]);
	206	return 0;
	207	}
	208
	209	/*
	210	* nilfs_destroy_gccache() - free gc_inode hash table
	211	* @nilfs - the nilfs
	212	*/
	213	void nilfs_destroy_gccache(struct the_nilfs *nilfs)
	214	{
	215	if (nilfs->ns_gc_inodes_h) {
	216	nilfs_remove_all_gcinode(nilfs);
	217	kfree(nilfs->ns_gc_inodes_h);
	218	nilfs->ns_gc_inodes_h = NULL;
	219	}
	220	}
	221
	222	static struct inode alloc_gcinode(struct the_nilfs nilfs, ino_t ino,
	223	__u64 cno)
	224	{
5731e191	225	struct inode *inode;
a3d93f70 RK	226	struct nilfs_inode_info *ii;
a3d93f70 RK	227
5731e191	228	inode = nilfs_mdt_new_common(nilfs, NULL, ino, GFP_NOFS, 0);
a3d93f70 RK	229	if (!inode)
	230	return NULL;
	231
	232	inode->i_op = NULL;
	233	inode->i_fop = NULL;
	234	inode->i_mapping->a_ops = &def_gcinode_aops;
	235
	236	ii = NILFS_I(inode);
	237	ii->i_cno = cno;
	238	ii->i_flags = 0;
	239	ii->i_state = 1 << NILFS_I_GCINODE;
	240	ii->i_bh = NULL;
a3d93f70 RK	241	nilfs_bmap_init_gc(ii->i_bmap);
	242
	243	return inode;
	244	}
	245
	246	static unsigned long ihash(ino_t ino, __u64 cno)
	247	{
	248	return hash_long((unsigned long)((ino << 2) + cno),
	249	NILFS_GCINODE_HASH_BITS);
	250	}
	251
	252	/*
	253	* nilfs_gc_iget() - find or create gc inode with specified (ino,cno)
	254	*/
	255	struct inode nilfs_gc_iget(struct the_nilfs nilfs, ino_t ino, __u64 cno)
	256	{
	257	struct hlist_head *head = nilfs->ns_gc_inodes_h + ihash(ino, cno);
	258	struct hlist_node *node;
	259	struct inode *inode;
	260
	261	hlist_for_each_entry(inode, node, head, i_hash) {
	262	if (inode->i_ino == ino && NILFS_I(inode)->i_cno == cno)
	263	return inode;
	264	}
	265
	266	inode = alloc_gcinode(nilfs, ino, cno);
	267	if (likely(inode)) {
	268	hlist_add_head(&inode->i_hash, head);
	269	list_add(&NILFS_I(inode)->i_dirty, &nilfs->ns_gc_inodes);
	270	}
	271	return inode;
	272	}
	273
	274	/*
	275	* nilfs_clear_gcinode() - clear and free a gc inode
	276	*/
	277	void nilfs_clear_gcinode(struct inode *inode)
	278	{
a3d93f70 RK	279	nilfs_mdt_destroy(inode);
	280	}
	281
	282	/*
	283	* nilfs_remove_all_gcinode() - remove all inodes from the_nilfs
	284	*/
	285	void nilfs_remove_all_gcinode(struct the_nilfs *nilfs)
	286	{
	287	struct hlist_head *head = nilfs->ns_gc_inodes_h;
	288	struct hlist_node node, n;
	289	struct inode *inode;
	290	int loop;
	291
	292	for (loop = 0; loop < NILFS_GCINODE_HASH_SIZE; loop++, head++) {
	293	hlist_for_each_entry_safe(inode, node, n, head, i_hash) {
	294	hlist_del_init(&inode->i_hash);
	295	list_del_init(&NILFS_I(inode)->i_dirty);
	296	nilfs_clear_gcinode(inode); /* might sleep */
	297	}
	298	}
	299	}