David Sterba | 9888c34 | 2018-04-03 19:16:55 +0200 | [diff] [blame] | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
Chris Mason | 6cbd557 | 2007-06-12 09:07:21 -0400 | [diff] [blame] | 2 | /* |
| 3 | * Copyright (C) 2007 Oracle. All rights reserved. |
Chris Mason | 6cbd557 | 2007-06-12 09:07:21 -0400 | [diff] [blame] | 4 | */ |
| 5 | |
David Sterba | 9888c34 | 2018-04-03 19:16:55 +0200 | [diff] [blame] | 6 | #ifndef BTRFS_INODE_H |
| 7 | #define BTRFS_INODE_H |
Chris Mason | 2c90e5d6 | 2007-04-02 10:50:19 -0400 | [diff] [blame] | 8 | |
Filipe David Borba Manana | 778ba82 | 2013-10-06 22:22:33 +0100 | [diff] [blame] | 9 | #include <linux/hash.h> |
Omar Sandoval | e3b318d | 2020-04-16 14:46:20 -0700 | [diff] [blame] | 10 | #include <linux/refcount.h> |
Josef Bacik | 04cc63d | 2023-08-25 16:19:26 -0400 | [diff] [blame] | 11 | #include <linux/fscrypt.h> |
Josef Bacik | 5335f43 | 2023-08-25 16:19:28 -0400 | [diff] [blame] | 12 | #include <trace/events/btrfs.h> |
Chris Mason | a52d9a8 | 2007-08-27 16:49:44 -0400 | [diff] [blame] | 13 | #include "extent_map.h" |
Chris Mason | d1310b2 | 2008-01-24 16:13:08 -0500 | [diff] [blame] | 14 | #include "extent_io.h" |
Chris Mason | e6dcd2d | 2008-07-17 12:53:50 -0400 | [diff] [blame] | 15 | #include "ordered-data.h" |
Miao Xie | 16cdcec | 2011-04-22 18:12:22 +0800 | [diff] [blame] | 16 | #include "delayed-inode.h" |
Chris Mason | a52d9a8 | 2007-08-27 16:49:44 -0400 | [diff] [blame] | 17 | |
Josef Bacik | 72ac3c0 | 2012-05-23 14:13:11 -0400 | [diff] [blame] | 18 | /* |
Filipe Manana | 528ee69 | 2021-12-15 12:19:59 +0000 | [diff] [blame] | 19 | * 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 Bacik | 72ac3c0 | 2012-05-23 14:13:11 -0400 | [diff] [blame] | 26 | * 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 Sandoval | 7efc3e3 | 2018-05-11 13:13:39 -0700 | [diff] [blame] | 32 | enum { |
Nikolay Borisov | 1fd4033 | 2020-10-01 09:40:39 +0300 | [diff] [blame] | 33 | BTRFS_INODE_FLUSH_ON_CLOSE, |
Omar Sandoval | 7efc3e3 | 2018-05-11 13:13:39 -0700 | [diff] [blame] | 34 | BTRFS_INODE_DUMMY, |
| 35 | BTRFS_INODE_IN_DEFRAG, |
| 36 | BTRFS_INODE_HAS_ASYNC_EXTENT, |
Filipe Manana | 4877817 | 2020-08-11 12:43:58 +0100 | [diff] [blame] | 37 | /* |
| 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 Sandoval | 7efc3e3 | 2018-05-11 13:13:39 -0700 | [diff] [blame] | 42 | BTRFS_INODE_NEEDS_FULL_SYNC, |
| 43 | BTRFS_INODE_COPY_EVERYTHING, |
| 44 | BTRFS_INODE_IN_DELALLOC_LIST, |
Omar Sandoval | 7efc3e3 | 2018-05-11 13:13:39 -0700 | [diff] [blame] | 45 | BTRFS_INODE_HAS_PROPS, |
Ethan Lien | 3cd24c6 | 2018-11-01 14:49:03 +0800 | [diff] [blame] | 46 | BTRFS_INODE_SNAPSHOT_FLUSH, |
Filipe Manana | f2f121a | 2020-11-13 11:21:49 +0000 | [diff] [blame] | 47 | /* |
| 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 Manana | 3d45f22 | 2020-12-02 11:55:58 +0000 | [diff] [blame] | 54 | /* |
| 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 Burkov | 1460540 | 2021-06-30 13:01:49 -0700 | [diff] [blame] | 63 | /* |
| 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 Bacik | 9b9b885 | 2022-09-14 19:04:50 -0400 | [diff] [blame] | 70 | /* Set when this inode is a free space inode. */ |
| 71 | BTRFS_INODE_FREE_SPACE_INODE, |
Josef Bacik | ed9b50a | 2023-11-29 13:10:31 -0500 | [diff] [blame] | 72 | /* Set when there are no capabilities in XATTs for the inode. */ |
| 73 | BTRFS_INODE_NO_CAP_XATTR, |
Omar Sandoval | 7efc3e3 | 2018-05-11 13:13:39 -0700 | [diff] [blame] | 74 | }; |
Josef Bacik | 72ac3c0 | 2012-05-23 14:13:11 -0400 | [diff] [blame] | 75 | |
Aneesh | f1ace24 | 2007-06-13 16:18:26 -0400 | [diff] [blame] | 76 | /* in memory btrfs inode */ |
Chris Mason | 2c90e5d6 | 2007-04-02 10:50:19 -0400 | [diff] [blame] | 77 | struct btrfs_inode { |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 78 | /* which subvolume this inode belongs to */ |
Chris Mason | d6e4a42 | 2007-04-06 15:37:36 -0400 | [diff] [blame] | 79 | struct btrfs_root *root; |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 80 | |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 81 | /* key used to find this inode on disk. This is used by the code |
| 82 | * to read in roots of subvolumes |
| 83 | */ |
Chris Mason | d6e4a42 | 2007-04-06 15:37:36 -0400 | [diff] [blame] | 84 | struct btrfs_key location; |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 85 | |
David Sterba | e41570d | 2023-09-08 01:09:38 +0200 | [diff] [blame] | 86 | /* 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 Manana | 2f2ff0e | 2015-03-20 17:19:46 +0000 | [diff] [blame] | 95 | /* |
| 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 Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 98 | * 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 Manana | 2f2ff0e | 2015-03-20 17:19:46 +0000 | [diff] [blame] | 101 | */ |
Josef Bacik | 9e0baf6 | 2011-07-15 15:16:44 +0000 | [diff] [blame] | 102 | spinlock_t lock; |
| 103 | |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 104 | /* the extent_tree has caches of all the extent mappings to disk */ |
Chris Mason | a52d9a8 | 2007-08-27 16:49:44 -0400 | [diff] [blame] | 105 | struct extent_map_tree extent_tree; |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 106 | |
| 107 | /* the io_tree does range state (DIRTY, LOCKED etc) */ |
Chris Mason | d1310b2 | 2008-01-24 16:13:08 -0500 | [diff] [blame] | 108 | struct extent_io_tree io_tree; |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 109 | |
Josef Bacik | 41a2ee7 | 2020-01-17 09:02:21 -0500 | [diff] [blame] | 110 | /* |
| 111 | * Keep track of where the inode has extent items mapped in order to |
David Sterba | 637e6e0 | 2023-11-30 23:42:01 +0100 | [diff] [blame] | 112 | * 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 Bacik | 41a2ee7 | 2020-01-17 09:02:21 -0500 | [diff] [blame] | 115 | */ |
David Sterba | 637e6e0 | 2023-11-30 23:42:01 +0100 | [diff] [blame] | 116 | struct extent_io_tree *file_extent_tree; |
Josef Bacik | 41a2ee7 | 2020-01-17 09:02:21 -0500 | [diff] [blame] | 117 | |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 118 | /* held while logging the inode in tree-log.c */ |
Chris Mason | e02119d | 2008-09-05 16:13:11 -0400 | [diff] [blame] | 119 | struct mutex log_mutex; |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 120 | |
David Sterba | 398fb91 | 2023-09-27 21:04:32 +0200 | [diff] [blame] | 121 | /* |
| 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 Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 125 | * of extent items we've reserved metadata for. Protected by 'lock'. |
David Sterba | 398fb91 | 2023-09-27 21:04:32 +0200 | [diff] [blame] | 126 | */ |
| 127 | unsigned outstanding_extents; |
| 128 | |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 129 | /* used to order data wrt metadata */ |
David Sterba | 54c65371 | 2023-09-27 14:22:39 +0200 | [diff] [blame] | 130 | spinlock_t ordered_tree_lock; |
| 131 | struct rb_root ordered_tree; |
| 132 | struct rb_node *ordered_tree_last; |
Josef Bacik | 15ee9bc | 2007-08-10 16:22:09 -0400 | [diff] [blame] | 133 | |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 134 | /* 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 Mason | ea8c281 | 2008-08-04 23:17:27 -0400 | [diff] [blame] | 138 | struct list_head delalloc_inodes; |
| 139 | |
Yan Zheng | 5d4f98a | 2009-06-10 10:45:14 -0400 | [diff] [blame] | 140 | /* node for the red-black tree that links inodes in subvolume root */ |
| 141 | struct rb_node rb_node; |
| 142 | |
Josef Bacik | 72ac3c0 | 2012-05-23 14:13:11 -0400 | [diff] [blame] | 143 | unsigned long runtime_flags; |
| 144 | |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 145 | /* full 64 bit generation number, struct vfs_inode doesn't have a big |
| 146 | * enough field for this. |
| 147 | */ |
Chris Mason | e02119d | 2008-09-05 16:13:11 -0400 | [diff] [blame] | 148 | u64 generation; |
| 149 | |
Josef Bacik | 15ee9bc | 2007-08-10 16:22:09 -0400 | [diff] [blame] | 150 | /* |
Filipe Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 151 | * ID of the transaction handle that last modified this inode. |
| 152 | * Protected by 'lock'. |
Josef Bacik | 15ee9bc | 2007-08-10 16:22:09 -0400 | [diff] [blame] | 153 | */ |
| 154 | u64 last_trans; |
Chris Mason | 257c62e | 2009-10-13 13:21:08 -0400 | [diff] [blame] | 155 | |
| 156 | /* |
Filipe Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 157 | * ID of the transaction that last logged this inode. |
| 158 | * Protected by 'lock'. |
Chris Mason | e02119d | 2008-09-05 16:13:11 -0400 | [diff] [blame] | 159 | */ |
| 160 | u64 logged_trans; |
Chris Mason | 49eb7e4 | 2008-09-11 15:53:12 -0400 | [diff] [blame] | 161 | |
Miao Xie | bb14a59 | 2014-02-20 18:08:56 +0800 | [diff] [blame] | 162 | /* |
Filipe Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 163 | * Log transaction ID when this inode was last modified. |
| 164 | * Protected by 'lock'. |
Miao Xie | bb14a59 | 2014-02-20 18:08:56 +0800 | [diff] [blame] | 165 | */ |
| 166 | int last_sub_trans; |
| 167 | |
Filipe Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 168 | /* A local copy of root's last_log_commit. Protected by 'lock'. */ |
Miao Xie | bb14a59 | 2014-02-20 18:08:56 +0800 | [diff] [blame] | 169 | int last_log_commit; |
| 170 | |
Filipe Manana | fa4b8cb | 2023-04-05 18:51:29 +0100 | [diff] [blame] | 171 | 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 Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 175 | * only for files. Protected by 'lock'. |
Filipe Manana | fa4b8cb | 2023-04-05 18:51:29 +0100 | [diff] [blame] | 176 | */ |
| 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 Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 187 | |
Filipe Manana | dc28722 | 2021-09-16 11:32:14 +0100 | [diff] [blame] | 188 | 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 Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 193 | * is used only for files. Protected by 'lock'. |
Filipe Manana | dc28722 | 2021-09-16 11:32:14 +0100 | [diff] [blame] | 194 | */ |
| 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 Manana | a7e3b97 | 2017-04-03 10:45:46 +0100 | [diff] [blame] | 202 | |
| 203 | /* |
Filipe Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 204 | * Total number of bytes pending defrag, used by stat to check whether |
| 205 | * it needs COW. Protected by 'lock'. |
Wang Shilong | 47059d9 | 2014-07-03 18:22:07 +0800 | [diff] [blame] | 206 | */ |
| 207 | u64 defrag_bytes; |
| 208 | |
| 209 | /* |
Filipe Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 210 | * The size of the file stored in the metadata on disk. data=ordered |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 211 | * means the in-memory i_size might be larger than the size on disk |
Filipe Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 212 | * because not all the blocks are written yet. Protected by 'lock'. |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 213 | */ |
Chris Mason | dbe674a | 2008-07-17 12:54:05 -0400 | [diff] [blame] | 214 | u64 disk_i_size; |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 215 | |
Josef Bacik | aec7477 | 2008-07-24 12:12:38 -0400 | [diff] [blame] | 216 | /* |
Filipe Manana | 528ee69 | 2021-12-15 12:19:59 +0000 | [diff] [blame] | 217 | * 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 Bacik | aec7477 | 2008-07-24 12:12:38 -0400 | [diff] [blame] | 220 | */ |
| 221 | u64 index_cnt; |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 222 | |
Miao Xie | 67de117 | 2013-12-26 13:07:06 +0800 | [diff] [blame] | 223 | /* Cache the directory index number to speed the dir/file remove */ |
| 224 | u64 dir_index; |
| 225 | |
Chris Mason | 12fcfd2 | 2009-03-24 10:24:20 -0400 | [diff] [blame] | 226 | /* 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 Bacik | 7709cde | 2011-08-04 10:25:02 -0400 | [diff] [blame] | 233 | /* |
Filipe Manana | 3ebac17 | 2020-07-15 12:30:43 +0100 | [diff] [blame] | 234 | * 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 Bacik | 7709cde | 2011-08-04 10:25:02 -0400 | [diff] [blame] | 245 | * Number of bytes outstanding that are going to need csums. This is |
Filipe Manana | 68539bd | 2023-10-04 11:38:53 +0100 | [diff] [blame] | 246 | * used in ENOSPC accounting. Protected by 'lock'. |
Josef Bacik | 7709cde | 2011-08-04 10:25:02 -0400 | [diff] [blame] | 247 | */ |
| 248 | u64 csum_bytes; |
| 249 | |
Boris Burkov | 77eea05 | 2021-06-30 13:01:48 -0700 | [diff] [blame] | 250 | /* Backwards incompatible flags, lower half of inode_item::flags */ |
Josef Bacik | f1bdcc0 | 2011-07-14 14:28:08 -0400 | [diff] [blame] | 251 | u32 flags; |
Boris Burkov | 77eea05 | 2021-06-30 13:01:48 -0700 | [diff] [blame] | 252 | /* Read-only compatibility flags, upper half of inode_item::flags */ |
| 253 | u32 ro_flags; |
Josef Bacik | f1bdcc0 | 2011-07-14 14:28:08 -0400 | [diff] [blame] | 254 | |
Josef Bacik | 69fe2d7 | 2017-10-19 14:15:57 -0400 | [diff] [blame] | 255 | struct btrfs_block_rsv block_rsv; |
Josef Bacik | 9ed74f2 | 2009-09-11 16:12:44 -0400 | [diff] [blame] | 256 | |
Miao Xie | 16cdcec | 2011-04-22 18:12:22 +0800 | [diff] [blame] | 257 | struct btrfs_delayed_node *delayed_node; |
| 258 | |
chandan r | 9cc97d6 | 2012-07-04 12:48:07 +0530 | [diff] [blame] | 259 | /* File creation time. */ |
David Sterba | c6e8f89 | 2023-09-29 21:18:44 +0200 | [diff] [blame] | 260 | u64 i_otime_sec; |
| 261 | u32 i_otime_nsec; |
chandan r | 9cc97d6 | 2012-07-04 12:48:07 +0530 | [diff] [blame] | 262 | |
David Sterba | 8089fe6 | 2015-11-19 14:15:51 +0100 | [diff] [blame] | 263 | /* Hook into fs_info->delayed_iputs */ |
| 264 | struct list_head delayed_iput; |
David Sterba | 8089fe6 | 2015-11-19 14:15:51 +0100 | [diff] [blame] | 265 | |
Josef Bacik | 8318ba7 | 2021-02-10 17:14:33 -0500 | [diff] [blame] | 266 | struct rw_semaphore i_mmap_lock; |
Chris Mason | d352ac6 | 2008-09-29 15:18:18 -0400 | [diff] [blame] | 267 | struct inode vfs_inode; |
Chris Mason | 2c90e5d6 | 2007-04-02 10:50:19 -0400 | [diff] [blame] | 268 | }; |
Chris Mason | dbe674a | 2008-07-17 12:54:05 -0400 | [diff] [blame] | 269 | |
Filipe Manana | fa4b8cb | 2023-04-05 18:51:29 +0100 | [diff] [blame] | 270 | static 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 | |
| 275 | static 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 Mahoney | 9a35b63 | 2017-06-28 21:56:54 -0600 | [diff] [blame] | 281 | static inline struct btrfs_inode *BTRFS_I(const struct inode *inode) |
Chris Mason | 2c90e5d6 | 2007-04-02 10:50:19 -0400 | [diff] [blame] | 282 | { |
| 283 | return container_of(inode, struct btrfs_inode, vfs_inode); |
| 284 | } |
| 285 | |
Filipe David Borba Manana | 778ba82 | 2013-10-06 22:22:33 +0100 | [diff] [blame] | 286 | static inline unsigned long btrfs_inode_hash(u64 objectid, |
| 287 | const struct btrfs_root *root) |
| 288 | { |
Misono Tomohiro | 4fd786e | 2018-08-06 14:25:24 +0900 | [diff] [blame] | 289 | u64 h = objectid ^ (root->root_key.objectid * GOLDEN_RATIO_PRIME); |
Filipe David Borba Manana | 778ba82 | 2013-10-06 22:22:33 +0100 | [diff] [blame] | 290 | |
| 291 | #if BITS_PER_LONG == 32 |
| 292 | h = (h >> 32) ^ (h & 0xffffffff); |
| 293 | #endif |
| 294 | |
| 295 | return (unsigned long)h; |
| 296 | } |
| 297 | |
Filipe Manana | cf2404a | 2022-07-11 15:22:50 +0100 | [diff] [blame] | 298 | #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 Mahoney | 9a35b63 | 2017-06-28 21:56:54 -0600 | [diff] [blame] | 304 | static inline u64 btrfs_ino(const struct btrfs_inode *inode) |
Li Zefan | 33345d01 | 2011-04-20 10:31:50 +0800 | [diff] [blame] | 305 | { |
Nikolay Borisov | 4a0cc7c | 2017-01-10 20:35:31 +0200 | [diff] [blame] | 306 | u64 ino = inode->location.objectid; |
Li Zefan | 33345d01 | 2011-04-20 10:31:50 +0800 | [diff] [blame] | 307 | |
Filipe Manana | adac558 | 2022-07-11 15:22:49 +0100 | [diff] [blame] | 308 | /* type == BTRFS_ROOT_ITEM_KEY: subvol dir */ |
| 309 | if (inode->location.type == BTRFS_ROOT_ITEM_KEY) |
Nikolay Borisov | 4a0cc7c | 2017-01-10 20:35:31 +0200 | [diff] [blame] | 310 | ino = inode->vfs_inode.i_ino; |
Li Zefan | 33345d01 | 2011-04-20 10:31:50 +0800 | [diff] [blame] | 311 | return ino; |
| 312 | } |
| 313 | |
Filipe Manana | cf2404a | 2022-07-11 15:22:50 +0100 | [diff] [blame] | 314 | #else |
| 315 | |
| 316 | static inline u64 btrfs_ino(const struct btrfs_inode *inode) |
| 317 | { |
| 318 | return inode->vfs_inode.i_ino; |
| 319 | } |
| 320 | |
| 321 | #endif |
| 322 | |
Nikolay Borisov | 6ef06d2 | 2017-02-20 13:50:34 +0200 | [diff] [blame] | 323 | static inline void btrfs_i_size_write(struct btrfs_inode *inode, u64 size) |
Chris Mason | dbe674a | 2008-07-17 12:54:05 -0400 | [diff] [blame] | 324 | { |
Nikolay Borisov | 6ef06d2 | 2017-02-20 13:50:34 +0200 | [diff] [blame] | 325 | i_size_write(&inode->vfs_inode, size); |
| 326 | inode->disk_i_size = size; |
Chris Mason | dbe674a | 2008-07-17 12:54:05 -0400 | [diff] [blame] | 327 | } |
| 328 | |
Nikolay Borisov | 70ddc55 | 2017-02-20 13:50:35 +0200 | [diff] [blame] | 329 | static inline bool btrfs_is_free_space_inode(struct btrfs_inode *inode) |
Chris Mason | 2cf8572 | 2011-07-26 15:35:09 -0400 | [diff] [blame] | 330 | { |
Josef Bacik | 9b9b885 | 2022-09-14 19:04:50 -0400 | [diff] [blame] | 331 | return test_bit(BTRFS_INODE_FREE_SPACE_INODE, &inode->runtime_flags); |
Chris Mason | 2cf8572 | 2011-07-26 15:35:09 -0400 | [diff] [blame] | 332 | } |
| 333 | |
Nikolay Borisov | 06f2548 | 2018-11-09 16:08:30 +0200 | [diff] [blame] | 334 | static inline bool is_data_inode(struct inode *inode) |
| 335 | { |
| 336 | return btrfs_ino(BTRFS_I(inode)) != BTRFS_BTREE_INODE_OBJECTID; |
| 337 | } |
| 338 | |
Josef Bacik | 8b62f87b | 2017-10-19 14:15:55 -0400 | [diff] [blame] | 339 | static 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 Bacik | dd48d40 | 2017-10-19 14:15:56 -0400 | [diff] [blame] | 346 | trace_btrfs_inode_mod_outstanding_extents(inode->root, btrfs_ino(inode), |
Naohiro Aota | 6442550 | 2023-06-19 11:15:31 +0900 | [diff] [blame] | 347 | mod, inode->outstanding_extents); |
Josef Bacik | 8b62f87b | 2017-10-19 14:15:55 -0400 | [diff] [blame] | 348 | } |
| 349 | |
Filipe Manana | bc0939f | 2021-02-23 12:08:48 +0000 | [diff] [blame] | 350 | /* |
| 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 | */ |
| 358 | static 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 Manana | 23e3337 | 2022-02-17 12:12:06 +0000 | [diff] [blame] | 365 | /* |
| 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 | */ |
| 370 | static 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 Manana | 209ecbb | 2021-02-23 12:08:49 +0000 | [diff] [blame] | 395 | static inline bool btrfs_inode_in_log(struct btrfs_inode *inode, u64 generation) |
Josef Bacik | 22ee6985d | 2012-05-29 16:57:49 -0400 | [diff] [blame] | 396 | { |
Filipe Manana | 209ecbb | 2021-02-23 12:08:49 +0000 | [diff] [blame] | 397 | bool ret = false; |
Filipe Manana | 2f2ff0e | 2015-03-20 17:19:46 +0000 | [diff] [blame] | 398 | |
Nikolay Borisov | 0f8939b | 2017-01-18 00:31:30 +0200 | [diff] [blame] | 399 | spin_lock(&inode->lock); |
| 400 | if (inode->logged_trans == generation && |
| 401 | inode->last_sub_trans <= inode->last_log_commit && |
Filipe Manana | f985078 | 2023-10-04 11:38:48 +0100 | [diff] [blame] | 402 | inode->last_sub_trans <= btrfs_get_root_last_log_commit(inode->root)) |
Filipe Manana | 209ecbb | 2021-02-23 12:08:49 +0000 | [diff] [blame] | 403 | ret = true; |
Nikolay Borisov | 0f8939b | 2017-01-18 00:31:30 +0200 | [diff] [blame] | 404 | spin_unlock(&inode->lock); |
Filipe Manana | 2f2ff0e | 2015-03-20 17:19:46 +0000 | [diff] [blame] | 405 | return ret; |
Josef Bacik | 22ee6985d | 2012-05-29 16:57:49 -0400 | [diff] [blame] | 406 | } |
| 407 | |
Chung-Chiang Cheng | e6f9d69 | 2022-04-15 16:04:05 +0800 | [diff] [blame] | 408 | /* |
| 409 | * Check if the inode has flags compatible with compression |
| 410 | */ |
| 411 | static 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 Thumshirn | 7ebc7e5 | 2019-06-03 16:58:52 +0200 | [diff] [blame] | 419 | /* 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 Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 422 | |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 423 | int btrfs_check_sector_csum(struct btrfs_fs_info *fs_info, struct page *page, |
| 424 | u32 pgoff, u8 *csum, const u8 * const csum_expected); |
Christoph Hellwig | e521904 | 2023-01-21 07:50:06 +0100 | [diff] [blame] | 425 | bool btrfs_data_csum_ok(struct btrfs_bio *bbio, struct btrfs_device *dev, |
| 426 | u32 bio_offset, struct bio_vec *bv); |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 427 | noinline 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 | |
| 431 | void __btrfs_del_delalloc_inode(struct btrfs_root *root, struct btrfs_inode *inode); |
| 432 | struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry); |
| 433 | int btrfs_set_inode_index(struct btrfs_inode *dir, u64 *index); |
| 434 | int btrfs_unlink_inode(struct btrfs_trans_handle *trans, |
| 435 | struct btrfs_inode *dir, struct btrfs_inode *inode, |
| 436 | const struct fscrypt_str *name); |
| 437 | int 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 Sterba | 3c4f91e | 2022-10-27 02:41:32 +0200 | [diff] [blame] | 440 | int btrfs_delete_subvolume(struct btrfs_inode *dir, struct dentry *dentry); |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 441 | int btrfs_truncate_block(struct btrfs_inode *inode, loff_t from, loff_t len, |
| 442 | int front); |
| 443 | |
| 444 | int btrfs_start_delalloc_snapshot(struct btrfs_root *root, bool in_reclaim_context); |
| 445 | int btrfs_start_delalloc_roots(struct btrfs_fs_info *fs_info, long nr, |
| 446 | bool in_reclaim_context); |
| 447 | int btrfs_set_extent_delalloc(struct btrfs_inode *inode, u64 start, u64 end, |
| 448 | unsigned int extra_bits, |
| 449 | struct extent_state **cached_state); |
| 450 | |
| 451 | struct 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 | |
| 465 | int btrfs_new_inode_prepare(struct btrfs_new_inode_args *args, |
| 466 | unsigned int *trans_num_items); |
| 467 | int btrfs_create_new_inode(struct btrfs_trans_handle *trans, |
| 468 | struct btrfs_new_inode_args *args); |
| 469 | void btrfs_new_inode_args_destroy(struct btrfs_new_inode_args *args); |
Christian Brauner | f2d4014 | 2023-01-13 12:49:25 +0100 | [diff] [blame] | 470 | struct inode *btrfs_new_subvol_inode(struct mnt_idmap *idmap, |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 471 | struct inode *dir); |
David Sterba | 4c5d166 | 2022-10-27 02:41:32 +0200 | [diff] [blame] | 472 | void btrfs_set_delalloc_extent(struct btrfs_inode *inode, struct extent_state *state, |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 473 | u32 bits); |
David Sterba | bd54766 | 2022-10-27 02:41:32 +0200 | [diff] [blame] | 474 | void btrfs_clear_delalloc_extent(struct btrfs_inode *inode, |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 475 | struct extent_state *state, u32 bits); |
David Sterba | 2454151 | 2022-10-27 02:41:32 +0200 | [diff] [blame] | 476 | void btrfs_merge_delalloc_extent(struct btrfs_inode *inode, struct extent_state *new, |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 477 | struct extent_state *other); |
David Sterba | 62798a4 | 2022-10-27 02:41:32 +0200 | [diff] [blame] | 478 | void btrfs_split_delalloc_extent(struct btrfs_inode *inode, |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 479 | struct extent_state *orig, u64 split); |
| 480 | void btrfs_set_range_writeback(struct btrfs_inode *inode, u64 start, u64 end); |
| 481 | vm_fault_t btrfs_page_mkwrite(struct vm_fault *vmf); |
| 482 | void btrfs_evict_inode(struct inode *inode); |
| 483 | struct inode *btrfs_alloc_inode(struct super_block *sb); |
| 484 | void btrfs_destroy_inode(struct inode *inode); |
| 485 | void btrfs_free_inode(struct inode *inode); |
| 486 | int btrfs_drop_inode(struct inode *inode); |
| 487 | int __init btrfs_init_cachep(void); |
| 488 | void __cold btrfs_destroy_cachep(void); |
| 489 | struct inode *btrfs_iget_path(struct super_block *s, u64 ino, |
| 490 | struct btrfs_root *root, struct btrfs_path *path); |
| 491 | struct inode *btrfs_iget(struct super_block *s, u64 ino, struct btrfs_root *root); |
| 492 | struct extent_map *btrfs_get_extent(struct btrfs_inode *inode, |
| 493 | struct page *page, size_t pg_offset, |
Qu Wenruo | 4618d0a | 2023-12-05 18:21:29 +1030 | [diff] [blame] | 494 | u64 start, u64 len); |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 495 | int btrfs_update_inode(struct btrfs_trans_handle *trans, |
Filipe Manana | 8b9d032 | 2023-09-22 11:37:22 +0100 | [diff] [blame] | 496 | struct btrfs_inode *inode); |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 497 | int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans, |
Filipe Manana | 0a5d0dc | 2023-09-22 11:37:21 +0100 | [diff] [blame] | 498 | struct btrfs_inode *inode); |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 499 | int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct btrfs_inode *inode); |
| 500 | int btrfs_orphan_cleanup(struct btrfs_root *root); |
| 501 | int btrfs_cont_expand(struct btrfs_inode *inode, loff_t oldsize, loff_t size); |
David Sterba | e55cf7c | 2022-10-28 03:53:04 +0200 | [diff] [blame] | 502 | void btrfs_add_delayed_iput(struct btrfs_inode *inode); |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 503 | void btrfs_run_delayed_iputs(struct btrfs_fs_info *fs_info); |
| 504 | int btrfs_wait_on_delayed_iputs(struct btrfs_fs_info *fs_info); |
| 505 | int 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); |
| 508 | int 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); |
| 512 | int btrfs_run_delalloc_range(struct btrfs_inode *inode, struct page *locked_page, |
Christoph Hellwig | c56cbe9 | 2023-06-28 17:31:31 +0200 | [diff] [blame] | 513 | u64 start, u64 end, struct writeback_control *wbc); |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 514 | int btrfs_writepage_cow_fixup(struct page *page); |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 515 | int btrfs_encoded_io_compression_from_extent(struct btrfs_fs_info *fs_info, |
| 516 | int compress_type); |
| 517 | int 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); |
| 521 | ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter, |
| 522 | struct btrfs_ioctl_encoded_io_args *encoded); |
| 523 | ssize_t btrfs_do_encoded_write(struct kiocb *iocb, struct iov_iter *from, |
| 524 | const struct btrfs_ioctl_encoded_io_args *encoded); |
| 525 | |
| 526 | ssize_t btrfs_dio_read(struct kiocb *iocb, struct iov_iter *iter, |
| 527 | size_t done_before); |
| 528 | struct iomap_dio *btrfs_dio_write(struct kiocb *iocb, struct iov_iter *iter, |
| 529 | size_t done_before); |
| 530 | |
| 531 | extern const struct dentry_operations btrfs_dentry_operations; |
| 532 | |
| 533 | /* Inode locking type flags, by default the exclusive lock is taken. */ |
| 534 | enum btrfs_ilock_type { |
| 535 | ENUM_BIT(BTRFS_ILOCK_SHARED), |
| 536 | ENUM_BIT(BTRFS_ILOCK_TRY), |
| 537 | ENUM_BIT(BTRFS_ILOCK_MMAP), |
| 538 | }; |
| 539 | |
David Sterba | 29b6352 | 2022-10-27 02:41:32 +0200 | [diff] [blame] | 540 | int btrfs_inode_lock(struct btrfs_inode *inode, unsigned int ilock_flags); |
David Sterba | e5d4d75b | 2022-10-27 02:41:32 +0200 | [diff] [blame] | 541 | void btrfs_inode_unlock(struct btrfs_inode *inode, unsigned int ilock_flags); |
Josef Bacik | 2885fd6 | 2022-10-26 15:08:21 -0400 | [diff] [blame] | 542 | void btrfs_update_inode_bytes(struct btrfs_inode *inode, const u64 add_bytes, |
| 543 | const u64 del_bytes); |
| 544 | void btrfs_assert_inode_range_clean(struct btrfs_inode *inode, u64 start, u64 end); |
| 545 | |
Chris Mason | 2c90e5d6 | 2007-04-02 10:50:19 -0400 | [diff] [blame] | 546 | #endif |