Loading...
1/*
2 * Copyright (C) 2007 Oracle. All rights reserved.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
7 *
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
12 *
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
17 */
18
19#ifndef __BTRFS_I__
20#define __BTRFS_I__
21
22#include "extent_map.h"
23#include "extent_io.h"
24#include "ordered-data.h"
25#include "delayed-inode.h"
26
27/* in memory btrfs inode */
28struct btrfs_inode {
29 /* which subvolume this inode belongs to */
30 struct btrfs_root *root;
31
32 /* key used to find this inode on disk. This is used by the code
33 * to read in roots of subvolumes
34 */
35 struct btrfs_key location;
36
37 /* Lock for counters */
38 spinlock_t lock;
39
40 /* the extent_tree has caches of all the extent mappings to disk */
41 struct extent_map_tree extent_tree;
42
43 /* the io_tree does range state (DIRTY, LOCKED etc) */
44 struct extent_io_tree io_tree;
45
46 /* special utility tree used to record which mirrors have already been
47 * tried when checksums fail for a given block
48 */
49 struct extent_io_tree io_failure_tree;
50
51 /* held while logging the inode in tree-log.c */
52 struct mutex log_mutex;
53
54 /* used to order data wrt metadata */
55 struct btrfs_ordered_inode_tree ordered_tree;
56
57 /* for keeping track of orphaned inodes */
58 struct list_head i_orphan;
59
60 /* list of all the delalloc inodes in the FS. There are times we need
61 * to write all the delalloc pages to disk, and this list is used
62 * to walk them all.
63 */
64 struct list_head delalloc_inodes;
65
66 /*
67 * list for tracking inodes that must be sent to disk before a
68 * rename or truncate commit
69 */
70 struct list_head ordered_operations;
71
72 /* node for the red-black tree that links inodes in subvolume root */
73 struct rb_node rb_node;
74
75 /* the space_info for where this inode's data allocations are done */
76 struct btrfs_space_info *space_info;
77
78 /* full 64 bit generation number, struct vfs_inode doesn't have a big
79 * enough field for this.
80 */
81 u64 generation;
82
83 /* sequence number for NFS changes */
84 u64 sequence;
85
86 /*
87 * transid of the trans_handle that last modified this inode
88 */
89 u64 last_trans;
90
91 /*
92 * log transid when this inode was last modified
93 */
94 u64 last_sub_trans;
95
96 /*
97 * transid that last logged this inode
98 */
99 u64 logged_trans;
100
101 /* total number of bytes pending delalloc, used by stat to calc the
102 * real block usage of the file
103 */
104 u64 delalloc_bytes;
105
106 /* total number of bytes that may be used for this inode for
107 * delalloc
108 */
109 u64 reserved_bytes;
110
111 /*
112 * the size of the file stored in the metadata on disk. data=ordered
113 * means the in-memory i_size might be larger than the size on disk
114 * because not all the blocks are written yet.
115 */
116 u64 disk_i_size;
117
118 /* flags field from the on disk inode */
119 u32 flags;
120
121 /*
122 * if this is a directory then index_cnt is the counter for the index
123 * number for new files that are created
124 */
125 u64 index_cnt;
126
127 /* the fsync log has some corner cases that mean we have to check
128 * directories to see if any unlinks have been done before
129 * the directory was logged. See tree-log.c for all the
130 * details
131 */
132 u64 last_unlink_trans;
133
134 /*
135 * Counters to keep track of the number of extent item's we may use due
136 * to delalloc and such. outstanding_extents is the number of extent
137 * items we think we'll end up using, and reserved_extents is the number
138 * of extent items we've reserved metadata for.
139 */
140 unsigned outstanding_extents;
141 unsigned reserved_extents;
142
143 /*
144 * ordered_data_close is set by truncate when a file that used
145 * to have good data has been truncated to zero. When it is set
146 * the btrfs file release call will add this inode to the
147 * ordered operations list so that we make sure to flush out any
148 * new data the application may have written before commit.
149 *
150 * yes, its silly to have a single bitflag, but we might grow more
151 * of these.
152 */
153 unsigned ordered_data_close:1;
154 unsigned orphan_meta_reserved:1;
155 unsigned dummy_inode:1;
156 unsigned in_defrag:1;
157
158 /*
159 * always compress this one file
160 */
161 unsigned force_compress:4;
162
163 struct btrfs_delayed_node *delayed_node;
164
165 struct inode vfs_inode;
166};
167
168extern unsigned char btrfs_filetype_table[];
169
170static inline struct btrfs_inode *BTRFS_I(struct inode *inode)
171{
172 return container_of(inode, struct btrfs_inode, vfs_inode);
173}
174
175static inline u64 btrfs_ino(struct inode *inode)
176{
177 u64 ino = BTRFS_I(inode)->location.objectid;
178
179 /*
180 * !ino: btree_inode
181 * type == BTRFS_ROOT_ITEM_KEY: subvol dir
182 */
183 if (!ino || BTRFS_I(inode)->location.type == BTRFS_ROOT_ITEM_KEY)
184 ino = inode->i_ino;
185 return ino;
186}
187
188static inline void btrfs_i_size_write(struct inode *inode, u64 size)
189{
190 i_size_write(inode, size);
191 BTRFS_I(inode)->disk_i_size = size;
192}
193
194static inline bool btrfs_is_free_space_inode(struct btrfs_root *root,
195 struct inode *inode)
196{
197 if (root == root->fs_info->tree_root ||
198 BTRFS_I(inode)->location.objectid == BTRFS_FREE_INO_OBJECTID)
199 return true;
200 return false;
201}
202
203#endif
1/*
2 * Copyright (C) 2007 Oracle. All rights reserved.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
7 *
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
12 *
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
17 */
18
19#ifndef __BTRFS_I__
20#define __BTRFS_I__
21
22#include <linux/hash.h>
23#include "extent_map.h"
24#include "extent_io.h"
25#include "ordered-data.h"
26#include "delayed-inode.h"
27
28/*
29 * ordered_data_close is set by truncate when a file that used
30 * to have good data has been truncated to zero. When it is set
31 * the btrfs file release call will add this inode to the
32 * ordered operations list so that we make sure to flush out any
33 * new data the application may have written before commit.
34 */
35#define BTRFS_INODE_ORDERED_DATA_CLOSE 0
36#define BTRFS_INODE_ORPHAN_META_RESERVED 1
37#define BTRFS_INODE_DUMMY 2
38#define BTRFS_INODE_IN_DEFRAG 3
39#define BTRFS_INODE_DELALLOC_META_RESERVED 4
40#define BTRFS_INODE_HAS_ORPHAN_ITEM 5
41#define BTRFS_INODE_HAS_ASYNC_EXTENT 6
42#define BTRFS_INODE_NEEDS_FULL_SYNC 7
43#define BTRFS_INODE_COPY_EVERYTHING 8
44#define BTRFS_INODE_IN_DELALLOC_LIST 9
45#define BTRFS_INODE_READDIO_NEED_LOCK 10
46#define BTRFS_INODE_HAS_PROPS 11
47
48/* in memory btrfs inode */
49struct btrfs_inode {
50 /* which subvolume this inode belongs to */
51 struct btrfs_root *root;
52
53 /* key used to find this inode on disk. This is used by the code
54 * to read in roots of subvolumes
55 */
56 struct btrfs_key location;
57
58 /* Lock for counters */
59 spinlock_t lock;
60
61 /* the extent_tree has caches of all the extent mappings to disk */
62 struct extent_map_tree extent_tree;
63
64 /* the io_tree does range state (DIRTY, LOCKED etc) */
65 struct extent_io_tree io_tree;
66
67 /* special utility tree used to record which mirrors have already been
68 * tried when checksums fail for a given block
69 */
70 struct extent_io_tree io_failure_tree;
71
72 /* held while logging the inode in tree-log.c */
73 struct mutex log_mutex;
74
75 /* held while doing delalloc reservations */
76 struct mutex delalloc_mutex;
77
78 /* used to order data wrt metadata */
79 struct btrfs_ordered_inode_tree ordered_tree;
80
81 /* list of all the delalloc inodes in the FS. There are times we need
82 * to write all the delalloc pages to disk, and this list is used
83 * to walk them all.
84 */
85 struct list_head delalloc_inodes;
86
87 /*
88 * list for tracking inodes that must be sent to disk before a
89 * rename or truncate commit
90 */
91 struct list_head ordered_operations;
92
93 /* node for the red-black tree that links inodes in subvolume root */
94 struct rb_node rb_node;
95
96 unsigned long runtime_flags;
97
98 /* Keep track of who's O_SYNC/fsyncing currently */
99 atomic_t sync_writers;
100
101 /* full 64 bit generation number, struct vfs_inode doesn't have a big
102 * enough field for this.
103 */
104 u64 generation;
105
106 /*
107 * transid of the trans_handle that last modified this inode
108 */
109 u64 last_trans;
110
111 /*
112 * transid that last logged this inode
113 */
114 u64 logged_trans;
115
116 /*
117 * log transid when this inode was last modified
118 */
119 int last_sub_trans;
120
121 /* a local copy of root's last_log_commit */
122 int last_log_commit;
123
124 /* total number of bytes pending delalloc, used by stat to calc the
125 * real block usage of the file
126 */
127 u64 delalloc_bytes;
128
129 /*
130 * the size of the file stored in the metadata on disk. data=ordered
131 * means the in-memory i_size might be larger than the size on disk
132 * because not all the blocks are written yet.
133 */
134 u64 disk_i_size;
135
136 /*
137 * if this is a directory then index_cnt is the counter for the index
138 * number for new files that are created
139 */
140 u64 index_cnt;
141
142 /* Cache the directory index number to speed the dir/file remove */
143 u64 dir_index;
144
145 /* the fsync log has some corner cases that mean we have to check
146 * directories to see if any unlinks have been done before
147 * the directory was logged. See tree-log.c for all the
148 * details
149 */
150 u64 last_unlink_trans;
151
152 /*
153 * Number of bytes outstanding that are going to need csums. This is
154 * used in ENOSPC accounting.
155 */
156 u64 csum_bytes;
157
158 /* flags field from the on disk inode */
159 u32 flags;
160
161 /*
162 * Counters to keep track of the number of extent item's we may use due
163 * to delalloc and such. outstanding_extents is the number of extent
164 * items we think we'll end up using, and reserved_extents is the number
165 * of extent items we've reserved metadata for.
166 */
167 unsigned outstanding_extents;
168 unsigned reserved_extents;
169
170 /*
171 * always compress this one file
172 */
173 unsigned force_compress;
174
175 struct btrfs_delayed_node *delayed_node;
176
177 struct inode vfs_inode;
178};
179
180extern unsigned char btrfs_filetype_table[];
181
182static inline struct btrfs_inode *BTRFS_I(struct inode *inode)
183{
184 return container_of(inode, struct btrfs_inode, vfs_inode);
185}
186
187static inline unsigned long btrfs_inode_hash(u64 objectid,
188 const struct btrfs_root *root)
189{
190 u64 h = objectid ^ (root->objectid * GOLDEN_RATIO_PRIME);
191
192#if BITS_PER_LONG == 32
193 h = (h >> 32) ^ (h & 0xffffffff);
194#endif
195
196 return (unsigned long)h;
197}
198
199static inline void btrfs_insert_inode_hash(struct inode *inode)
200{
201 unsigned long h = btrfs_inode_hash(inode->i_ino, BTRFS_I(inode)->root);
202
203 __insert_inode_hash(inode, h);
204}
205
206static inline u64 btrfs_ino(struct inode *inode)
207{
208 u64 ino = BTRFS_I(inode)->location.objectid;
209
210 /*
211 * !ino: btree_inode
212 * type == BTRFS_ROOT_ITEM_KEY: subvol dir
213 */
214 if (!ino || BTRFS_I(inode)->location.type == BTRFS_ROOT_ITEM_KEY)
215 ino = inode->i_ino;
216 return ino;
217}
218
219static inline void btrfs_i_size_write(struct inode *inode, u64 size)
220{
221 i_size_write(inode, size);
222 BTRFS_I(inode)->disk_i_size = size;
223}
224
225static inline bool btrfs_is_free_space_inode(struct inode *inode)
226{
227 struct btrfs_root *root = BTRFS_I(inode)->root;
228
229 if (root == root->fs_info->tree_root &&
230 btrfs_ino(inode) != BTRFS_BTREE_INODE_OBJECTID)
231 return true;
232 if (BTRFS_I(inode)->location.objectid == BTRFS_FREE_INO_OBJECTID)
233 return true;
234 return false;
235}
236
237static inline int btrfs_inode_in_log(struct inode *inode, u64 generation)
238{
239 if (BTRFS_I(inode)->logged_trans == generation &&
240 BTRFS_I(inode)->last_sub_trans <=
241 BTRFS_I(inode)->last_log_commit &&
242 BTRFS_I(inode)->last_sub_trans <=
243 BTRFS_I(inode)->root->last_log_commit)
244 return 1;
245 return 0;
246}
247
248struct btrfs_dio_private {
249 struct inode *inode;
250 u64 logical_offset;
251 u64 disk_bytenr;
252 u64 bytes;
253 void *private;
254
255 /* number of bios pending for this dio */
256 atomic_t pending_bios;
257
258 /* IO errors */
259 int errors;
260
261 /* orig_bio is our btrfs_io_bio */
262 struct bio *orig_bio;
263
264 /* dio_bio came from fs/direct-io.c */
265 struct bio *dio_bio;
266 u8 csum[0];
267};
268
269/*
270 * Disable DIO read nolock optimization, so new dio readers will be forced
271 * to grab i_mutex. It is used to avoid the endless truncate due to
272 * nonlocked dio read.
273 */
274static inline void btrfs_inode_block_unlocked_dio(struct inode *inode)
275{
276 set_bit(BTRFS_INODE_READDIO_NEED_LOCK, &BTRFS_I(inode)->runtime_flags);
277 smp_mb();
278}
279
280static inline void btrfs_inode_resume_unlocked_dio(struct inode *inode)
281{
282 smp_mb__before_clear_bit();
283 clear_bit(BTRFS_INODE_READDIO_NEED_LOCK,
284 &BTRFS_I(inode)->runtime_flags);
285}
286
287#endif