| 1 | /* | 
|---|
| 2 | * Ext4 orphan inode handling | 
|---|
| 3 | */ | 
|---|
| 4 | #include <linux/fs.h> | 
|---|
| 5 | #include <linux/quotaops.h> | 
|---|
| 6 | #include <linux/buffer_head.h> | 
|---|
| 7 |  | 
|---|
| 8 | #include "ext4.h" | 
|---|
| 9 | #include "ext4_jbd2.h" | 
|---|
| 10 |  | 
|---|
| 11 | static int ext4_orphan_file_add(handle_t *handle, struct inode *inode) | 
|---|
| 12 | { | 
|---|
| 13 | int i, j, start; | 
|---|
| 14 | struct ext4_orphan_info *oi = &EXT4_SB(sb: inode->i_sb)->s_orphan_info; | 
|---|
| 15 | int ret = 0; | 
|---|
| 16 | bool found = false; | 
|---|
| 17 | __le32 *bdata; | 
|---|
| 18 | int inodes_per_ob = ext4_inodes_per_orphan_block(sb: inode->i_sb); | 
|---|
| 19 | int looped = 0; | 
|---|
| 20 |  | 
|---|
| 21 | /* | 
|---|
| 22 | * Find block with free orphan entry. Use CPU number for a naive hash | 
|---|
| 23 | * for a search start in the orphan file | 
|---|
| 24 | */ | 
|---|
| 25 | start = raw_smp_processor_id()*13 % oi->of_blocks; | 
|---|
| 26 | i = start; | 
|---|
| 27 | do { | 
|---|
| 28 | if (atomic_dec_if_positive(v: &oi->of_binfo[i].ob_free_entries) | 
|---|
| 29 | >= 0) { | 
|---|
| 30 | found = true; | 
|---|
| 31 | break; | 
|---|
| 32 | } | 
|---|
| 33 | if (++i >= oi->of_blocks) | 
|---|
| 34 | i = 0; | 
|---|
| 35 | } while (i != start); | 
|---|
| 36 |  | 
|---|
| 37 | if (!found) { | 
|---|
| 38 | /* | 
|---|
| 39 | * For now we don't grow or shrink orphan file. We just use | 
|---|
| 40 | * whatever was allocated at mke2fs time. The additional | 
|---|
| 41 | * credits we would have to reserve for each orphan inode | 
|---|
| 42 | * operation just don't seem worth it. | 
|---|
| 43 | */ | 
|---|
| 44 | return -ENOSPC; | 
|---|
| 45 | } | 
|---|
| 46 |  | 
|---|
| 47 | ret = ext4_journal_get_write_access(handle, inode->i_sb, | 
|---|
| 48 | oi->of_binfo[i].ob_bh, EXT4_JTR_ORPHAN_FILE); | 
|---|
| 49 | if (ret) { | 
|---|
| 50 | atomic_inc(v: &oi->of_binfo[i].ob_free_entries); | 
|---|
| 51 | return ret; | 
|---|
| 52 | } | 
|---|
| 53 |  | 
|---|
| 54 | bdata = (__le32 *)(oi->of_binfo[i].ob_bh->b_data); | 
|---|
| 55 | /* Find empty slot in a block */ | 
|---|
| 56 | j = 0; | 
|---|
| 57 | do { | 
|---|
| 58 | if (looped) { | 
|---|
| 59 | /* | 
|---|
| 60 | * Did we walk through the block several times without | 
|---|
| 61 | * finding free entry? It is theoretically possible | 
|---|
| 62 | * if entries get constantly allocated and freed or | 
|---|
| 63 | * if the block is corrupted. Avoid indefinite looping | 
|---|
| 64 | * and bail. We'll use orphan list instead. | 
|---|
| 65 | */ | 
|---|
| 66 | if (looped > 3) { | 
|---|
| 67 | atomic_inc(v: &oi->of_binfo[i].ob_free_entries); | 
|---|
| 68 | return -ENOSPC; | 
|---|
| 69 | } | 
|---|
| 70 | cond_resched(); | 
|---|
| 71 | } | 
|---|
| 72 | while (bdata[j]) { | 
|---|
| 73 | if (++j >= inodes_per_ob) { | 
|---|
| 74 | j = 0; | 
|---|
| 75 | looped++; | 
|---|
| 76 | } | 
|---|
| 77 | } | 
|---|
| 78 | } while (cmpxchg(&bdata[j], (__le32)0, cpu_to_le32(inode->i_ino)) != | 
|---|
| 79 | (__le32)0); | 
|---|
| 80 |  | 
|---|
| 81 | EXT4_I(inode)->i_orphan_idx = i * inodes_per_ob + j; | 
|---|
| 82 | ext4_set_inode_state(inode, bit: EXT4_STATE_ORPHAN_FILE); | 
|---|
| 83 |  | 
|---|
| 84 | return ext4_handle_dirty_metadata(handle, NULL, oi->of_binfo[i].ob_bh); | 
|---|
| 85 | } | 
|---|
| 86 |  | 
|---|
| 87 | /* | 
|---|
| 88 | * ext4_orphan_add() links an unlinked or truncated inode into a list of | 
|---|
| 89 | * such inodes, starting at the superblock, in case we crash before the | 
|---|
| 90 | * file is closed/deleted, or in case the inode truncate spans multiple | 
|---|
| 91 | * transactions and the last transaction is not recovered after a crash. | 
|---|
| 92 | * | 
|---|
| 93 | * At filesystem recovery time, we walk this list deleting unlinked | 
|---|
| 94 | * inodes and truncating linked inodes in ext4_orphan_cleanup(). | 
|---|
| 95 | * | 
|---|
| 96 | * Orphan list manipulation functions must be called under i_rwsem unless | 
|---|
| 97 | * we are just creating the inode or deleting it. | 
|---|
| 98 | */ | 
|---|
| 99 | int ext4_orphan_add(handle_t *handle, struct inode *inode) | 
|---|
| 100 | { | 
|---|
| 101 | struct super_block *sb = inode->i_sb; | 
|---|
| 102 | struct ext4_sb_info *sbi = EXT4_SB(sb); | 
|---|
| 103 | struct ext4_iloc iloc; | 
|---|
| 104 | int err = 0, rc; | 
|---|
| 105 | bool dirty = false; | 
|---|
| 106 |  | 
|---|
| 107 | if (!sbi->s_journal || is_bad_inode(inode)) | 
|---|
| 108 | return 0; | 
|---|
| 109 |  | 
|---|
| 110 | WARN_ON_ONCE(!(inode->i_state & (I_NEW | I_FREEING)) && | 
|---|
| 111 | !inode_is_locked(inode)); | 
|---|
| 112 | if (ext4_inode_orphan_tracked(inode)) | 
|---|
| 113 | return 0; | 
|---|
| 114 |  | 
|---|
| 115 | /* | 
|---|
| 116 | * Orphan handling is only valid for files with data blocks | 
|---|
| 117 | * being truncated, or files being unlinked. Note that we either | 
|---|
| 118 | * hold i_rwsem, or the inode can not be referenced from outside, | 
|---|
| 119 | * so i_nlink should not be bumped due to race | 
|---|
| 120 | */ | 
|---|
| 121 | ASSERT((S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || | 
|---|
| 122 | S_ISLNK(inode->i_mode)) || inode->i_nlink == 0); | 
|---|
| 123 |  | 
|---|
| 124 | if (sbi->s_orphan_info.of_blocks) { | 
|---|
| 125 | err = ext4_orphan_file_add(handle, inode); | 
|---|
| 126 | /* | 
|---|
| 127 | * Fallback to normal orphan list of orphan file is | 
|---|
| 128 | * out of space | 
|---|
| 129 | */ | 
|---|
| 130 | if (err != -ENOSPC) | 
|---|
| 131 | return err; | 
|---|
| 132 | } | 
|---|
| 133 |  | 
|---|
| 134 | BUFFER_TRACE(sbi->s_sbh, "get_write_access"); | 
|---|
| 135 | err = ext4_journal_get_write_access(handle, sb, sbi->s_sbh, | 
|---|
| 136 | EXT4_JTR_NONE); | 
|---|
| 137 | if (err) | 
|---|
| 138 | goto out; | 
|---|
| 139 |  | 
|---|
| 140 | err = ext4_reserve_inode_write(handle, inode, iloc: &iloc); | 
|---|
| 141 | if (err) | 
|---|
| 142 | goto out; | 
|---|
| 143 |  | 
|---|
| 144 | mutex_lock(lock: &sbi->s_orphan_lock); | 
|---|
| 145 | /* | 
|---|
| 146 | * Due to previous errors inode may be already a part of on-disk | 
|---|
| 147 | * orphan list. If so skip on-disk list modification. | 
|---|
| 148 | */ | 
|---|
| 149 | if (!NEXT_ORPHAN(inode) || NEXT_ORPHAN(inode) > | 
|---|
| 150 | (le32_to_cpu(sbi->s_es->s_inodes_count))) { | 
|---|
| 151 | /* Insert this inode at the head of the on-disk orphan list */ | 
|---|
| 152 | NEXT_ORPHAN(inode) = le32_to_cpu(sbi->s_es->s_last_orphan); | 
|---|
| 153 | lock_buffer(bh: sbi->s_sbh); | 
|---|
| 154 | sbi->s_es->s_last_orphan = cpu_to_le32(inode->i_ino); | 
|---|
| 155 | ext4_superblock_csum_set(sb); | 
|---|
| 156 | unlock_buffer(bh: sbi->s_sbh); | 
|---|
| 157 | dirty = true; | 
|---|
| 158 | } | 
|---|
| 159 | list_add(new: &EXT4_I(inode)->i_orphan, head: &sbi->s_orphan); | 
|---|
| 160 | mutex_unlock(lock: &sbi->s_orphan_lock); | 
|---|
| 161 |  | 
|---|
| 162 | if (dirty) { | 
|---|
| 163 | err = ext4_handle_dirty_metadata(handle, NULL, sbi->s_sbh); | 
|---|
| 164 | rc = ext4_mark_iloc_dirty(handle, inode, iloc: &iloc); | 
|---|
| 165 | if (!err) | 
|---|
| 166 | err = rc; | 
|---|
| 167 | if (err) { | 
|---|
| 168 | /* | 
|---|
| 169 | * We have to remove inode from in-memory list if | 
|---|
| 170 | * addition to on disk orphan list failed. Stray orphan | 
|---|
| 171 | * list entries can cause panics at unmount time. | 
|---|
| 172 | */ | 
|---|
| 173 | mutex_lock(lock: &sbi->s_orphan_lock); | 
|---|
| 174 | list_del_init(entry: &EXT4_I(inode)->i_orphan); | 
|---|
| 175 | mutex_unlock(lock: &sbi->s_orphan_lock); | 
|---|
| 176 | } | 
|---|
| 177 | } else | 
|---|
| 178 | brelse(bh: iloc.bh); | 
|---|
| 179 |  | 
|---|
| 180 | ext4_debug( "superblock will point to %lu\n", inode->i_ino); | 
|---|
| 181 | ext4_debug( "orphan inode %lu will point to %d\n", | 
|---|
| 182 | inode->i_ino, NEXT_ORPHAN(inode)); | 
|---|
| 183 | out: | 
|---|
| 184 | ext4_std_error(sb, err); | 
|---|
| 185 | return err; | 
|---|
| 186 | } | 
|---|
| 187 |  | 
|---|
| 188 | static int ext4_orphan_file_del(handle_t *handle, struct inode *inode) | 
|---|
| 189 | { | 
|---|
| 190 | struct ext4_orphan_info *oi = &EXT4_SB(sb: inode->i_sb)->s_orphan_info; | 
|---|
| 191 | __le32 *bdata; | 
|---|
| 192 | int blk, off; | 
|---|
| 193 | int inodes_per_ob = ext4_inodes_per_orphan_block(sb: inode->i_sb); | 
|---|
| 194 | int ret = 0; | 
|---|
| 195 |  | 
|---|
| 196 | if (!handle) | 
|---|
| 197 | goto out; | 
|---|
| 198 | blk = EXT4_I(inode)->i_orphan_idx / inodes_per_ob; | 
|---|
| 199 | off = EXT4_I(inode)->i_orphan_idx % inodes_per_ob; | 
|---|
| 200 | if (WARN_ON_ONCE(blk >= oi->of_blocks)) | 
|---|
| 201 | goto out; | 
|---|
| 202 |  | 
|---|
| 203 | ret = ext4_journal_get_write_access(handle, inode->i_sb, | 
|---|
| 204 | oi->of_binfo[blk].ob_bh, EXT4_JTR_ORPHAN_FILE); | 
|---|
| 205 | if (ret) | 
|---|
| 206 | goto out; | 
|---|
| 207 |  | 
|---|
| 208 | bdata = (__le32 *)(oi->of_binfo[blk].ob_bh->b_data); | 
|---|
| 209 | bdata[off] = 0; | 
|---|
| 210 | atomic_inc(v: &oi->of_binfo[blk].ob_free_entries); | 
|---|
| 211 | ret = ext4_handle_dirty_metadata(handle, NULL, oi->of_binfo[blk].ob_bh); | 
|---|
| 212 | out: | 
|---|
| 213 | ext4_clear_inode_state(inode, bit: EXT4_STATE_ORPHAN_FILE); | 
|---|
| 214 | INIT_LIST_HEAD(list: &EXT4_I(inode)->i_orphan); | 
|---|
| 215 |  | 
|---|
| 216 | return ret; | 
|---|
| 217 | } | 
|---|
| 218 |  | 
|---|
| 219 | /* | 
|---|
| 220 | * ext4_orphan_del() removes an unlinked or truncated inode from the list | 
|---|
| 221 | * of such inodes stored on disk, because it is finally being cleaned up. | 
|---|
| 222 | */ | 
|---|
| 223 | int ext4_orphan_del(handle_t *handle, struct inode *inode) | 
|---|
| 224 | { | 
|---|
| 225 | struct list_head *prev; | 
|---|
| 226 | struct ext4_inode_info *ei = EXT4_I(inode); | 
|---|
| 227 | struct ext4_sb_info *sbi = EXT4_SB(sb: inode->i_sb); | 
|---|
| 228 | __u32 ino_next; | 
|---|
| 229 | struct ext4_iloc iloc; | 
|---|
| 230 | int err = 0; | 
|---|
| 231 |  | 
|---|
| 232 | if (!sbi->s_journal && !(sbi->s_mount_state & EXT4_ORPHAN_FS)) | 
|---|
| 233 | return 0; | 
|---|
| 234 |  | 
|---|
| 235 | WARN_ON_ONCE(!(inode->i_state & (I_NEW | I_FREEING)) && | 
|---|
| 236 | !inode_is_locked(inode)); | 
|---|
| 237 | if (ext4_test_inode_state(inode, bit: EXT4_STATE_ORPHAN_FILE)) | 
|---|
| 238 | return ext4_orphan_file_del(handle, inode); | 
|---|
| 239 |  | 
|---|
| 240 | /* Do this quick check before taking global s_orphan_lock. */ | 
|---|
| 241 | if (list_empty(head: &ei->i_orphan)) | 
|---|
| 242 | return 0; | 
|---|
| 243 |  | 
|---|
| 244 | if (handle) { | 
|---|
| 245 | /* Grab inode buffer early before taking global s_orphan_lock */ | 
|---|
| 246 | err = ext4_reserve_inode_write(handle, inode, iloc: &iloc); | 
|---|
| 247 | } | 
|---|
| 248 |  | 
|---|
| 249 | mutex_lock(lock: &sbi->s_orphan_lock); | 
|---|
| 250 | ext4_debug( "remove inode %lu from orphan list\n", inode->i_ino); | 
|---|
| 251 |  | 
|---|
| 252 | prev = ei->i_orphan.prev; | 
|---|
| 253 | list_del_init(entry: &ei->i_orphan); | 
|---|
| 254 |  | 
|---|
| 255 | /* If we're on an error path, we may not have a valid | 
|---|
| 256 | * transaction handle with which to update the orphan list on | 
|---|
| 257 | * disk, but we still need to remove the inode from the linked | 
|---|
| 258 | * list in memory. */ | 
|---|
| 259 | if (!handle || err) { | 
|---|
| 260 | mutex_unlock(lock: &sbi->s_orphan_lock); | 
|---|
| 261 | goto out_err; | 
|---|
| 262 | } | 
|---|
| 263 |  | 
|---|
| 264 | ino_next = NEXT_ORPHAN(inode); | 
|---|
| 265 | if (prev == &sbi->s_orphan) { | 
|---|
| 266 | ext4_debug( "superblock will point to %u\n", ino_next); | 
|---|
| 267 | BUFFER_TRACE(sbi->s_sbh, "get_write_access"); | 
|---|
| 268 | err = ext4_journal_get_write_access(handle, inode->i_sb, | 
|---|
| 269 | sbi->s_sbh, EXT4_JTR_NONE); | 
|---|
| 270 | if (err) { | 
|---|
| 271 | mutex_unlock(lock: &sbi->s_orphan_lock); | 
|---|
| 272 | goto out_brelse; | 
|---|
| 273 | } | 
|---|
| 274 | lock_buffer(bh: sbi->s_sbh); | 
|---|
| 275 | sbi->s_es->s_last_orphan = cpu_to_le32(ino_next); | 
|---|
| 276 | ext4_superblock_csum_set(sb: inode->i_sb); | 
|---|
| 277 | unlock_buffer(bh: sbi->s_sbh); | 
|---|
| 278 | mutex_unlock(lock: &sbi->s_orphan_lock); | 
|---|
| 279 | err = ext4_handle_dirty_metadata(handle, NULL, sbi->s_sbh); | 
|---|
| 280 | } else { | 
|---|
| 281 | struct ext4_iloc iloc2; | 
|---|
| 282 | struct inode *i_prev = | 
|---|
| 283 | &list_entry(prev, struct ext4_inode_info, i_orphan)->vfs_inode; | 
|---|
| 284 |  | 
|---|
| 285 | ext4_debug( "orphan inode %lu will point to %u\n", | 
|---|
| 286 | i_prev->i_ino, ino_next); | 
|---|
| 287 | err = ext4_reserve_inode_write(handle, inode: i_prev, iloc: &iloc2); | 
|---|
| 288 | if (err) { | 
|---|
| 289 | mutex_unlock(lock: &sbi->s_orphan_lock); | 
|---|
| 290 | goto out_brelse; | 
|---|
| 291 | } | 
|---|
| 292 | NEXT_ORPHAN(i_prev) = ino_next; | 
|---|
| 293 | err = ext4_mark_iloc_dirty(handle, inode: i_prev, iloc: &iloc2); | 
|---|
| 294 | mutex_unlock(lock: &sbi->s_orphan_lock); | 
|---|
| 295 | } | 
|---|
| 296 | if (err) | 
|---|
| 297 | goto out_brelse; | 
|---|
| 298 | NEXT_ORPHAN(inode) = 0; | 
|---|
| 299 | err = ext4_mark_iloc_dirty(handle, inode, iloc: &iloc); | 
|---|
| 300 | out_err: | 
|---|
| 301 | ext4_std_error(inode->i_sb, err); | 
|---|
| 302 | return err; | 
|---|
| 303 |  | 
|---|
| 304 | out_brelse: | 
|---|
| 305 | brelse(bh: iloc.bh); | 
|---|
| 306 | goto out_err; | 
|---|
| 307 | } | 
|---|
| 308 |  | 
|---|
| 309 | #ifdef CONFIG_QUOTA | 
|---|
| 310 | static int ext4_quota_on_mount(struct super_block *sb, int type) | 
|---|
| 311 | { | 
|---|
| 312 | return dquot_quota_on_mount(sb, | 
|---|
| 313 | rcu_dereference_protected(EXT4_SB(sb)->s_qf_names[type], | 
|---|
| 314 | lockdep_is_held(&sb->s_umount)), | 
|---|
| 315 | format_id: EXT4_SB(sb)->s_jquota_fmt, type); | 
|---|
| 316 | } | 
|---|
| 317 | #endif | 
|---|
| 318 |  | 
|---|
| 319 | static void ext4_process_orphan(struct inode *inode, | 
|---|
| 320 | int *nr_truncates, int *nr_orphans) | 
|---|
| 321 | { | 
|---|
| 322 | struct super_block *sb = inode->i_sb; | 
|---|
| 323 | int ret; | 
|---|
| 324 |  | 
|---|
| 325 | dquot_initialize(inode); | 
|---|
| 326 | if (inode->i_nlink) { | 
|---|
| 327 | if (test_opt(sb, DEBUG)) | 
|---|
| 328 | ext4_msg(sb, KERN_DEBUG, | 
|---|
| 329 | "%s: truncating inode %lu to %lld bytes", | 
|---|
| 330 | __func__, inode->i_ino, inode->i_size); | 
|---|
| 331 | ext4_debug( "truncating inode %lu to %lld bytes\n", | 
|---|
| 332 | inode->i_ino, inode->i_size); | 
|---|
| 333 | inode_lock(inode); | 
|---|
| 334 | truncate_inode_pages(inode->i_mapping, inode->i_size); | 
|---|
| 335 | ret = ext4_truncate(inode); | 
|---|
| 336 | if (ret) { | 
|---|
| 337 | /* | 
|---|
| 338 | * We need to clean up the in-core orphan list | 
|---|
| 339 | * manually if ext4_truncate() failed to get a | 
|---|
| 340 | * transaction handle. | 
|---|
| 341 | */ | 
|---|
| 342 | ext4_orphan_del(NULL, inode); | 
|---|
| 343 | ext4_std_error(inode->i_sb, ret); | 
|---|
| 344 | } | 
|---|
| 345 | inode_unlock(inode); | 
|---|
| 346 | (*nr_truncates)++; | 
|---|
| 347 | } else { | 
|---|
| 348 | if (test_opt(sb, DEBUG)) | 
|---|
| 349 | ext4_msg(sb, KERN_DEBUG, | 
|---|
| 350 | "%s: deleting unreferenced inode %lu", | 
|---|
| 351 | __func__, inode->i_ino); | 
|---|
| 352 | ext4_debug( "deleting unreferenced inode %lu\n", | 
|---|
| 353 | inode->i_ino); | 
|---|
| 354 | (*nr_orphans)++; | 
|---|
| 355 | } | 
|---|
| 356 | iput(inode);  /* The delete magic happens here! */ | 
|---|
| 357 | } | 
|---|
| 358 |  | 
|---|
| 359 | /* ext4_orphan_cleanup() walks a singly-linked list of inodes (starting at | 
|---|
| 360 | * the superblock) which were deleted from all directories, but held open by | 
|---|
| 361 | * a process at the time of a crash.  We walk the list and try to delete these | 
|---|
| 362 | * inodes at recovery time (only with a read-write filesystem). | 
|---|
| 363 | * | 
|---|
| 364 | * In order to keep the orphan inode chain consistent during traversal (in | 
|---|
| 365 | * case of crash during recovery), we link each inode into the superblock | 
|---|
| 366 | * orphan list_head and handle it the same way as an inode deletion during | 
|---|
| 367 | * normal operation (which journals the operations for us). | 
|---|
| 368 | * | 
|---|
| 369 | * We only do an iget() and an iput() on each inode, which is very safe if we | 
|---|
| 370 | * accidentally point at an in-use or already deleted inode.  The worst that | 
|---|
| 371 | * can happen in this case is that we get a "bit already cleared" message from | 
|---|
| 372 | * ext4_free_inode().  The only reason we would point at a wrong inode is if | 
|---|
| 373 | * e2fsck was run on this filesystem, and it must have already done the orphan | 
|---|
| 374 | * inode cleanup for us, so we can safely abort without any further action. | 
|---|
| 375 | */ | 
|---|
| 376 | void ext4_orphan_cleanup(struct super_block *sb, struct ext4_super_block *es) | 
|---|
| 377 | { | 
|---|
| 378 | unsigned int s_flags = sb->s_flags; | 
|---|
| 379 | int nr_orphans = 0, nr_truncates = 0; | 
|---|
| 380 | struct inode *inode; | 
|---|
| 381 | int i, j; | 
|---|
| 382 | #ifdef CONFIG_QUOTA | 
|---|
| 383 | int quota_update = 0; | 
|---|
| 384 | #endif | 
|---|
| 385 | __le32 *bdata; | 
|---|
| 386 | struct ext4_orphan_info *oi = &EXT4_SB(sb)->s_orphan_info; | 
|---|
| 387 | int inodes_per_ob = ext4_inodes_per_orphan_block(sb); | 
|---|
| 388 |  | 
|---|
| 389 | if (!es->s_last_orphan && !oi->of_blocks) { | 
|---|
| 390 | ext4_debug( "no orphan inodes to clean up\n"); | 
|---|
| 391 | return; | 
|---|
| 392 | } | 
|---|
| 393 |  | 
|---|
| 394 | if (bdev_read_only(bdev: sb->s_bdev)) { | 
|---|
| 395 | ext4_msg(sb, KERN_ERR, "write access " | 
|---|
| 396 | "unavailable, skipping orphan cleanup"); | 
|---|
| 397 | return; | 
|---|
| 398 | } | 
|---|
| 399 |  | 
|---|
| 400 | /* Check if feature set would not allow a r/w mount */ | 
|---|
| 401 | if (!ext4_feature_set_ok(sb, readonly: 0)) { | 
|---|
| 402 | ext4_msg(sb, KERN_INFO, "Skipping orphan cleanup due to " | 
|---|
| 403 | "unknown ROCOMPAT features"); | 
|---|
| 404 | return; | 
|---|
| 405 | } | 
|---|
| 406 |  | 
|---|
| 407 | if (EXT4_SB(sb)->s_mount_state & EXT4_ERROR_FS) { | 
|---|
| 408 | /* don't clear list on RO mount w/ errors */ | 
|---|
| 409 | if (es->s_last_orphan && !(s_flags & SB_RDONLY)) { | 
|---|
| 410 | ext4_msg(sb, KERN_INFO, "Errors on filesystem, " | 
|---|
| 411 | "clearing orphan list."); | 
|---|
| 412 | es->s_last_orphan = 0; | 
|---|
| 413 | } | 
|---|
| 414 | ext4_debug( "Skipping orphan recovery on fs with errors.\n"); | 
|---|
| 415 | return; | 
|---|
| 416 | } | 
|---|
| 417 |  | 
|---|
| 418 | if (s_flags & SB_RDONLY) { | 
|---|
| 419 | ext4_msg(sb, KERN_INFO, "orphan cleanup on readonly fs"); | 
|---|
| 420 | sb->s_flags &= ~SB_RDONLY; | 
|---|
| 421 | } | 
|---|
| 422 | #ifdef CONFIG_QUOTA | 
|---|
| 423 | /* | 
|---|
| 424 | * Turn on quotas which were not enabled for read-only mounts if | 
|---|
| 425 | * filesystem has quota feature, so that they are updated correctly. | 
|---|
| 426 | */ | 
|---|
| 427 | if (ext4_has_feature_quota(sb) && (s_flags & SB_RDONLY)) { | 
|---|
| 428 | int ret = ext4_enable_quotas(sb); | 
|---|
| 429 |  | 
|---|
| 430 | if (!ret) | 
|---|
| 431 | quota_update = 1; | 
|---|
| 432 | else | 
|---|
| 433 | ext4_msg(sb, KERN_ERR, | 
|---|
| 434 | "Cannot turn on quotas: error %d", ret); | 
|---|
| 435 | } | 
|---|
| 436 |  | 
|---|
| 437 | /* Turn on journaled quotas used for old sytle */ | 
|---|
| 438 | for (i = 0; i < EXT4_MAXQUOTAS; i++) { | 
|---|
| 439 | if (EXT4_SB(sb)->s_qf_names[i]) { | 
|---|
| 440 | int ret = ext4_quota_on_mount(sb, type: i); | 
|---|
| 441 |  | 
|---|
| 442 | if (!ret) | 
|---|
| 443 | quota_update = 1; | 
|---|
| 444 | else | 
|---|
| 445 | ext4_msg(sb, KERN_ERR, | 
|---|
| 446 | "Cannot turn on journaled " | 
|---|
| 447 | "quota: type %d: error %d", i, ret); | 
|---|
| 448 | } | 
|---|
| 449 | } | 
|---|
| 450 | #endif | 
|---|
| 451 |  | 
|---|
| 452 | while (es->s_last_orphan) { | 
|---|
| 453 | /* | 
|---|
| 454 | * We may have encountered an error during cleanup; if | 
|---|
| 455 | * so, skip the rest. | 
|---|
| 456 | */ | 
|---|
| 457 | if (EXT4_SB(sb)->s_mount_state & EXT4_ERROR_FS) { | 
|---|
| 458 | ext4_debug( "Skipping orphan recovery on fs with errors.\n"); | 
|---|
| 459 | es->s_last_orphan = 0; | 
|---|
| 460 | break; | 
|---|
| 461 | } | 
|---|
| 462 |  | 
|---|
| 463 | inode = ext4_orphan_get(sb, le32_to_cpu(es->s_last_orphan)); | 
|---|
| 464 | if (IS_ERR(ptr: inode)) { | 
|---|
| 465 | es->s_last_orphan = 0; | 
|---|
| 466 | break; | 
|---|
| 467 | } | 
|---|
| 468 |  | 
|---|
| 469 | list_add(new: &EXT4_I(inode)->i_orphan, head: &EXT4_SB(sb)->s_orphan); | 
|---|
| 470 | ext4_process_orphan(inode, nr_truncates: &nr_truncates, nr_orphans: &nr_orphans); | 
|---|
| 471 | } | 
|---|
| 472 |  | 
|---|
| 473 | for (i = 0; i < oi->of_blocks; i++) { | 
|---|
| 474 | bdata = (__le32 *)(oi->of_binfo[i].ob_bh->b_data); | 
|---|
| 475 | for (j = 0; j < inodes_per_ob; j++) { | 
|---|
| 476 | if (!bdata[j]) | 
|---|
| 477 | continue; | 
|---|
| 478 | inode = ext4_orphan_get(sb, le32_to_cpu(bdata[j])); | 
|---|
| 479 | if (IS_ERR(ptr: inode)) | 
|---|
| 480 | continue; | 
|---|
| 481 | ext4_set_inode_state(inode, bit: EXT4_STATE_ORPHAN_FILE); | 
|---|
| 482 | EXT4_I(inode)->i_orphan_idx = i * inodes_per_ob + j; | 
|---|
| 483 | ext4_process_orphan(inode, nr_truncates: &nr_truncates, nr_orphans: &nr_orphans); | 
|---|
| 484 | } | 
|---|
| 485 | } | 
|---|
| 486 |  | 
|---|
| 487 | #define PLURAL(x) (x), ((x) == 1) ? "" : "s" | 
|---|
| 488 |  | 
|---|
| 489 | if (nr_orphans) | 
|---|
| 490 | ext4_msg(sb, KERN_INFO, "%d orphan inode%s deleted", | 
|---|
| 491 | PLURAL(nr_orphans)); | 
|---|
| 492 | if (nr_truncates) | 
|---|
| 493 | ext4_msg(sb, KERN_INFO, "%d truncate%s cleaned up", | 
|---|
| 494 | PLURAL(nr_truncates)); | 
|---|
| 495 | #ifdef CONFIG_QUOTA | 
|---|
| 496 | /* Turn off quotas if they were enabled for orphan cleanup */ | 
|---|
| 497 | if (quota_update) { | 
|---|
| 498 | for (i = 0; i < EXT4_MAXQUOTAS; i++) { | 
|---|
| 499 | if (sb_dqopt(sb)->files[i]) | 
|---|
| 500 | dquot_quota_off(sb, type: i); | 
|---|
| 501 | } | 
|---|
| 502 | } | 
|---|
| 503 | #endif | 
|---|
| 504 | sb->s_flags = s_flags; /* Restore SB_RDONLY status */ | 
|---|
| 505 | } | 
|---|
| 506 |  | 
|---|
| 507 | void ext4_release_orphan_info(struct super_block *sb) | 
|---|
| 508 | { | 
|---|
| 509 | int i; | 
|---|
| 510 | struct ext4_orphan_info *oi = &EXT4_SB(sb)->s_orphan_info; | 
|---|
| 511 |  | 
|---|
| 512 | if (!oi->of_blocks) | 
|---|
| 513 | return; | 
|---|
| 514 | for (i = 0; i < oi->of_blocks; i++) | 
|---|
| 515 | brelse(bh: oi->of_binfo[i].ob_bh); | 
|---|
| 516 | kfree(objp: oi->of_binfo); | 
|---|
| 517 | } | 
|---|
| 518 |  | 
|---|
| 519 | static struct ext4_orphan_block_tail *ext4_orphan_block_tail( | 
|---|
| 520 | struct super_block *sb, | 
|---|
| 521 | struct buffer_head *bh) | 
|---|
| 522 | { | 
|---|
| 523 | return (struct ext4_orphan_block_tail *)(bh->b_data + sb->s_blocksize - | 
|---|
| 524 | sizeof(struct ext4_orphan_block_tail)); | 
|---|
| 525 | } | 
|---|
| 526 |  | 
|---|
| 527 | static int ext4_orphan_file_block_csum_verify(struct super_block *sb, | 
|---|
| 528 | struct buffer_head *bh) | 
|---|
| 529 | { | 
|---|
| 530 | __u32 calculated; | 
|---|
| 531 | int inodes_per_ob = ext4_inodes_per_orphan_block(sb); | 
|---|
| 532 | struct ext4_orphan_info *oi = &EXT4_SB(sb)->s_orphan_info; | 
|---|
| 533 | struct ext4_orphan_block_tail *ot; | 
|---|
| 534 | __le64 dsk_block_nr = cpu_to_le64(bh->b_blocknr); | 
|---|
| 535 |  | 
|---|
| 536 | if (!ext4_has_feature_metadata_csum(sb)) | 
|---|
| 537 | return 1; | 
|---|
| 538 |  | 
|---|
| 539 | ot = ext4_orphan_block_tail(sb, bh); | 
|---|
| 540 | calculated = ext4_chksum(crc: oi->of_csum_seed, address: (__u8 *)&dsk_block_nr, | 
|---|
| 541 | length: sizeof(dsk_block_nr)); | 
|---|
| 542 | calculated = ext4_chksum(crc: calculated, address: (__u8 *)bh->b_data, | 
|---|
| 543 | length: inodes_per_ob * sizeof(__u32)); | 
|---|
| 544 | return le32_to_cpu(ot->ob_checksum) == calculated; | 
|---|
| 545 | } | 
|---|
| 546 |  | 
|---|
| 547 | /* This gets called only when checksumming is enabled */ | 
|---|
| 548 | void ext4_orphan_file_block_trigger(struct jbd2_buffer_trigger_type *triggers, | 
|---|
| 549 | struct buffer_head *bh, | 
|---|
| 550 | void *data, size_t size) | 
|---|
| 551 | { | 
|---|
| 552 | struct super_block *sb = EXT4_TRIGGER(trigger: triggers)->sb; | 
|---|
| 553 | __u32 csum; | 
|---|
| 554 | int inodes_per_ob = ext4_inodes_per_orphan_block(sb); | 
|---|
| 555 | struct ext4_orphan_info *oi = &EXT4_SB(sb)->s_orphan_info; | 
|---|
| 556 | struct ext4_orphan_block_tail *ot; | 
|---|
| 557 | __le64 dsk_block_nr = cpu_to_le64(bh->b_blocknr); | 
|---|
| 558 |  | 
|---|
| 559 | csum = ext4_chksum(crc: oi->of_csum_seed, address: (__u8 *)&dsk_block_nr, | 
|---|
| 560 | length: sizeof(dsk_block_nr)); | 
|---|
| 561 | csum = ext4_chksum(crc: csum, address: (__u8 *)data, length: inodes_per_ob * sizeof(__u32)); | 
|---|
| 562 | ot = ext4_orphan_block_tail(sb, bh); | 
|---|
| 563 | ot->ob_checksum = cpu_to_le32(csum); | 
|---|
| 564 | } | 
|---|
| 565 |  | 
|---|
| 566 | int ext4_init_orphan_info(struct super_block *sb) | 
|---|
| 567 | { | 
|---|
| 568 | struct ext4_orphan_info *oi = &EXT4_SB(sb)->s_orphan_info; | 
|---|
| 569 | struct inode *inode; | 
|---|
| 570 | int i, j; | 
|---|
| 571 | int ret; | 
|---|
| 572 | int free; | 
|---|
| 573 | __le32 *bdata; | 
|---|
| 574 | int inodes_per_ob = ext4_inodes_per_orphan_block(sb); | 
|---|
| 575 | struct ext4_orphan_block_tail *ot; | 
|---|
| 576 | ino_t orphan_ino = le32_to_cpu(EXT4_SB(sb)->s_es->s_orphan_file_inum); | 
|---|
| 577 |  | 
|---|
| 578 | if (!ext4_has_feature_orphan_file(sb)) | 
|---|
| 579 | return 0; | 
|---|
| 580 |  | 
|---|
| 581 | inode = ext4_iget(sb, orphan_ino, EXT4_IGET_SPECIAL); | 
|---|
| 582 | if (IS_ERR(ptr: inode)) { | 
|---|
| 583 | ext4_msg(sb, KERN_ERR, "get orphan inode failed"); | 
|---|
| 584 | return PTR_ERR(ptr: inode); | 
|---|
| 585 | } | 
|---|
| 586 | /* | 
|---|
| 587 | * This is just an artificial limit to prevent corrupted fs from | 
|---|
| 588 | * consuming absurd amounts of memory when pinning blocks of orphan | 
|---|
| 589 | * file in memory. | 
|---|
| 590 | */ | 
|---|
| 591 | if (inode->i_size > 8 << 20) { | 
|---|
| 592 | ext4_msg(sb, KERN_ERR, "orphan file too big: %llu", | 
|---|
| 593 | (unsigned long long)inode->i_size); | 
|---|
| 594 | ret = -EFSCORRUPTED; | 
|---|
| 595 | goto out_put; | 
|---|
| 596 | } | 
|---|
| 597 | oi->of_blocks = inode->i_size >> sb->s_blocksize_bits; | 
|---|
| 598 | oi->of_csum_seed = EXT4_I(inode)->i_csum_seed; | 
|---|
| 599 | oi->of_binfo = kvmalloc_array(oi->of_blocks, | 
|---|
| 600 | sizeof(struct ext4_orphan_block), | 
|---|
| 601 | GFP_KERNEL); | 
|---|
| 602 | if (!oi->of_binfo) { | 
|---|
| 603 | ret = -ENOMEM; | 
|---|
| 604 | goto out_put; | 
|---|
| 605 | } | 
|---|
| 606 | for (i = 0; i < oi->of_blocks; i++) { | 
|---|
| 607 | oi->of_binfo[i].ob_bh = ext4_bread(NULL, inode, i, 0); | 
|---|
| 608 | if (IS_ERR(ptr: oi->of_binfo[i].ob_bh)) { | 
|---|
| 609 | ret = PTR_ERR(ptr: oi->of_binfo[i].ob_bh); | 
|---|
| 610 | goto out_free; | 
|---|
| 611 | } | 
|---|
| 612 | if (!oi->of_binfo[i].ob_bh) { | 
|---|
| 613 | ret = -EIO; | 
|---|
| 614 | goto out_free; | 
|---|
| 615 | } | 
|---|
| 616 | ot = ext4_orphan_block_tail(sb, bh: oi->of_binfo[i].ob_bh); | 
|---|
| 617 | if (le32_to_cpu(ot->ob_magic) != EXT4_ORPHAN_BLOCK_MAGIC) { | 
|---|
| 618 | ext4_error(sb, "orphan file block %d: bad magic", i); | 
|---|
| 619 | ret = -EIO; | 
|---|
| 620 | goto out_free; | 
|---|
| 621 | } | 
|---|
| 622 | if (!ext4_orphan_file_block_csum_verify(sb, | 
|---|
| 623 | bh: oi->of_binfo[i].ob_bh)) { | 
|---|
| 624 | ext4_error(sb, "orphan file block %d: bad checksum", i); | 
|---|
| 625 | ret = -EIO; | 
|---|
| 626 | goto out_free; | 
|---|
| 627 | } | 
|---|
| 628 | bdata = (__le32 *)(oi->of_binfo[i].ob_bh->b_data); | 
|---|
| 629 | free = 0; | 
|---|
| 630 | for (j = 0; j < inodes_per_ob; j++) | 
|---|
| 631 | if (bdata[j] == 0) | 
|---|
| 632 | free++; | 
|---|
| 633 | atomic_set(v: &oi->of_binfo[i].ob_free_entries, i: free); | 
|---|
| 634 | } | 
|---|
| 635 | iput(inode); | 
|---|
| 636 | return 0; | 
|---|
| 637 | out_free: | 
|---|
| 638 | for (i--; i >= 0; i--) | 
|---|
| 639 | brelse(bh: oi->of_binfo[i].ob_bh); | 
|---|
| 640 | kfree(objp: oi->of_binfo); | 
|---|
| 641 | out_put: | 
|---|
| 642 | iput(inode); | 
|---|
| 643 | return ret; | 
|---|
| 644 | } | 
|---|
| 645 |  | 
|---|
| 646 | int ext4_orphan_file_empty(struct super_block *sb) | 
|---|
| 647 | { | 
|---|
| 648 | struct ext4_orphan_info *oi = &EXT4_SB(sb)->s_orphan_info; | 
|---|
| 649 | int i; | 
|---|
| 650 | int inodes_per_ob = ext4_inodes_per_orphan_block(sb); | 
|---|
| 651 |  | 
|---|
| 652 | if (!ext4_has_feature_orphan_file(sb)) | 
|---|
| 653 | return 1; | 
|---|
| 654 | for (i = 0; i < oi->of_blocks; i++) | 
|---|
| 655 | if (atomic_read(v: &oi->of_binfo[i].ob_free_entries) != | 
|---|
| 656 | inodes_per_ob) | 
|---|
| 657 | return 0; | 
|---|
| 658 | return 1; | 
|---|
| 659 | } | 
|---|
| 660 |  | 
|---|