]>
Commit | Line | Data |
---|---|---|
328970de | 1 | /* SPDX-License-Identifier: GPL-2.0-or-later */ |
fa60ce2c | 2 | /* |
ccd979bd MF |
3 | * ocfs2.h |
4 | * | |
5 | * Defines macros and structures used in OCFS2 | |
6 | * | |
7 | * Copyright (C) 2002, 2004 Oracle. All rights reserved. | |
ccd979bd MF |
8 | */ |
9 | ||
10 | #ifndef OCFS2_H | |
11 | #define OCFS2_H | |
12 | ||
13 | #include <linux/spinlock.h> | |
14 | #include <linux/sched.h> | |
15 | #include <linux/wait.h> | |
16 | #include <linux/list.h> | |
e3a767b6 | 17 | #include <linux/llist.h> |
ccd979bd MF |
18 | #include <linux/rbtree.h> |
19 | #include <linux/workqueue.h> | |
20 | #include <linux/kref.h> | |
c74ec2f7 | 21 | #include <linux/mutex.h> |
cb25797d | 22 | #include <linux/lockdep.h> |
7aee47b0 | 23 | #include <linux/jbd2.h> |
ccd979bd | 24 | |
8f2c9c1b JB |
25 | /* For union ocfs2_dlm_lksb */ |
26 | #include "stackglue.h" | |
ccd979bd MF |
27 | |
28 | #include "ocfs2_fs.h" | |
ccd979bd | 29 | #include "ocfs2_lockid.h" |
9df5778e | 30 | #include "ocfs2_ioctl.h" |
ccd979bd | 31 | |
73be192b JB |
32 | /* For struct ocfs2_blockcheck_stats */ |
33 | #include "blockcheck.h" | |
34 | ||
d02f00cc | 35 | #include "reservations.h" |
47460d65 | 36 | |
5f483c4a GH |
37 | #include "filecheck.h" |
38 | ||
47460d65 JB |
39 | /* Caching of metadata buffers */ |
40 | ||
ccd979bd MF |
41 | /* Most user visible OCFS2 inodes will have very few pieces of |
42 | * metadata, but larger files (including bitmaps, etc) must be taken | |
43 | * into account when designing an access scheme. We allow a small | |
44 | * amount of inlined blocks to be stored on an array and grow the | |
45 | * structure into a rb tree when necessary. */ | |
47460d65 JB |
46 | #define OCFS2_CACHE_INFO_MAX_ARRAY 2 |
47 | ||
48 | /* Flags for ocfs2_caching_info */ | |
49 | ||
50 | enum ocfs2_caching_info_flags { | |
51 | /* Indicates that the metadata cache is using the inline array */ | |
52 | OCFS2_CACHE_FL_INLINE = 1<<1, | |
53 | }; | |
ccd979bd | 54 | |
6e5a3d75 | 55 | struct ocfs2_caching_operations; |
ccd979bd | 56 | struct ocfs2_caching_info { |
47460d65 JB |
57 | /* |
58 | * The parent structure provides the locks, but because the | |
6e5a3d75 JB |
59 | * parent structure can differ, it provides locking operations |
60 | * to struct ocfs2_caching_info. | |
47460d65 | 61 | */ |
6e5a3d75 | 62 | const struct ocfs2_caching_operations *ci_ops; |
47460d65 | 63 | |
292dd27e JB |
64 | /* next two are protected by trans_inc_lock */ |
65 | /* which transaction were we created on? Zero if none. */ | |
66 | unsigned long ci_created_trans; | |
66fb345d JB |
67 | /* last transaction we were a part of. */ |
68 | unsigned long ci_last_trans; | |
69 | ||
70 | /* Cache structures */ | |
47460d65 | 71 | unsigned int ci_flags; |
ccd979bd MF |
72 | unsigned int ci_num_cached; |
73 | union { | |
47460d65 | 74 | sector_t ci_array[OCFS2_CACHE_INFO_MAX_ARRAY]; |
ccd979bd MF |
75 | struct rb_root ci_tree; |
76 | } ci_cache; | |
77 | }; | |
0cf2f763 JB |
78 | /* |
79 | * Need this prototype here instead of in uptodate.h because journal.h | |
80 | * uses it. | |
81 | */ | |
82 | struct super_block *ocfs2_metadata_cache_get_super(struct ocfs2_caching_info *ci); | |
ccd979bd MF |
83 | |
84 | /* this limits us to 256 nodes | |
85 | * if we need more, we can do a kmalloc for the map */ | |
86 | #define OCFS2_NODE_MAP_MAX_NODES 256 | |
87 | struct ocfs2_node_map { | |
88 | u16 num_nodes; | |
89 | unsigned long map[BITS_TO_LONGS(OCFS2_NODE_MAP_MAX_NODES)]; | |
90 | }; | |
91 | ||
92 | enum ocfs2_ast_action { | |
93 | OCFS2_AST_INVALID = 0, | |
94 | OCFS2_AST_ATTACH, | |
95 | OCFS2_AST_CONVERT, | |
96 | OCFS2_AST_DOWNCONVERT, | |
97 | }; | |
98 | ||
99 | /* actions for an unlockast function to take. */ | |
100 | enum ocfs2_unlock_action { | |
101 | OCFS2_UNLOCK_INVALID = 0, | |
102 | OCFS2_UNLOCK_CANCEL_CONVERT, | |
103 | OCFS2_UNLOCK_DROP_LOCK, | |
104 | }; | |
105 | ||
106 | /* ocfs2_lock_res->l_flags flags. */ | |
3b5da018 | 107 | #define OCFS2_LOCK_ATTACHED (0x00000001) /* we have initialized |
ccd979bd MF |
108 | * the lvb */ |
109 | #define OCFS2_LOCK_BUSY (0x00000002) /* we are currently in | |
110 | * dlm_lock */ | |
111 | #define OCFS2_LOCK_BLOCKED (0x00000004) /* blocked waiting to | |
112 | * downconvert*/ | |
113 | #define OCFS2_LOCK_LOCAL (0x00000008) /* newly created inode */ | |
114 | #define OCFS2_LOCK_NEEDS_REFRESH (0x00000010) | |
115 | #define OCFS2_LOCK_REFRESHING (0x00000020) | |
116 | #define OCFS2_LOCK_INITIALIZED (0x00000040) /* track initialization | |
117 | * for shutdown paths */ | |
118 | #define OCFS2_LOCK_FREEING (0x00000080) /* help dlmglue track | |
119 | * when to skip queueing | |
120 | * a lock because it's | |
121 | * about to be | |
122 | * dropped. */ | |
123 | #define OCFS2_LOCK_QUEUED (0x00000100) /* queued for downconvert */ | |
cf8e06f1 | 124 | #define OCFS2_LOCK_NOCACHE (0x00000200) /* don't use a holder count */ |
de551246 JB |
125 | #define OCFS2_LOCK_PENDING (0x00000400) /* This lockres is pending a |
126 | call to dlm_lock. Only | |
127 | exists with BUSY set. */ | |
a1912826 SM |
128 | #define OCFS2_LOCK_UPCONVERT_FINISHING (0x00000800) /* blocks the dc thread |
129 | * from downconverting | |
130 | * before the upconvert | |
131 | * has completed */ | |
ccd979bd | 132 | |
d1e78238 X |
133 | #define OCFS2_LOCK_NONBLOCK_FINISHED (0x00001000) /* NONBLOCK cluster |
134 | * lock has already | |
135 | * returned, do not block | |
136 | * dc thread from | |
137 | * downconverting */ | |
138 | ||
ccd979bd MF |
139 | struct ocfs2_lock_res_ops; |
140 | ||
141 | typedef void (*ocfs2_lock_callback)(int status, unsigned long data); | |
142 | ||
5bc970e8 SM |
143 | #ifdef CONFIG_OCFS2_FS_STATS |
144 | struct ocfs2_lock_stats { | |
145 | u64 ls_total; /* Total wait in NSEC */ | |
146 | u32 ls_gets; /* Num acquires */ | |
147 | u32 ls_fail; /* Num failed acquires */ | |
148 | ||
149 | /* Storing max wait in usecs saves 24 bytes per inode */ | |
150 | u32 ls_max; /* Max wait in USEC */ | |
8a7f5f4c | 151 | u64 ls_last; /* Last unlock time in USEC */ |
5bc970e8 SM |
152 | }; |
153 | #endif | |
154 | ||
ccd979bd MF |
155 | struct ocfs2_lock_res { |
156 | void *l_priv; | |
157 | struct ocfs2_lock_res_ops *l_ops; | |
83fd9c7f | 158 | |
ccd979bd MF |
159 | |
160 | struct list_head l_blocked_list; | |
161 | struct list_head l_mask_waiters; | |
439a36b8 | 162 | struct list_head l_holders; |
ccd979bd | 163 | |
ccd979bd MF |
164 | unsigned long l_flags; |
165 | char l_name[OCFS2_LOCK_ID_MAX_LEN]; | |
ccd979bd MF |
166 | unsigned int l_ro_holders; |
167 | unsigned int l_ex_holders; | |
a2a2f552 MM |
168 | signed char l_level; |
169 | signed char l_requested; | |
170 | signed char l_blocking; | |
83fd9c7f GR |
171 | |
172 | /* Data packed - type enum ocfs2_lock_type */ | |
173 | unsigned char l_type; | |
ccd979bd MF |
174 | |
175 | /* used from AST/BAST funcs. */ | |
83fd9c7f GR |
176 | /* Data packed - enum type ocfs2_ast_action */ |
177 | unsigned char l_action; | |
178 | /* Data packed - enum type ocfs2_unlock_action */ | |
179 | unsigned char l_unlock_action; | |
de551246 | 180 | unsigned int l_pending_gen; |
ccd979bd | 181 | |
83fd9c7f GR |
182 | spinlock_t l_lock; |
183 | ||
184 | struct ocfs2_dlm_lksb l_lksb; | |
185 | ||
ccd979bd MF |
186 | wait_queue_head_t l_event; |
187 | ||
188 | struct list_head l_debug_list; | |
8ddb7b00 SM |
189 | |
190 | #ifdef CONFIG_OCFS2_FS_STATS | |
5bc970e8 SM |
191 | struct ocfs2_lock_stats l_lock_prmode; /* PR mode stats */ |
192 | u32 l_lock_refresh; /* Disk refreshes */ | |
5da844a2 | 193 | u64 l_lock_wait; /* First lock wait time */ |
5bc970e8 | 194 | struct ocfs2_lock_stats l_lock_exmode; /* EX mode stats */ |
8ddb7b00 | 195 | #endif |
cb25797d JK |
196 | #ifdef CONFIG_DEBUG_LOCK_ALLOC |
197 | struct lockdep_map l_lockdep_map; | |
198 | #endif | |
ccd979bd MF |
199 | }; |
200 | ||
ed460cff JQ |
201 | enum ocfs2_orphan_reco_type { |
202 | ORPHAN_NO_NEED_TRUNCATE = 0, | |
203 | ORPHAN_NEED_TRUNCATE, | |
204 | }; | |
205 | ||
692684e1 SM |
206 | enum ocfs2_orphan_scan_state { |
207 | ORPHAN_SCAN_ACTIVE, | |
208 | ORPHAN_SCAN_INACTIVE | |
209 | }; | |
210 | ||
83273932 SE |
211 | struct ocfs2_orphan_scan { |
212 | struct mutex os_lock; | |
213 | struct ocfs2_super *os_osb; | |
214 | struct ocfs2_lock_res os_lockres; /* lock to synchronize scans */ | |
215 | struct delayed_work os_orphan_scan_work; | |
395627b0 | 216 | time64_t os_scantime; /* time this node ran the scan */ |
15633a22 SE |
217 | u32 os_count; /* tracks node specific scans */ |
218 | u32 os_seqno; /* tracks cluster wide scans */ | |
692684e1 | 219 | atomic_t os_state; /* ACTIVE or INACTIVE */ |
83273932 SE |
220 | }; |
221 | ||
ccd979bd MF |
222 | struct ocfs2_dlm_debug { |
223 | struct kref d_refcnt; | |
8056773a | 224 | u32 d_filter_secs; |
ccd979bd MF |
225 | struct list_head d_lockres_tracking; |
226 | }; | |
227 | ||
228 | enum ocfs2_vol_state | |
229 | { | |
230 | VOLUME_INIT = 0, | |
231 | VOLUME_MOUNTED, | |
19ece546 | 232 | VOLUME_MOUNTED_QUOTAS, |
ccd979bd MF |
233 | VOLUME_DISMOUNTED, |
234 | VOLUME_DISABLED | |
235 | }; | |
236 | ||
237 | struct ocfs2_alloc_stats | |
238 | { | |
239 | atomic_t moves; | |
240 | atomic_t local_data; | |
241 | atomic_t bitmap_data; | |
242 | atomic_t bg_allocs; | |
243 | atomic_t bg_extends; | |
244 | }; | |
245 | ||
246 | enum ocfs2_local_alloc_state | |
247 | { | |
9c7af40b MF |
248 | OCFS2_LA_UNUSED = 0, /* Local alloc will never be used for |
249 | * this mountpoint. */ | |
250 | OCFS2_LA_ENABLED, /* Local alloc is in use. */ | |
251 | OCFS2_LA_THROTTLED, /* Local alloc is in use, but number | |
252 | * of bits has been reduced. */ | |
253 | OCFS2_LA_DISABLED /* Local alloc has temporarily been | |
254 | * disabled. */ | |
ccd979bd MF |
255 | }; |
256 | ||
257 | enum ocfs2_mount_options | |
258 | { | |
2c442719 | 259 | OCFS2_MOUNT_HB_LOCAL = 1 << 0, /* Local heartbeat */ |
ccd979bd MF |
260 | OCFS2_MOUNT_BARRIER = 1 << 1, /* Use block barriers */ |
261 | OCFS2_MOUNT_NOINTR = 1 << 2, /* Don't catch signals */ | |
262 | OCFS2_MOUNT_ERRORS_PANIC = 1 << 3, /* Panic on errors */ | |
263 | OCFS2_MOUNT_DATA_WRITEBACK = 1 << 4, /* No data ordering */ | |
53fc622b | 264 | OCFS2_MOUNT_LOCALFLOCKS = 1 << 5, /* No cluster aware user file locks */ |
cf1d6c76 | 265 | OCFS2_MOUNT_NOUSERXATTR = 1 << 6, /* No user xattr */ |
12462f1d | 266 | OCFS2_MOUNT_INODE64 = 1 << 7, /* Allow inode numbers > 2^32 */ |
5297aad8 JK |
267 | OCFS2_MOUNT_POSIX_ACL = 1 << 8, /* Force POSIX access control lists */ |
268 | OCFS2_MOUNT_NO_POSIX_ACL = 1 << 9, /* Disable POSIX access | |
269 | control lists */ | |
270 | OCFS2_MOUNT_USRQUOTA = 1 << 10, /* We support user quotas */ | |
271 | OCFS2_MOUNT_GRPQUOTA = 1 << 11, /* We support group quotas */ | |
fc371891 JB |
272 | OCFS2_MOUNT_COHERENCY_BUFFERED = 1 << 12, /* Allow concurrent O_DIRECT |
273 | writes */ | |
274 | OCFS2_MOUNT_HB_NONE = 1 << 13, /* No heartbeat */ | |
275 | OCFS2_MOUNT_HB_GLOBAL = 1 << 14, /* Global heartbeat */ | |
1dfeb768 | 276 | |
277 | OCFS2_MOUNT_JOURNAL_ASYNC_COMMIT = 1 << 15, /* Journal Async Commit */ | |
7d0fb914 GR |
278 | OCFS2_MOUNT_ERRORS_CONT = 1 << 16, /* Return EIO to the calling process on error */ |
279 | OCFS2_MOUNT_ERRORS_ROFS = 1 << 17, /* Change filesystem to read-only on error */ | |
912f655d | 280 | OCFS2_MOUNT_NOCLUSTER = 1 << 18, /* No cluster aware filesystem mount */ |
ccd979bd MF |
281 | }; |
282 | ||
8ed6b237 GR |
283 | #define OCFS2_OSB_SOFT_RO 0x0001 |
284 | #define OCFS2_OSB_HARD_RO 0x0002 | |
285 | #define OCFS2_OSB_ERROR_FS 0x0004 | |
286 | #define OCFS2_DEFAULT_ATIME_QUANTUM 60 | |
ccd979bd MF |
287 | |
288 | struct ocfs2_journal; | |
d85b20e4 | 289 | struct ocfs2_slot_info; |
553abd04 | 290 | struct ocfs2_recovery_map; |
9140db04 | 291 | struct ocfs2_replay_map; |
2205363d | 292 | struct ocfs2_quota_recovery; |
ccd979bd MF |
293 | struct ocfs2_super |
294 | { | |
ccd979bd MF |
295 | struct task_struct *commit_task; |
296 | struct super_block *sb; | |
297 | struct inode *root_inode; | |
298 | struct inode *sys_root_inode; | |
b4d693fc TM |
299 | struct inode *global_system_inodes[NUM_GLOBAL_SYSTEM_INODES]; |
300 | struct inode **local_system_inodes; | |
ccd979bd MF |
301 | |
302 | struct ocfs2_slot_info *slot_info; | |
303 | ||
539d8264 SM |
304 | u32 *slot_recovery_generations; |
305 | ||
ccd979bd | 306 | spinlock_t node_map_lock; |
ccd979bd | 307 | |
ccd979bd MF |
308 | u64 root_blkno; |
309 | u64 system_dir_blkno; | |
310 | u64 bitmap_blkno; | |
311 | u32 bitmap_cpg; | |
ccd979bd | 312 | char *uuid_str; |
cf1d6c76 | 313 | u32 uuid_hash; |
ccd979bd MF |
314 | u8 *vol_label; |
315 | u64 first_cluster_group_blkno; | |
316 | u32 fs_generation; | |
317 | ||
318 | u32 s_feature_compat; | |
319 | u32 s_feature_incompat; | |
320 | u32 s_feature_ro_compat; | |
321 | ||
4d0ddb2c TM |
322 | /* Protects s_next_generation, osb_flags and s_inode_steal_slot. |
323 | * Could protect more on osb as it's very short lived. | |
324 | */ | |
ccd979bd MF |
325 | spinlock_t osb_lock; |
326 | u32 s_next_generation; | |
327 | unsigned long osb_flags; | |
38d51b2d JB |
328 | u16 s_inode_steal_slot; |
329 | u16 s_meta_steal_slot; | |
4d0ddb2c | 330 | atomic_t s_num_inodes_stolen; |
b89c5428 | 331 | atomic_t s_num_meta_stolen; |
ccd979bd MF |
332 | |
333 | unsigned long s_mount_opt; | |
7f1a37e3 | 334 | unsigned int s_atime_quantum; |
ccd979bd | 335 | |
fc881fa0 | 336 | unsigned int max_slots; |
19fdb624 | 337 | unsigned int node_num; |
fc881fa0 JB |
338 | int slot_num; |
339 | int preferred_slot; | |
ccd979bd MF |
340 | int s_sectsize_bits; |
341 | int s_clustersize; | |
342 | int s_clustersize_bits; | |
fdd77704 | 343 | unsigned int s_xattr_inline_size; |
ccd979bd MF |
344 | |
345 | atomic_t vol_state; | |
c74ec2f7 | 346 | struct mutex recovery_lock; |
553abd04 | 347 | struct ocfs2_recovery_map *recovery_map; |
9140db04 | 348 | struct ocfs2_replay_map *replay_map; |
ccd979bd MF |
349 | struct task_struct *recovery_thread_task; |
350 | int disable_recovery; | |
351 | wait_queue_head_t checkpoint_event; | |
ccd979bd | 352 | struct ocfs2_journal *journal; |
d147b3d6 | 353 | unsigned long osb_commit_interval; |
ccd979bd | 354 | |
9c7af40b MF |
355 | struct delayed_work la_enable_wq; |
356 | ||
357 | /* | |
358 | * Must hold local alloc i_mutex and osb->osb_lock to change | |
359 | * local_alloc_bits. Reads can be done under either lock. | |
360 | */ | |
ebcee4b5 | 361 | unsigned int local_alloc_bits; |
9c7af40b | 362 | unsigned int local_alloc_default_bits; |
6b82021b MF |
363 | /* osb_clusters_at_boot can become stale! Do not trust it to |
364 | * be up to date. */ | |
365 | unsigned int osb_clusters_at_boot; | |
9c7af40b MF |
366 | |
367 | enum ocfs2_local_alloc_state local_alloc_state; /* protected | |
368 | * by osb_lock */ | |
369 | ||
ccd979bd | 370 | struct buffer_head *local_alloc_bh; |
9c7af40b | 371 | |
883d4cae | 372 | u64 la_last_gd; |
ccd979bd | 373 | |
d02f00cc MF |
374 | struct ocfs2_reservation_map osb_la_resmap; |
375 | ||
376 | unsigned int osb_resv_level; | |
83f92318 | 377 | unsigned int osb_dir_resv_level; |
d02f00cc | 378 | |
964f14a0 | 379 | /* Next two fields are for local node slot recovery during |
ccd979bd | 380 | * mount. */ |
ccd979bd | 381 | struct ocfs2_dinode *local_alloc_copy; |
2205363d | 382 | struct ocfs2_quota_recovery *quota_rec; |
ccd979bd | 383 | |
73be192b | 384 | struct ocfs2_blockcheck_stats osb_ecc_stats; |
ccd979bd MF |
385 | struct ocfs2_alloc_stats alloc_stats; |
386 | char dev_str[20]; /* "major,minor" of the device */ | |
387 | ||
98f486f2 SM |
388 | u8 osb_stackflags; |
389 | ||
b61817e1 | 390 | char osb_cluster_stack[OCFS2_STACK_LABEL_LEN + 1]; |
c74a3bdd | 391 | char osb_cluster_name[OCFS2_CLUSTER_NAME_LEN + 1]; |
4670c46d | 392 | struct ocfs2_cluster_connection *cconn; |
ccd979bd MF |
393 | struct ocfs2_lock_res osb_super_lockres; |
394 | struct ocfs2_lock_res osb_rename_lockres; | |
6ca497a8 | 395 | struct ocfs2_lock_res osb_nfs_sync_lockres; |
4cd9973f | 396 | struct rw_semaphore nfs_sync_rwlock; |
4882abeb | 397 | struct ocfs2_lock_res osb_trim_fs_lockres; |
5500ab4e | 398 | struct mutex obs_trim_fs_mutex; |
ccd979bd MF |
399 | struct ocfs2_dlm_debug *osb_dlm_debug; |
400 | ||
401 | struct dentry *osb_debug_root; | |
402 | ||
403 | wait_queue_head_t recovery_event; | |
404 | ||
34d024f8 MF |
405 | spinlock_t dc_task_lock; |
406 | struct task_struct *dc_task; | |
407 | wait_queue_head_t dc_event; | |
408 | unsigned long dc_wake_sequence; | |
409 | unsigned long dc_work_sequence; | |
ccd979bd | 410 | |
7ec373cf MF |
411 | /* |
412 | * Any thread can add locks to the list, but the downconvert | |
413 | * thread is the only one allowed to remove locks. Any change | |
414 | * to this rule requires updating | |
415 | * ocfs2_downconvert_thread_do_work(). | |
416 | */ | |
ccd979bd MF |
417 | struct list_head blocked_lock_list; |
418 | unsigned long blocked_lock_count; | |
419 | ||
e3a767b6 JK |
420 | /* List of dquot structures to drop last reference to */ |
421 | struct llist_head dquot_drop_list; | |
422 | struct work_struct dquot_drop_work; | |
423 | ||
ccd979bd MF |
424 | wait_queue_head_t osb_mount_event; |
425 | ||
426 | /* Truncate log info */ | |
427 | struct inode *osb_tl_inode; | |
428 | struct buffer_head *osb_tl_bh; | |
c4028958 | 429 | struct delayed_work osb_truncate_log_wq; |
a9e9acae | 430 | atomic_t osb_tl_disable; |
50308d81 TM |
431 | /* |
432 | * How many clusters in our truncate log. | |
433 | * It must be protected by osb_tl_inode->i_mutex. | |
434 | */ | |
435 | unsigned int truncated_clusters; | |
b4df6ed8 MF |
436 | |
437 | struct ocfs2_node_map osb_recovering_orphan_dirs; | |
438 | unsigned int *osb_orphan_wipes; | |
439 | wait_queue_head_t osb_wipe_event; | |
c8b9cf9a | 440 | |
83273932 SE |
441 | struct ocfs2_orphan_scan osb_orphan_scan; |
442 | ||
c8b9cf9a TM |
443 | /* used to protect metaecc calculation check of xattr. */ |
444 | spinlock_t osb_xattr_lock; | |
9b7895ef MF |
445 | |
446 | unsigned int osb_dx_mask; | |
447 | u32 osb_dx_seed[4]; | |
feb473a6 TM |
448 | |
449 | /* the group we used to allocate inodes. */ | |
450 | u64 osb_inode_alloc_group; | |
374a263e TM |
451 | |
452 | /* rb tree root for refcount lock. */ | |
453 | struct rb_root osb_rf_lock_tree; | |
454 | struct ocfs2_refcount_tree *osb_ref_tree_lru; | |
43b10a20 | 455 | |
456 | struct mutex system_file_mutex; | |
35ddf78e | 457 | |
458 | /* | |
459 | * OCFS2 needs to schedule several different types of work which | |
460 | * require cluster locking, disk I/O, recovery waits, etc. Since these | |
461 | * types of work tend to be heavy we avoid using the kernel events | |
462 | * workqueue and schedule on our own. | |
463 | */ | |
464 | struct workqueue_struct *ocfs2_wq; | |
5f483c4a GH |
465 | |
466 | /* sysfs directory per partition */ | |
467 | struct kset *osb_dev_kset; | |
468 | ||
469 | /* file check related stuff */ | |
470 | struct ocfs2_filecheck_sysfs_entry osb_fc_ent; | |
ccd979bd MF |
471 | }; |
472 | ||
473 | #define OCFS2_SB(sb) ((struct ocfs2_super *)(sb)->s_fs_info) | |
ccd979bd | 474 | |
13723d00 | 475 | /* Useful typedef for passing around journal access functions */ |
0cf2f763 JB |
476 | typedef int (*ocfs2_journal_access_func)(handle_t *handle, |
477 | struct ocfs2_caching_info *ci, | |
13723d00 JB |
478 | struct buffer_head *bh, int type); |
479 | ||
ccd979bd MF |
480 | static inline int ocfs2_should_order_data(struct inode *inode) |
481 | { | |
482 | if (!S_ISREG(inode->i_mode)) | |
483 | return 0; | |
484 | if (OCFS2_SB(inode->i_sb)->s_mount_opt & OCFS2_MOUNT_DATA_WRITEBACK) | |
485 | return 0; | |
486 | return 1; | |
487 | } | |
488 | ||
dcd0538f MF |
489 | static inline int ocfs2_sparse_alloc(struct ocfs2_super *osb) |
490 | { | |
491 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC) | |
492 | return 1; | |
493 | return 0; | |
494 | } | |
495 | ||
328d5752 MF |
496 | static inline int ocfs2_writes_unwritten_extents(struct ocfs2_super *osb) |
497 | { | |
498 | /* | |
499 | * Support for sparse files is a pre-requisite | |
500 | */ | |
501 | if (!ocfs2_sparse_alloc(osb)) | |
502 | return 0; | |
503 | ||
504 | if (osb->s_feature_ro_compat & OCFS2_FEATURE_RO_COMPAT_UNWRITTEN) | |
505 | return 1; | |
506 | return 0; | |
507 | } | |
508 | ||
160cc266 JQ |
509 | static inline int ocfs2_supports_append_dio(struct ocfs2_super *osb) |
510 | { | |
18d585f0 | 511 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_APPEND_DIO) |
160cc266 JQ |
512 | return 1; |
513 | return 0; | |
514 | } | |
515 | ||
516 | ||
15b1e36b MF |
517 | static inline int ocfs2_supports_inline_data(struct ocfs2_super *osb) |
518 | { | |
519 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_INLINE_DATA) | |
520 | return 1; | |
521 | return 0; | |
522 | } | |
523 | ||
8154da3d TY |
524 | static inline int ocfs2_supports_xattr(struct ocfs2_super *osb) |
525 | { | |
526 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_XATTR) | |
527 | return 1; | |
528 | return 0; | |
529 | } | |
530 | ||
70ad1ba7 JB |
531 | static inline int ocfs2_meta_ecc(struct ocfs2_super *osb) |
532 | { | |
533 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_META_ECC) | |
534 | return 1; | |
535 | return 0; | |
536 | } | |
537 | ||
9b7895ef MF |
538 | static inline int ocfs2_supports_indexed_dirs(struct ocfs2_super *osb) |
539 | { | |
540 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS) | |
541 | return 1; | |
542 | return 0; | |
543 | } | |
544 | ||
4711954e | 545 | static inline int ocfs2_supports_discontig_bg(struct ocfs2_super *osb) |
798db35f JB |
546 | { |
547 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG) | |
548 | return 1; | |
549 | return 0; | |
550 | } | |
551 | ||
198a1ca3 MF |
552 | static inline unsigned int ocfs2_link_max(struct ocfs2_super *osb) |
553 | { | |
554 | if (ocfs2_supports_indexed_dirs(osb)) | |
555 | return OCFS2_DX_LINK_MAX; | |
556 | return OCFS2_LINK_MAX; | |
557 | } | |
558 | ||
559 | static inline unsigned int ocfs2_read_links_count(struct ocfs2_dinode *di) | |
560 | { | |
561 | u32 nlink = le16_to_cpu(di->i_links_count); | |
562 | u32 hi = le16_to_cpu(di->i_links_count_hi); | |
563 | ||
564 | if (di->i_dyn_features & cpu_to_le16(OCFS2_INDEXED_DIR_FL)) | |
565 | nlink |= (hi << OCFS2_LINKS_HI_SHIFT); | |
566 | ||
567 | return nlink; | |
568 | } | |
569 | ||
570 | static inline void ocfs2_set_links_count(struct ocfs2_dinode *di, u32 nlink) | |
571 | { | |
572 | u16 lo, hi; | |
573 | ||
574 | lo = nlink; | |
575 | hi = nlink >> OCFS2_LINKS_HI_SHIFT; | |
576 | ||
577 | di->i_links_count = cpu_to_le16(lo); | |
578 | di->i_links_count_hi = cpu_to_le16(hi); | |
579 | } | |
580 | ||
581 | static inline void ocfs2_add_links_count(struct ocfs2_dinode *di, int n) | |
582 | { | |
583 | u32 links = ocfs2_read_links_count(di); | |
584 | ||
585 | links += n; | |
586 | ||
587 | ocfs2_set_links_count(di, links); | |
588 | } | |
589 | ||
1aa75fea TM |
590 | static inline int ocfs2_refcount_tree(struct ocfs2_super *osb) |
591 | { | |
592 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE) | |
593 | return 1; | |
594 | return 0; | |
595 | } | |
596 | ||
ccd979bd MF |
597 | /* set / clear functions because cluster events can make these happen |
598 | * in parallel so we want the transitions to be atomic. this also | |
599 | * means that any future flags osb_flags must be protected by spinlock | |
600 | * too! */ | |
601 | static inline void ocfs2_set_osb_flag(struct ocfs2_super *osb, | |
602 | unsigned long flag) | |
603 | { | |
604 | spin_lock(&osb->osb_lock); | |
605 | osb->osb_flags |= flag; | |
606 | spin_unlock(&osb->osb_lock); | |
607 | } | |
608 | ||
609 | static inline void ocfs2_set_ro_flag(struct ocfs2_super *osb, | |
610 | int hard) | |
611 | { | |
612 | spin_lock(&osb->osb_lock); | |
613 | osb->osb_flags &= ~(OCFS2_OSB_SOFT_RO|OCFS2_OSB_HARD_RO); | |
614 | if (hard) | |
615 | osb->osb_flags |= OCFS2_OSB_HARD_RO; | |
616 | else | |
617 | osb->osb_flags |= OCFS2_OSB_SOFT_RO; | |
618 | spin_unlock(&osb->osb_lock); | |
619 | } | |
620 | ||
621 | static inline int ocfs2_is_hard_readonly(struct ocfs2_super *osb) | |
622 | { | |
623 | int ret; | |
624 | ||
625 | spin_lock(&osb->osb_lock); | |
626 | ret = osb->osb_flags & OCFS2_OSB_HARD_RO; | |
627 | spin_unlock(&osb->osb_lock); | |
628 | ||
629 | return ret; | |
630 | } | |
631 | ||
632 | static inline int ocfs2_is_soft_readonly(struct ocfs2_super *osb) | |
633 | { | |
634 | int ret; | |
635 | ||
636 | spin_lock(&osb->osb_lock); | |
637 | ret = osb->osb_flags & OCFS2_OSB_SOFT_RO; | |
638 | spin_unlock(&osb->osb_lock); | |
639 | ||
640 | return ret; | |
641 | } | |
642 | ||
98f486f2 | 643 | static inline int ocfs2_clusterinfo_valid(struct ocfs2_super *osb) |
b61817e1 JB |
644 | { |
645 | return (osb->s_feature_incompat & | |
98f486f2 SM |
646 | (OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK | |
647 | OCFS2_FEATURE_INCOMPAT_CLUSTERINFO)); | |
648 | } | |
649 | ||
650 | static inline int ocfs2_userspace_stack(struct ocfs2_super *osb) | |
651 | { | |
652 | if (ocfs2_clusterinfo_valid(osb) && | |
653 | memcmp(osb->osb_cluster_stack, OCFS2_CLASSIC_CLUSTER_STACK, | |
654 | OCFS2_STACK_LABEL_LEN)) | |
655 | return 1; | |
656 | return 0; | |
657 | } | |
658 | ||
659 | static inline int ocfs2_o2cb_stack(struct ocfs2_super *osb) | |
660 | { | |
661 | if (ocfs2_clusterinfo_valid(osb) && | |
662 | !memcmp(osb->osb_cluster_stack, OCFS2_CLASSIC_CLUSTER_STACK, | |
663 | OCFS2_STACK_LABEL_LEN)) | |
664 | return 1; | |
665 | return 0; | |
666 | } | |
667 | ||
668 | static inline int ocfs2_cluster_o2cb_global_heartbeat(struct ocfs2_super *osb) | |
669 | { | |
670 | return ocfs2_o2cb_stack(osb) && | |
671 | (osb->osb_stackflags & OCFS2_CLUSTER_O2CB_GLOBAL_HEARTBEAT); | |
b61817e1 JB |
672 | } |
673 | ||
c271c5c2 SM |
674 | static inline int ocfs2_mount_local(struct ocfs2_super *osb) |
675 | { | |
912f655d GH |
676 | return ((osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT) |
677 | || (osb->s_mount_opt & OCFS2_MOUNT_NOCLUSTER)); | |
c271c5c2 SM |
678 | } |
679 | ||
386a2ef8 JB |
680 | static inline int ocfs2_uses_extended_slot_map(struct ocfs2_super *osb) |
681 | { | |
682 | return (osb->s_feature_incompat & | |
683 | OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP); | |
684 | } | |
685 | ||
686 | ||
ccd979bd MF |
687 | #define OCFS2_IS_VALID_DINODE(ptr) \ |
688 | (!strcmp((ptr)->i_signature, OCFS2_INODE_SIGNATURE)) | |
689 | ||
ccd979bd MF |
690 | #define OCFS2_IS_VALID_EXTENT_BLOCK(ptr) \ |
691 | (!strcmp((ptr)->h_signature, OCFS2_EXTENT_BLOCK_SIGNATURE)) | |
692 | ||
ccd979bd MF |
693 | #define OCFS2_IS_VALID_GROUP_DESC(ptr) \ |
694 | (!strcmp((ptr)->bg_signature, OCFS2_GROUP_DESC_SIGNATURE)) | |
695 | ||
ccd979bd | 696 | |
f6087fb7 JB |
697 | #define OCFS2_IS_VALID_XATTR_BLOCK(ptr) \ |
698 | (!strcmp((ptr)->xb_signature, OCFS2_XATTR_BLOCK_SIGNATURE)) | |
699 | ||
87d35a74 MF |
700 | #define OCFS2_IS_VALID_DIR_TRAILER(ptr) \ |
701 | (!strcmp((ptr)->db_signature, OCFS2_DIR_TRAILER_SIGNATURE)) | |
702 | ||
9b7895ef MF |
703 | #define OCFS2_IS_VALID_DX_ROOT(ptr) \ |
704 | (!strcmp((ptr)->dr_signature, OCFS2_DX_ROOT_SIGNATURE)) | |
705 | ||
706 | #define OCFS2_IS_VALID_DX_LEAF(ptr) \ | |
707 | (!strcmp((ptr)->dl_signature, OCFS2_DX_LEAF_SIGNATURE)) | |
708 | ||
f2c870e3 TM |
709 | #define OCFS2_IS_VALID_REFCOUNT_BLOCK(ptr) \ |
710 | (!strcmp((ptr)->rf_signature, OCFS2_REFCOUNT_BLOCK_SIGNATURE)) | |
711 | ||
ccd979bd MF |
712 | static inline unsigned long ino_from_blkno(struct super_block *sb, |
713 | u64 blkno) | |
714 | { | |
715 | return (unsigned long)(blkno & (u64)ULONG_MAX); | |
716 | } | |
717 | ||
718 | static inline u64 ocfs2_clusters_to_blocks(struct super_block *sb, | |
719 | u32 clusters) | |
720 | { | |
721 | int c_to_b_bits = OCFS2_SB(sb)->s_clustersize_bits - | |
722 | sb->s_blocksize_bits; | |
723 | ||
724 | return (u64)clusters << c_to_b_bits; | |
725 | } | |
726 | ||
ae1f0814 JQ |
727 | static inline u32 ocfs2_clusters_for_blocks(struct super_block *sb, |
728 | u64 blocks) | |
729 | { | |
730 | int b_to_c_bits = OCFS2_SB(sb)->s_clustersize_bits - | |
731 | sb->s_blocksize_bits; | |
732 | ||
733 | blocks += (1 << b_to_c_bits) - 1; | |
734 | return (u32)(blocks >> b_to_c_bits); | |
735 | } | |
736 | ||
ccd979bd MF |
737 | static inline u32 ocfs2_blocks_to_clusters(struct super_block *sb, |
738 | u64 blocks) | |
739 | { | |
740 | int b_to_c_bits = OCFS2_SB(sb)->s_clustersize_bits - | |
741 | sb->s_blocksize_bits; | |
742 | ||
743 | return (u32)(blocks >> b_to_c_bits); | |
744 | } | |
745 | ||
746 | static inline unsigned int ocfs2_clusters_for_bytes(struct super_block *sb, | |
747 | u64 bytes) | |
748 | { | |
749 | int cl_bits = OCFS2_SB(sb)->s_clustersize_bits; | |
750 | unsigned int clusters; | |
751 | ||
752 | bytes += OCFS2_SB(sb)->s_clustersize - 1; | |
753 | /* OCFS2 just cannot have enough clusters to overflow this */ | |
754 | clusters = (unsigned int)(bytes >> cl_bits); | |
755 | ||
756 | return clusters; | |
757 | } | |
758 | ||
24c40b32 JQ |
759 | static inline unsigned int ocfs2_bytes_to_clusters(struct super_block *sb, |
760 | u64 bytes) | |
761 | { | |
762 | int cl_bits = OCFS2_SB(sb)->s_clustersize_bits; | |
763 | unsigned int clusters; | |
764 | ||
765 | clusters = (unsigned int)(bytes >> cl_bits); | |
766 | return clusters; | |
767 | } | |
768 | ||
ccd979bd MF |
769 | static inline u64 ocfs2_blocks_for_bytes(struct super_block *sb, |
770 | u64 bytes) | |
771 | { | |
772 | bytes += sb->s_blocksize - 1; | |
773 | return bytes >> sb->s_blocksize_bits; | |
774 | } | |
775 | ||
776 | static inline u64 ocfs2_clusters_to_bytes(struct super_block *sb, | |
777 | u32 clusters) | |
778 | { | |
779 | return (u64)clusters << OCFS2_SB(sb)->s_clustersize_bits; | |
780 | } | |
781 | ||
1d46dc08 MF |
782 | static inline u64 ocfs2_block_to_cluster_start(struct super_block *sb, |
783 | u64 blocks) | |
784 | { | |
785 | int bits = OCFS2_SB(sb)->s_clustersize_bits - sb->s_blocksize_bits; | |
786 | unsigned int clusters; | |
787 | ||
788 | clusters = ocfs2_blocks_to_clusters(sb, blocks); | |
789 | return (u64)clusters << bits; | |
790 | } | |
791 | ||
ccd979bd MF |
792 | static inline u64 ocfs2_align_bytes_to_clusters(struct super_block *sb, |
793 | u64 bytes) | |
794 | { | |
795 | int cl_bits = OCFS2_SB(sb)->s_clustersize_bits; | |
796 | unsigned int clusters; | |
797 | ||
798 | clusters = ocfs2_clusters_for_bytes(sb, bytes); | |
799 | return (u64)clusters << cl_bits; | |
800 | } | |
801 | ||
802 | static inline u64 ocfs2_align_bytes_to_blocks(struct super_block *sb, | |
803 | u64 bytes) | |
804 | { | |
805 | u64 blocks; | |
806 | ||
807 | blocks = ocfs2_blocks_for_bytes(sb, bytes); | |
808 | return blocks << sb->s_blocksize_bits; | |
809 | } | |
810 | ||
811 | static inline unsigned long ocfs2_align_bytes_to_sectors(u64 bytes) | |
812 | { | |
813 | return (unsigned long)((bytes + 511) >> 9); | |
814 | } | |
815 | ||
9517bac6 MF |
816 | static inline unsigned int ocfs2_page_index_to_clusters(struct super_block *sb, |
817 | unsigned long pg_index) | |
818 | { | |
819 | u32 clusters = pg_index; | |
820 | unsigned int cbits = OCFS2_SB(sb)->s_clustersize_bits; | |
821 | ||
09cbfeaf KS |
822 | if (unlikely(PAGE_SHIFT > cbits)) |
823 | clusters = pg_index << (PAGE_SHIFT - cbits); | |
824 | else if (PAGE_SHIFT < cbits) | |
825 | clusters = pg_index >> (cbits - PAGE_SHIFT); | |
9517bac6 MF |
826 | |
827 | return clusters; | |
828 | } | |
829 | ||
830 | /* | |
831 | * Find the 1st page index which covers the given clusters. | |
832 | */ | |
7c08d70c | 833 | static inline pgoff_t ocfs2_align_clusters_to_page_index(struct super_block *sb, |
9517bac6 MF |
834 | u32 clusters) |
835 | { | |
836 | unsigned int cbits = OCFS2_SB(sb)->s_clustersize_bits; | |
7c08d70c | 837 | pgoff_t index = clusters; |
9517bac6 | 838 | |
09cbfeaf KS |
839 | if (PAGE_SHIFT > cbits) { |
840 | index = (pgoff_t)clusters >> (PAGE_SHIFT - cbits); | |
841 | } else if (PAGE_SHIFT < cbits) { | |
842 | index = (pgoff_t)clusters << (cbits - PAGE_SHIFT); | |
9517bac6 MF |
843 | } |
844 | ||
845 | return index; | |
846 | } | |
847 | ||
60b11392 MF |
848 | static inline unsigned int ocfs2_pages_per_cluster(struct super_block *sb) |
849 | { | |
850 | unsigned int cbits = OCFS2_SB(sb)->s_clustersize_bits; | |
851 | unsigned int pages_per_cluster = 1; | |
852 | ||
09cbfeaf KS |
853 | if (PAGE_SHIFT < cbits) |
854 | pages_per_cluster = 1 << (cbits - PAGE_SHIFT); | |
60b11392 MF |
855 | |
856 | return pages_per_cluster; | |
857 | } | |
858 | ||
ebcee4b5 MF |
859 | static inline unsigned int ocfs2_megabytes_to_clusters(struct super_block *sb, |
860 | unsigned int megs) | |
861 | { | |
862 | BUILD_BUG_ON(OCFS2_MAX_CLUSTERSIZE > 1048576); | |
863 | ||
864 | return megs << (20 - OCFS2_SB(sb)->s_clustersize_bits); | |
865 | } | |
866 | ||
73c8a800 MF |
867 | static inline unsigned int ocfs2_clusters_to_megabytes(struct super_block *sb, |
868 | unsigned int clusters) | |
869 | { | |
870 | return clusters >> (20 - OCFS2_SB(sb)->s_clustersize_bits); | |
871 | } | |
872 | ||
b4414eea MF |
873 | static inline void _ocfs2_set_bit(unsigned int bit, unsigned long *bitmap) |
874 | { | |
730e663b | 875 | __set_bit_le(bit, bitmap); |
b4414eea MF |
876 | } |
877 | #define ocfs2_set_bit(bit, addr) _ocfs2_set_bit((bit), (unsigned long *)(addr)) | |
878 | ||
879 | static inline void _ocfs2_clear_bit(unsigned int bit, unsigned long *bitmap) | |
880 | { | |
730e663b | 881 | __clear_bit_le(bit, bitmap); |
b4414eea MF |
882 | } |
883 | #define ocfs2_clear_bit(bit, addr) _ocfs2_clear_bit((bit), (unsigned long *)(addr)) | |
884 | ||
c4354d0d AM |
885 | #define ocfs2_test_bit test_bit_le |
886 | #define ocfs2_find_next_zero_bit find_next_zero_bit_le | |
887 | #define ocfs2_find_next_bit find_next_bit_le | |
93925579 AM |
888 | |
889 | static inline void *correct_addr_and_bit_unaligned(int *bit, void *addr) | |
890 | { | |
891 | #if BITS_PER_LONG == 64 | |
892 | *bit += ((unsigned long) addr & 7UL) << 3; | |
893 | addr = (void *) ((unsigned long) addr & ~7UL); | |
894 | #elif BITS_PER_LONG == 32 | |
895 | *bit += ((unsigned long) addr & 3UL) << 3; | |
896 | addr = (void *) ((unsigned long) addr & ~3UL); | |
897 | #else | |
898 | #error "how many bits you are?!" | |
899 | #endif | |
900 | return addr; | |
901 | } | |
902 | ||
903 | static inline void ocfs2_set_bit_unaligned(int bit, void *bitmap) | |
904 | { | |
905 | bitmap = correct_addr_and_bit_unaligned(&bit, bitmap); | |
906 | ocfs2_set_bit(bit, bitmap); | |
907 | } | |
908 | ||
909 | static inline void ocfs2_clear_bit_unaligned(int bit, void *bitmap) | |
910 | { | |
911 | bitmap = correct_addr_and_bit_unaligned(&bit, bitmap); | |
912 | ocfs2_clear_bit(bit, bitmap); | |
913 | } | |
914 | ||
915 | static inline int ocfs2_test_bit_unaligned(int bit, void *bitmap) | |
916 | { | |
917 | bitmap = correct_addr_and_bit_unaligned(&bit, bitmap); | |
918 | return ocfs2_test_bit(bit, bitmap); | |
919 | } | |
920 | ||
921 | static inline int ocfs2_find_next_zero_bit_unaligned(void *bitmap, int max, | |
922 | int start) | |
923 | { | |
924 | int fix = 0, ret, tmpmax; | |
925 | bitmap = correct_addr_and_bit_unaligned(&fix, bitmap); | |
926 | tmpmax = max + fix; | |
927 | start += fix; | |
928 | ||
929 | ret = ocfs2_find_next_zero_bit(bitmap, tmpmax, start) - fix; | |
930 | if (ret > max) | |
931 | return max; | |
932 | return ret; | |
933 | } | |
934 | ||
ccd979bd MF |
935 | #endif /* OCFS2_H */ |
936 |