blob: 1c2a3e23f8b2dec6b1f895098239864d7f5e9946 [file] [log] [blame]
Ryusuke Konishia3d93f72009-04-06 19:01:40 -07001/*
Ryusuke Konishi047180f2009-04-06 19:01:45 -07002 * gcinode.c - dummy inodes to buffer blocks for garbage collection
Ryusuke Konishia3d93f72009-04-06 19:01:40 -07003 *
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 */
Ryusuke Konishi047180f2009-04-06 19:01:45 -070025/*
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 *
Ryusuke Konishi047180f2009-04-06 19:01:45 -070031 * Buffers and pages held by the dummy inodes will be released each
32 * time after they are copied to a new log. Dirty blocks made on the
33 * current generation and the blocks to be moved by GC never overlap
34 * because the dirty blocks make a new generation; they rather must be
35 * written individually.
36 */
Ryusuke Konishia3d93f72009-04-06 19:01:40 -070037
38#include <linux/buffer_head.h>
39#include <linux/mpage.h>
40#include <linux/hash.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +090041#include <linux/slab.h>
Ryusuke Konishia3d93f72009-04-06 19:01:40 -070042#include <linux/swap.h>
43#include "nilfs.h"
Ryusuke Konishi05d0e942010-07-10 20:52:09 +090044#include "btree.h"
45#include "btnode.h"
Ryusuke Konishia3d93f72009-04-06 19:01:40 -070046#include "page.h"
47#include "mdt.h"
48#include "dat.h"
49#include "ifile.h"
50
Alexey Dobriyan7f094102009-09-21 17:01:10 -070051static const struct address_space_operations def_gcinode_aops = {
Ryusuke Konishifa032742009-05-27 22:44:34 +090052};
Ryusuke Konishia3d93f72009-04-06 19:01:40 -070053
54/*
55 * nilfs_gccache_submit_read_data() - add data buffer and submit read request
56 * @inode - gc inode
57 * @blkoff - dummy offset treated as the key for the page cache
58 * @pbn - physical block number of the block
59 * @vbn - virtual block number of the block, 0 for non-virtual block
60 * @out_bh - indirect pointer to a buffer_head struct to receive the results
61 *
62 * Description: nilfs_gccache_submit_read_data() registers the data buffer
63 * specified by @pbn to the GC pagecache with the key @blkoff.
64 * This function sets @vbn (@pbn if @vbn is zero) in b_blocknr of the buffer.
65 *
66 * Return Value: On success, 0 is returned. On Error, one of the following
67 * negative error code is returned.
68 *
69 * %-EIO - I/O error.
70 *
71 * %-ENOMEM - Insufficient amount of memory available.
72 *
73 * %-ENOENT - The block specified with @pbn does not exist.
74 */
75int nilfs_gccache_submit_read_data(struct inode *inode, sector_t blkoff,
76 sector_t pbn, __u64 vbn,
77 struct buffer_head **out_bh)
78{
79 struct buffer_head *bh;
80 int err;
81
82 bh = nilfs_grab_buffer(inode, inode->i_mapping, blkoff, 0);
83 if (unlikely(!bh))
84 return -ENOMEM;
85
86 if (buffer_uptodate(bh))
87 goto out;
88
89 if (pbn == 0) {
90 struct inode *dat_inode = NILFS_I_NILFS(inode)->ns_dat;
91 /* use original dat, not gc dat. */
92 err = nilfs_dat_translate(dat_inode, vbn, &pbn);
93 if (unlikely(err)) { /* -EIO, -ENOMEM, -ENOENT */
94 brelse(bh);
95 goto failed;
96 }
97 }
98
99 lock_buffer(bh);
100 if (buffer_uptodate(bh)) {
101 unlock_buffer(bh);
102 goto out;
103 }
104
105 if (!buffer_mapped(bh)) {
106 bh->b_bdev = NILFS_I_NILFS(inode)->ns_bdev;
107 set_buffer_mapped(bh);
108 }
109 bh->b_blocknr = pbn;
110 bh->b_end_io = end_buffer_read_sync;
111 get_bh(bh);
112 submit_bh(READ, bh);
113 if (vbn)
114 bh->b_blocknr = vbn;
115 out:
116 err = 0;
117 *out_bh = bh;
118
119 failed:
120 unlock_page(bh->b_page);
121 page_cache_release(bh->b_page);
122 return err;
123}
124
125/*
126 * nilfs_gccache_submit_read_node() - add node buffer and submit read request
127 * @inode - gc inode
128 * @pbn - physical block number for the block
129 * @vbn - virtual block number for the block
130 * @out_bh - indirect pointer to a buffer_head struct to receive the results
131 *
132 * Description: nilfs_gccache_submit_read_node() registers the node buffer
133 * specified by @vbn to the GC pagecache. @pbn can be supplied by the
134 * caller to avoid translation of the disk block address.
135 *
136 * Return Value: On success, 0 is returned. On Error, one of the following
137 * negative error code is returned.
138 *
139 * %-EIO - I/O error.
140 *
141 * %-ENOMEM - Insufficient amount of memory available.
142 */
143int nilfs_gccache_submit_read_node(struct inode *inode, sector_t pbn,
144 __u64 vbn, struct buffer_head **out_bh)
145{
Ryusuke Konishi26dfdd82010-07-18 10:42:23 +0900146 int ret;
147
148 ret = nilfs_btnode_submit_block(&NILFS_I(inode)->i_btnode_cache,
149 vbn ? : pbn, pbn, READ, out_bh, &pbn);
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700150 if (ret == -EEXIST) /* internal code (cache hit) */
151 ret = 0;
152 return ret;
153}
154
155int nilfs_gccache_wait_and_mark_dirty(struct buffer_head *bh)
156{
157 wait_on_buffer(bh);
158 if (!buffer_uptodate(bh))
159 return -EIO;
160 if (buffer_dirty(bh))
161 return -EEXIST;
162
Ryusuke Konishi1d5385b2010-07-16 23:52:40 +0900163 if (buffer_nilfs_node(bh)) {
164 if (nilfs_btree_broken_node_block(bh)) {
165 clear_buffer_uptodate(bh);
166 return -EIO;
167 }
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700168 nilfs_btnode_mark_dirty(bh);
Ryusuke Konishi1d5385b2010-07-16 23:52:40 +0900169 } else {
Ryusuke Konishiadbb39b2010-09-05 10:14:43 +0900170 nilfs_mark_buffer_dirty(bh);
Ryusuke Konishi1d5385b2010-07-16 23:52:40 +0900171 }
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700172 return 0;
173}
174
Ryusuke Konishi263d90c2010-08-20 19:06:11 +0900175int nilfs_init_gcinode(struct inode *inode)
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700176{
Ryusuke Konishi263d90c2010-08-20 19:06:11 +0900177 struct nilfs_inode_info *ii = NILFS_I(inode);
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700178
Ryusuke Konishiadbb39b2010-09-05 10:14:43 +0900179 inode->i_mode = S_IFREG;
180 mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
181 inode->i_mapping->a_ops = &def_gcinode_aops;
182 inode->i_mapping->backing_dev_info = inode->i_sb->s_bdi;
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700183
Ryusuke Konishiadbb39b2010-09-05 10:14:43 +0900184 ii->i_flags = 0;
185 nilfs_bmap_init_gc(ii->i_bmap);
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700186
Ryusuke Konishiadbb39b2010-09-05 10:14:43 +0900187 return 0;
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700188}
189
Ryusuke Konishi263d90c2010-08-20 19:06:11 +0900190/**
191 * nilfs_remove_all_gcinodes() - remove all unprocessed gc inodes
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700192 */
Ryusuke Konishi263d90c2010-08-20 19:06:11 +0900193void nilfs_remove_all_gcinodes(struct the_nilfs *nilfs)
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700194{
Ryusuke Konishi263d90c2010-08-20 19:06:11 +0900195 struct list_head *head = &nilfs->ns_gc_inodes;
196 struct nilfs_inode_info *ii;
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700197
Ryusuke Konishi263d90c2010-08-20 19:06:11 +0900198 while (!list_empty(head)) {
199 ii = list_first_entry(head, struct nilfs_inode_info, i_dirty);
200 list_del_init(&ii->i_dirty);
201 iput(&ii->vfs_inode);
Ryusuke Konishia3d93f72009-04-06 19:01:40 -0700202 }
203}