blob: 7f7c5a92d2b8799ce9da6ec0800b053133692595 [file] [log] [blame]
David Sterba9888c342018-04-03 19:16:55 +02001/* SPDX-License-Identifier: GPL-2.0 */
Chris Mason6cbd5572007-06-12 09:07:21 -04002/*
3 * Copyright (C) 2007 Oracle. All rights reserved.
Chris Mason6cbd5572007-06-12 09:07:21 -04004 */
5
David Sterba9888c342018-04-03 19:16:55 +02006#ifndef BTRFS_INODE_H
7#define BTRFS_INODE_H
Chris Mason2c90e5d62007-04-02 10:50:19 -04008
Filipe David Borba Manana778ba822013-10-06 22:22:33 +01009#include <linux/hash.h>
Omar Sandovale3b318d2020-04-16 14:46:20 -070010#include <linux/refcount.h>
Josef Bacik04cc63d2023-08-25 16:19:26 -040011#include <linux/fscrypt.h>
Josef Bacik5335f432023-08-25 16:19:28 -040012#include <trace/events/btrfs.h>
Chris Masona52d9a82007-08-27 16:49:44 -040013#include "extent_map.h"
Chris Masond1310b22008-01-24 16:13:08 -050014#include "extent_io.h"
Chris Masone6dcd2d2008-07-17 12:53:50 -040015#include "ordered-data.h"
Miao Xie16cdcec2011-04-22 18:12:22 +080016#include "delayed-inode.h"
Chris Masona52d9a82007-08-27 16:49:44 -040017
Josef Bacik72ac3c02012-05-23 14:13:11 -040018/*
Filipe Manana528ee692021-12-15 12:19:59 +000019 * Since we search a directory based on f_pos (struct dir_context::pos) we have
20 * to start at 2 since '.' and '..' have f_pos of 0 and 1 respectively, so
21 * everybody else has to start at 2 (see btrfs_real_readdir() and dir_emit_dots()).
22 */
23#define BTRFS_DIR_START_INDEX 2
24
25/*
Josef Bacik72ac3c02012-05-23 14:13:11 -040026 * ordered_data_close is set by truncate when a file that used
27 * to have good data has been truncated to zero. When it is set
28 * the btrfs file release call will add this inode to the
29 * ordered operations list so that we make sure to flush out any
30 * new data the application may have written before commit.
31 */
Omar Sandoval7efc3e32018-05-11 13:13:39 -070032enum {
Nikolay Borisov1fd40332020-10-01 09:40:39 +030033 BTRFS_INODE_FLUSH_ON_CLOSE,
Omar Sandoval7efc3e32018-05-11 13:13:39 -070034 BTRFS_INODE_DUMMY,
35 BTRFS_INODE_IN_DEFRAG,
36 BTRFS_INODE_HAS_ASYNC_EXTENT,
Filipe Manana48778172020-08-11 12:43:58 +010037 /*
38 * Always set under the VFS' inode lock, otherwise it can cause races
39 * during fsync (we start as a fast fsync and then end up in a full
40 * fsync racing with ordered extent completion).
41 */
Omar Sandoval7efc3e32018-05-11 13:13:39 -070042 BTRFS_INODE_NEEDS_FULL_SYNC,
43 BTRFS_INODE_COPY_EVERYTHING,
44 BTRFS_INODE_IN_DELALLOC_LIST,
Omar Sandoval7efc3e32018-05-11 13:13:39 -070045 BTRFS_INODE_HAS_PROPS,
Ethan Lien3cd24c62018-11-01 14:49:03 +080046 BTRFS_INODE_SNAPSHOT_FLUSH,
Filipe Mananaf2f121a2020-11-13 11:21:49 +000047 /*
48 * Set and used when logging an inode and it serves to signal that an
49 * inode does not have xattrs, so subsequent fsyncs can avoid searching
50 * for xattrs to log. This bit must be cleared whenever a xattr is added
51 * to an inode.
52 */
53 BTRFS_INODE_NO_XATTRS,
Filipe Manana3d45f222020-12-02 11:55:58 +000054 /*
55 * Set when we are in a context where we need to start a transaction and
56 * have dirty pages with the respective file range locked. This is to
57 * ensure that when reserving space for the transaction, if we are low
58 * on available space and need to flush delalloc, we will not flush
59 * delalloc for this inode, because that could result in a deadlock (on
60 * the file range, inode's io_tree).
61 */
62 BTRFS_INODE_NO_DELALLOC_FLUSH,
Boris Burkov14605402021-06-30 13:01:49 -070063 /*
64 * Set when we are working on enabling verity for a file. Computing and
65 * writing the whole Merkle tree can take a while so we want to prevent
66 * races where two separate tasks attempt to simultaneously start verity
67 * on the same file.
68 */
69 BTRFS_INODE_VERITY_IN_PROGRESS,
Josef Bacik9b9b8852022-09-14 19:04:50 -040070 /* Set when this inode is a free space inode. */
71 BTRFS_INODE_FREE_SPACE_INODE,
Josef Baciked9b50a2023-11-29 13:10:31 -050072 /* Set when there are no capabilities in XATTs for the inode. */
73 BTRFS_INODE_NO_CAP_XATTR,
Omar Sandoval7efc3e32018-05-11 13:13:39 -070074};
Josef Bacik72ac3c02012-05-23 14:13:11 -040075
Aneeshf1ace242007-06-13 16:18:26 -040076/* in memory btrfs inode */
Chris Mason2c90e5d62007-04-02 10:50:19 -040077struct btrfs_inode {
Chris Masond352ac62008-09-29 15:18:18 -040078 /* which subvolume this inode belongs to */
Chris Masond6e4a422007-04-06 15:37:36 -040079 struct btrfs_root *root;
Chris Masond352ac62008-09-29 15:18:18 -040080
Chris Masond352ac62008-09-29 15:18:18 -040081 /* key used to find this inode on disk. This is used by the code
82 * to read in roots of subvolumes
83 */
Chris Masond6e4a422007-04-06 15:37:36 -040084 struct btrfs_key location;
Chris Masond352ac62008-09-29 15:18:18 -040085
David Sterbae41570d2023-09-08 01:09:38 +020086 /* Cached value of inode property 'compression'. */
87 u8 prop_compress;
88
89 /*
90 * Force compression on the file using the defrag ioctl, could be
91 * different from prop_compress and takes precedence if set.
92 */
93 u8 defrag_compress;
94
Filipe Manana2f2ff0e2015-03-20 17:19:46 +000095 /*
96 * Lock for counters and all fields used to determine if the inode is in
97 * the log or not (last_trans, last_sub_trans, last_log_commit,
Filipe Manana68539bd2023-10-04 11:38:53 +010098 * logged_trans), to access/update delalloc_bytes, new_delalloc_bytes,
99 * defrag_bytes, disk_i_size, outstanding_extents, csum_bytes and to
100 * update the VFS' inode number of bytes used.
Filipe Manana2f2ff0e2015-03-20 17:19:46 +0000101 */
Josef Bacik9e0baf62011-07-15 15:16:44 +0000102 spinlock_t lock;
103
Chris Masond352ac62008-09-29 15:18:18 -0400104 /* the extent_tree has caches of all the extent mappings to disk */
Chris Masona52d9a82007-08-27 16:49:44 -0400105 struct extent_map_tree extent_tree;
Chris Masond352ac62008-09-29 15:18:18 -0400106
107 /* the io_tree does range state (DIRTY, LOCKED etc) */
Chris Masond1310b22008-01-24 16:13:08 -0500108 struct extent_io_tree io_tree;
Chris Masond352ac62008-09-29 15:18:18 -0400109
Josef Bacik41a2ee72020-01-17 09:02:21 -0500110 /*
111 * Keep track of where the inode has extent items mapped in order to
David Sterba637e6e02023-11-30 23:42:01 +0100112 * make sure the i_size adjustments are accurate. Not required when the
113 * filesystem is NO_HOLES, the status can't be set while mounted as
114 * it's a mkfs-time feature.
Josef Bacik41a2ee72020-01-17 09:02:21 -0500115 */
David Sterba637e6e02023-11-30 23:42:01 +0100116 struct extent_io_tree *file_extent_tree;
Josef Bacik41a2ee72020-01-17 09:02:21 -0500117
Chris Masond352ac62008-09-29 15:18:18 -0400118 /* held while logging the inode in tree-log.c */
Chris Masone02119d2008-09-05 16:13:11 -0400119 struct mutex log_mutex;
Chris Masond352ac62008-09-29 15:18:18 -0400120
David Sterba398fb912023-09-27 21:04:32 +0200121 /*
122 * Counters to keep track of the number of extent item's we may use due
123 * to delalloc and such. outstanding_extents is the number of extent
124 * items we think we'll end up using, and reserved_extents is the number
Filipe Manana68539bd2023-10-04 11:38:53 +0100125 * of extent items we've reserved metadata for. Protected by 'lock'.
David Sterba398fb912023-09-27 21:04:32 +0200126 */
127 unsigned outstanding_extents;
128
Chris Masond352ac62008-09-29 15:18:18 -0400129 /* used to order data wrt metadata */
David Sterba54c653712023-09-27 14:22:39 +0200130 spinlock_t ordered_tree_lock;
131 struct rb_root ordered_tree;
132 struct rb_node *ordered_tree_last;
Josef Bacik15ee9bc2007-08-10 16:22:09 -0400133
Chris Masond352ac62008-09-29 15:18:18 -0400134 /* list of all the delalloc inodes in the FS. There are times we need
135 * to write all the delalloc pages to disk, and this list is used
136 * to walk them all.
137 */
Chris Masonea8c2812008-08-04 23:17:27 -0400138 struct list_head delalloc_inodes;
139
Yan Zheng5d4f98a2009-06-10 10:45:14 -0400140 /* node for the red-black tree that links inodes in subvolume root */
141 struct rb_node rb_node;
142
Josef Bacik72ac3c02012-05-23 14:13:11 -0400143 unsigned long runtime_flags;
144
Chris Masond352ac62008-09-29 15:18:18 -0400145 /* full 64 bit generation number, struct vfs_inode doesn't have a big
146 * enough field for this.
147 */
Chris Masone02119d2008-09-05 16:13:11 -0400148 u64 generation;
149
Josef Bacik15ee9bc2007-08-10 16:22:09 -0400150 /*
Filipe Manana68539bd2023-10-04 11:38:53 +0100151 * ID of the transaction handle that last modified this inode.
152 * Protected by 'lock'.
Josef Bacik15ee9bc2007-08-10 16:22:09 -0400153 */
154 u64 last_trans;
Chris Mason257c62e2009-10-13 13:21:08 -0400155
156 /*
Filipe Manana68539bd2023-10-04 11:38:53 +0100157 * ID of the transaction that last logged this inode.
158 * Protected by 'lock'.
Chris Masone02119d2008-09-05 16:13:11 -0400159 */
160 u64 logged_trans;
Chris Mason49eb7e42008-09-11 15:53:12 -0400161
Miao Xiebb14a592014-02-20 18:08:56 +0800162 /*
Filipe Manana68539bd2023-10-04 11:38:53 +0100163 * Log transaction ID when this inode was last modified.
164 * Protected by 'lock'.
Miao Xiebb14a592014-02-20 18:08:56 +0800165 */
166 int last_sub_trans;
167
Filipe Manana68539bd2023-10-04 11:38:53 +0100168 /* A local copy of root's last_log_commit. Protected by 'lock'. */
Miao Xiebb14a592014-02-20 18:08:56 +0800169 int last_log_commit;
170
Filipe Mananafa4b8cb2023-04-05 18:51:29 +0100171 union {
172 /*
173 * Total number of bytes pending delalloc, used by stat to
174 * calculate the real block usage of the file. This is used
Filipe Manana68539bd2023-10-04 11:38:53 +0100175 * only for files. Protected by 'lock'.
Filipe Mananafa4b8cb2023-04-05 18:51:29 +0100176 */
177 u64 delalloc_bytes;
178 /*
179 * The lowest possible index of the next dir index key which
180 * points to an inode that needs to be logged.
181 * This is used only for directories.
182 * Use the helpers btrfs_get_first_dir_index_to_log() and
183 * btrfs_set_first_dir_index_to_log() to access this field.
184 */
185 u64 first_dir_index_to_log;
186 };
Chris Masond352ac62008-09-29 15:18:18 -0400187
Filipe Mananadc287222021-09-16 11:32:14 +0100188 union {
189 /*
190 * Total number of bytes pending delalloc that fall within a file
191 * range that is either a hole or beyond EOF (and no prealloc extent
192 * exists in the range). This is always <= delalloc_bytes and this
Filipe Manana68539bd2023-10-04 11:38:53 +0100193 * is used only for files. Protected by 'lock'.
Filipe Mananadc287222021-09-16 11:32:14 +0100194 */
195 u64 new_delalloc_bytes;
196 /*
197 * The offset of the last dir index key that was logged.
198 * This is used only for directories.
199 */
200 u64 last_dir_index_offset;
201 };
Filipe Mananaa7e3b972017-04-03 10:45:46 +0100202
203 /*
Filipe Manana68539bd2023-10-04 11:38:53 +0100204 * Total number of bytes pending defrag, used by stat to check whether
205 * it needs COW. Protected by 'lock'.
Wang Shilong47059d92014-07-03 18:22:07 +0800206 */
207 u64 defrag_bytes;
208
209 /*
Filipe Manana68539bd2023-10-04 11:38:53 +0100210 * The size of the file stored in the metadata on disk. data=ordered
Chris Masond352ac62008-09-29 15:18:18 -0400211 * means the in-memory i_size might be larger than the size on disk
Filipe Manana68539bd2023-10-04 11:38:53 +0100212 * because not all the blocks are written yet. Protected by 'lock'.
Chris Masond352ac62008-09-29 15:18:18 -0400213 */
Chris Masondbe674a2008-07-17 12:54:05 -0400214 u64 disk_i_size;
Chris Masond352ac62008-09-29 15:18:18 -0400215
Josef Bacikaec74772008-07-24 12:12:38 -0400216 /*
Filipe Manana528ee692021-12-15 12:19:59 +0000217 * If this is a directory then index_cnt is the counter for the index
218 * number for new files that are created. For an empty directory, this
219 * must be initialized to BTRFS_DIR_START_INDEX.
Josef Bacikaec74772008-07-24 12:12:38 -0400220 */
221 u64 index_cnt;
Chris Masond352ac62008-09-29 15:18:18 -0400222
Miao Xie67de1172013-12-26 13:07:06 +0800223 /* Cache the directory index number to speed the dir/file remove */
224 u64 dir_index;
225
Chris Mason12fcfd22009-03-24 10:24:20 -0400226 /* the fsync log has some corner cases that mean we have to check
227 * directories to see if any unlinks have been done before
228 * the directory was logged. See tree-log.c for all the
229 * details
230 */
231 u64 last_unlink_trans;
232
Josef Bacik7709cde2011-08-04 10:25:02 -0400233 /*
Filipe Manana3ebac172020-07-15 12:30:43 +0100234 * The id/generation of the last transaction where this inode was
235 * either the source or the destination of a clone/dedupe operation.
236 * Used when logging an inode to know if there are shared extents that
237 * need special care when logging checksum items, to avoid duplicate
238 * checksum items in a log (which can lead to a corruption where we end
239 * up with missing checksum ranges after log replay).
240 * Protected by the vfs inode lock.
241 */
242 u64 last_reflink_trans;
243
244 /*
Josef Bacik7709cde2011-08-04 10:25:02 -0400245 * Number of bytes outstanding that are going to need csums. This is
Filipe Manana68539bd2023-10-04 11:38:53 +0100246 * used in ENOSPC accounting. Protected by 'lock'.
Josef Bacik7709cde2011-08-04 10:25:02 -0400247 */
248 u64 csum_bytes;
249
Boris Burkov77eea052021-06-30 13:01:48 -0700250 /* Backwards incompatible flags, lower half of inode_item::flags */
Josef Bacikf1bdcc02011-07-14 14:28:08 -0400251 u32 flags;
Boris Burkov77eea052021-06-30 13:01:48 -0700252 /* Read-only compatibility flags, upper half of inode_item::flags */
253 u32 ro_flags;
Josef Bacikf1bdcc02011-07-14 14:28:08 -0400254
Josef Bacik69fe2d72017-10-19 14:15:57 -0400255 struct btrfs_block_rsv block_rsv;
Josef Bacik9ed74f22009-09-11 16:12:44 -0400256
Miao Xie16cdcec2011-04-22 18:12:22 +0800257 struct btrfs_delayed_node *delayed_node;
258
chandan r9cc97d62012-07-04 12:48:07 +0530259 /* File creation time. */
David Sterbac6e8f892023-09-29 21:18:44 +0200260 u64 i_otime_sec;
261 u32 i_otime_nsec;
chandan r9cc97d62012-07-04 12:48:07 +0530262
David Sterba8089fe62015-11-19 14:15:51 +0100263 /* Hook into fs_info->delayed_iputs */
264 struct list_head delayed_iput;
David Sterba8089fe62015-11-19 14:15:51 +0100265
Josef Bacik8318ba72021-02-10 17:14:33 -0500266 struct rw_semaphore i_mmap_lock;
Chris Masond352ac62008-09-29 15:18:18 -0400267 struct inode vfs_inode;
Chris Mason2c90e5d62007-04-02 10:50:19 -0400268};
Chris Masondbe674a2008-07-17 12:54:05 -0400269
Filipe Mananafa4b8cb2023-04-05 18:51:29 +0100270static inline u64 btrfs_get_first_dir_index_to_log(const struct btrfs_inode *inode)
271{
272 return READ_ONCE(inode->first_dir_index_to_log);
273}
274
275static inline void btrfs_set_first_dir_index_to_log(struct btrfs_inode *inode,
276 u64 index)
277{
278 WRITE_ONCE(inode->first_dir_index_to_log, index);
279}
280
Jeff Mahoney9a35b632017-06-28 21:56:54 -0600281static inline struct btrfs_inode *BTRFS_I(const struct inode *inode)
Chris Mason2c90e5d62007-04-02 10:50:19 -0400282{
283 return container_of(inode, struct btrfs_inode, vfs_inode);
284}
285
Filipe David Borba Manana778ba822013-10-06 22:22:33 +0100286static inline unsigned long btrfs_inode_hash(u64 objectid,
287 const struct btrfs_root *root)
288{
Misono Tomohiro4fd786e2018-08-06 14:25:24 +0900289 u64 h = objectid ^ (root->root_key.objectid * GOLDEN_RATIO_PRIME);
Filipe David Borba Manana778ba822013-10-06 22:22:33 +0100290
291#if BITS_PER_LONG == 32
292 h = (h >> 32) ^ (h & 0xffffffff);
293#endif
294
295 return (unsigned long)h;
296}
297
Filipe Mananacf2404a2022-07-11 15:22:50 +0100298#if BITS_PER_LONG == 32
299
300/*
301 * On 32 bit systems the i_ino of struct inode is 32 bits (unsigned long), so
302 * we use the inode's location objectid which is a u64 to avoid truncation.
303 */
Jeff Mahoney9a35b632017-06-28 21:56:54 -0600304static inline u64 btrfs_ino(const struct btrfs_inode *inode)
Li Zefan33345d012011-04-20 10:31:50 +0800305{
Nikolay Borisov4a0cc7c2017-01-10 20:35:31 +0200306 u64 ino = inode->location.objectid;
Li Zefan33345d012011-04-20 10:31:50 +0800307
Filipe Mananaadac5582022-07-11 15:22:49 +0100308 /* type == BTRFS_ROOT_ITEM_KEY: subvol dir */
309 if (inode->location.type == BTRFS_ROOT_ITEM_KEY)
Nikolay Borisov4a0cc7c2017-01-10 20:35:31 +0200310 ino = inode->vfs_inode.i_ino;
Li Zefan33345d012011-04-20 10:31:50 +0800311 return ino;
312}
313
Filipe Mananacf2404a2022-07-11 15:22:50 +0100314#else
315
316static inline u64 btrfs_ino(const struct btrfs_inode *inode)
317{
318 return inode->vfs_inode.i_ino;
319}
320
321#endif
322
Nikolay Borisov6ef06d22017-02-20 13:50:34 +0200323static inline void btrfs_i_size_write(struct btrfs_inode *inode, u64 size)
Chris Masondbe674a2008-07-17 12:54:05 -0400324{
Nikolay Borisov6ef06d22017-02-20 13:50:34 +0200325 i_size_write(&inode->vfs_inode, size);
326 inode->disk_i_size = size;
Chris Masondbe674a2008-07-17 12:54:05 -0400327}
328
Nikolay Borisov70ddc552017-02-20 13:50:35 +0200329static inline bool btrfs_is_free_space_inode(struct btrfs_inode *inode)
Chris Mason2cf85722011-07-26 15:35:09 -0400330{
Josef Bacik9b9b8852022-09-14 19:04:50 -0400331 return test_bit(BTRFS_INODE_FREE_SPACE_INODE, &inode->runtime_flags);
Chris Mason2cf85722011-07-26 15:35:09 -0400332}
333
Nikolay Borisov06f25482018-11-09 16:08:30 +0200334static inline bool is_data_inode(struct inode *inode)
335{
336 return btrfs_ino(BTRFS_I(inode)) != BTRFS_BTREE_INODE_OBJECTID;
337}
338
Josef Bacik8b62f87b2017-10-19 14:15:55 -0400339static inline void btrfs_mod_outstanding_extents(struct btrfs_inode *inode,
340 int mod)
341{
342 lockdep_assert_held(&inode->lock);
343 inode->outstanding_extents += mod;
344 if (btrfs_is_free_space_inode(inode))
345 return;
Josef Bacikdd48d402017-10-19 14:15:56 -0400346 trace_btrfs_inode_mod_outstanding_extents(inode->root, btrfs_ino(inode),
Naohiro Aota64425502023-06-19 11:15:31 +0900347 mod, inode->outstanding_extents);
Josef Bacik8b62f87b2017-10-19 14:15:55 -0400348}
349
Filipe Mananabc0939f2021-02-23 12:08:48 +0000350/*
351 * Called every time after doing a buffered, direct IO or memory mapped write.
352 *
353 * This is to ensure that if we write to a file that was previously fsynced in
354 * the current transaction, then try to fsync it again in the same transaction,
355 * we will know that there were changes in the file and that it needs to be
356 * logged.
357 */
358static inline void btrfs_set_inode_last_sub_trans(struct btrfs_inode *inode)
359{
360 spin_lock(&inode->lock);
361 inode->last_sub_trans = inode->root->log_transid;
362 spin_unlock(&inode->lock);
363}
364
Filipe Manana23e33372022-02-17 12:12:06 +0000365/*
366 * Should be called while holding the inode's VFS lock in exclusive mode or in a
367 * context where no one else can access the inode concurrently (during inode
368 * creation or when loading an inode from disk).
369 */
370static inline void btrfs_set_inode_full_sync(struct btrfs_inode *inode)
371{
372 set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags);
373 /*
374 * The inode may have been part of a reflink operation in the last
375 * transaction that modified it, and then a fsync has reset the
376 * last_reflink_trans to avoid subsequent fsyncs in the same
377 * transaction to do unnecessary work. So update last_reflink_trans
378 * to the last_trans value (we have to be pessimistic and assume a
379 * reflink happened).
380 *
381 * The ->last_trans is protected by the inode's spinlock and we can
382 * have a concurrent ordered extent completion update it. Also set
383 * last_reflink_trans to ->last_trans only if the former is less than
384 * the later, because we can be called in a context where
385 * last_reflink_trans was set to the current transaction generation
386 * while ->last_trans was not yet updated in the current transaction,
387 * and therefore has a lower value.
388 */
389 spin_lock(&inode->lock);
390 if (inode->last_reflink_trans < inode->last_trans)
391 inode->last_reflink_trans = inode->last_trans;
392 spin_unlock(&inode->lock);
393}
394
Filipe Manana209ecbb2021-02-23 12:08:49 +0000395static inline bool btrfs_inode_in_log(struct btrfs_inode *inode, u64 generation)
Josef Bacik22ee6985d2012-05-29 16:57:49 -0400396{
Filipe Manana209ecbb2021-02-23 12:08:49 +0000397 bool ret = false;
Filipe Manana2f2ff0e2015-03-20 17:19:46 +0000398
Nikolay Borisov0f8939b2017-01-18 00:31:30 +0200399 spin_lock(&inode->lock);
400 if (inode->logged_trans == generation &&
401 inode->last_sub_trans <= inode->last_log_commit &&
Filipe Mananaf9850782023-10-04 11:38:48 +0100402 inode->last_sub_trans <= btrfs_get_root_last_log_commit(inode->root))
Filipe Manana209ecbb2021-02-23 12:08:49 +0000403 ret = true;
Nikolay Borisov0f8939b2017-01-18 00:31:30 +0200404 spin_unlock(&inode->lock);
Filipe Manana2f2ff0e2015-03-20 17:19:46 +0000405 return ret;
Josef Bacik22ee6985d2012-05-29 16:57:49 -0400406}
407
Chung-Chiang Chenge6f9d692022-04-15 16:04:05 +0800408/*
409 * Check if the inode has flags compatible with compression
410 */
411static inline bool btrfs_inode_can_compress(const struct btrfs_inode *inode)
412{
413 if (inode->flags & BTRFS_INODE_NODATACOW ||
414 inode->flags & BTRFS_INODE_NODATASUM)
415 return false;
416 return true;
417}
418
Johannes Thumshirn7ebc7e52019-06-03 16:58:52 +0200419/* Array of bytes with variable length, hexadecimal format 0x1234 */
420#define CSUM_FMT "0x%*phN"
421#define CSUM_FMT_VALUE(size, bytes) size, bytes
Josef Bacik2885fd62022-10-26 15:08:21 -0400422
Josef Bacik2885fd62022-10-26 15:08:21 -0400423int btrfs_check_sector_csum(struct btrfs_fs_info *fs_info, struct page *page,
424 u32 pgoff, u8 *csum, const u8 * const csum_expected);
Christoph Hellwige5219042023-01-21 07:50:06 +0100425bool btrfs_data_csum_ok(struct btrfs_bio *bbio, struct btrfs_device *dev,
426 u32 bio_offset, struct bio_vec *bv);
Josef Bacik2885fd62022-10-26 15:08:21 -0400427noinline int can_nocow_extent(struct inode *inode, u64 offset, u64 *len,
428 u64 *orig_start, u64 *orig_block_len,
429 u64 *ram_bytes, bool nowait, bool strict);
430
431void __btrfs_del_delalloc_inode(struct btrfs_root *root, struct btrfs_inode *inode);
432struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry);
433int btrfs_set_inode_index(struct btrfs_inode *dir, u64 *index);
434int btrfs_unlink_inode(struct btrfs_trans_handle *trans,
435 struct btrfs_inode *dir, struct btrfs_inode *inode,
436 const struct fscrypt_str *name);
437int btrfs_add_link(struct btrfs_trans_handle *trans,
438 struct btrfs_inode *parent_inode, struct btrfs_inode *inode,
439 const struct fscrypt_str *name, int add_backref, u64 index);
David Sterba3c4f91e2022-10-27 02:41:32 +0200440int btrfs_delete_subvolume(struct btrfs_inode *dir, struct dentry *dentry);
Josef Bacik2885fd62022-10-26 15:08:21 -0400441int btrfs_truncate_block(struct btrfs_inode *inode, loff_t from, loff_t len,
442 int front);
443
444int btrfs_start_delalloc_snapshot(struct btrfs_root *root, bool in_reclaim_context);
445int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, long nr,
446 bool in_reclaim_context);
447int btrfs_set_extent_delalloc(struct btrfs_inode *inode, u64 start, u64 end,
448 unsigned int extra_bits,
449 struct extent_state **cached_state);
450
451struct btrfs_new_inode_args {
452 /* Input */
453 struct inode *dir;
454 struct dentry *dentry;
455 struct inode *inode;
456 bool orphan;
457 bool subvol;
458
459 /* Output from btrfs_new_inode_prepare(), input to btrfs_create_new_inode(). */
460 struct posix_acl *default_acl;
461 struct posix_acl *acl;
462 struct fscrypt_name fname;
463};
464
465int btrfs_new_inode_prepare(struct btrfs_new_inode_args *args,
466 unsigned int *trans_num_items);
467int btrfs_create_new_inode(struct btrfs_trans_handle *trans,
468 struct btrfs_new_inode_args *args);
469void btrfs_new_inode_args_destroy(struct btrfs_new_inode_args *args);
Christian Braunerf2d40142023-01-13 12:49:25 +0100470struct inode *btrfs_new_subvol_inode(struct mnt_idmap *idmap,
Josef Bacik2885fd62022-10-26 15:08:21 -0400471 struct inode *dir);
David Sterba4c5d1662022-10-27 02:41:32 +0200472 void btrfs_set_delalloc_extent(struct btrfs_inode *inode, struct extent_state *state,
Josef Bacik2885fd62022-10-26 15:08:21 -0400473 u32 bits);
David Sterbabd547662022-10-27 02:41:32 +0200474void btrfs_clear_delalloc_extent(struct btrfs_inode *inode,
Josef Bacik2885fd62022-10-26 15:08:21 -0400475 struct extent_state *state, u32 bits);
David Sterba24541512022-10-27 02:41:32 +0200476void btrfs_merge_delalloc_extent(struct btrfs_inode *inode, struct extent_state *new,
Josef Bacik2885fd62022-10-26 15:08:21 -0400477 struct extent_state *other);
David Sterba62798a42022-10-27 02:41:32 +0200478void btrfs_split_delalloc_extent(struct btrfs_inode *inode,
Josef Bacik2885fd62022-10-26 15:08:21 -0400479 struct extent_state *orig, u64 split);
480void btrfs_set_range_writeback(struct btrfs_inode *inode, u64 start, u64 end);
481vm_fault_t btrfs_page_mkwrite(struct vm_fault *vmf);
482void btrfs_evict_inode(struct inode *inode);
483struct inode *btrfs_alloc_inode(struct super_block *sb);
484void btrfs_destroy_inode(struct inode *inode);
485void btrfs_free_inode(struct inode *inode);
486int btrfs_drop_inode(struct inode *inode);
487int __init btrfs_init_cachep(void);
488void __cold btrfs_destroy_cachep(void);
489struct inode *btrfs_iget_path(struct super_block *s, u64 ino,
490 struct btrfs_root *root, struct btrfs_path *path);
491struct inode *btrfs_iget(struct super_block *s, u64 ino, struct btrfs_root *root);
492struct extent_map *btrfs_get_extent(struct btrfs_inode *inode,
493 struct page *page, size_t pg_offset,
Qu Wenruo4618d0a2023-12-05 18:21:29 +1030494 u64 start, u64 len);
Josef Bacik2885fd62022-10-26 15:08:21 -0400495int btrfs_update_inode(struct btrfs_trans_handle *trans,
Filipe Manana8b9d0322023-09-22 11:37:22 +0100496 struct btrfs_inode *inode);
Josef Bacik2885fd62022-10-26 15:08:21 -0400497int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans,
Filipe Manana0a5d0dc2023-09-22 11:37:21 +0100498 struct btrfs_inode *inode);
Josef Bacik2885fd62022-10-26 15:08:21 -0400499int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct btrfs_inode *inode);
500int btrfs_orphan_cleanup(struct btrfs_root *root);
501int btrfs_cont_expand(struct btrfs_inode *inode, loff_t oldsize, loff_t size);
David Sterbae55cf7c2022-10-28 03:53:04 +0200502void btrfs_add_delayed_iput(struct btrfs_inode *inode);
Josef Bacik2885fd62022-10-26 15:08:21 -0400503void btrfs_run_delayed_iputs(struct btrfs_fs_info *fs_info);
504int btrfs_wait_on_delayed_iputs(struct btrfs_fs_info *fs_info);
505int btrfs_prealloc_file_range(struct inode *inode, int mode,
506 u64 start, u64 num_bytes, u64 min_size,
507 loff_t actual_len, u64 *alloc_hint);
508int btrfs_prealloc_file_range_trans(struct inode *inode,
509 struct btrfs_trans_handle *trans, int mode,
510 u64 start, u64 num_bytes, u64 min_size,
511 loff_t actual_len, u64 *alloc_hint);
512int btrfs_run_delalloc_range(struct btrfs_inode *inode, struct page *locked_page,
Christoph Hellwigc56cbe92023-06-28 17:31:31 +0200513 u64 start, u64 end, struct writeback_control *wbc);
Josef Bacik2885fd62022-10-26 15:08:21 -0400514int btrfs_writepage_cow_fixup(struct page *page);
Josef Bacik2885fd62022-10-26 15:08:21 -0400515int btrfs_encoded_io_compression_from_extent(struct btrfs_fs_info *fs_info,
516 int compress_type);
517int btrfs_encoded_read_regular_fill_pages(struct btrfs_inode *inode,
518 u64 file_offset, u64 disk_bytenr,
519 u64 disk_io_size,
520 struct page **pages);
521ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter,
522 struct btrfs_ioctl_encoded_io_args *encoded);
523ssize_t btrfs_do_encoded_write(struct kiocb *iocb, struct iov_iter *from,
524 const struct btrfs_ioctl_encoded_io_args *encoded);
525
526ssize_t btrfs_dio_read(struct kiocb *iocb, struct iov_iter *iter,
527 size_t done_before);
528struct iomap_dio *btrfs_dio_write(struct kiocb *iocb, struct iov_iter *iter,
529 size_t done_before);
530
531extern const struct dentry_operations btrfs_dentry_operations;
532
533/* Inode locking type flags, by default the exclusive lock is taken. */
534enum btrfs_ilock_type {
535 ENUM_BIT(BTRFS_ILOCK_SHARED),
536 ENUM_BIT(BTRFS_ILOCK_TRY),
537 ENUM_BIT(BTRFS_ILOCK_MMAP),
538};
539
David Sterba29b63522022-10-27 02:41:32 +0200540int btrfs_inode_lock(struct btrfs_inode *inode, unsigned int ilock_flags);
David Sterbae5d4d75b2022-10-27 02:41:32 +0200541void btrfs_inode_unlock(struct btrfs_inode *inode, unsigned int ilock_flags);
Josef Bacik2885fd62022-10-26 15:08:21 -0400542void btrfs_update_inode_bytes(struct btrfs_inode *inode, const u64 add_bytes,
543 const u64 del_bytes);
544void btrfs_assert_inode_range_clean(struct btrfs_inode *inode, u64 start, u64 end);
545
Chris Mason2c90e5d62007-04-02 10:50:19 -0400546#endif