Commit | Line | Data |
---|---|---|
328970de | 1 | /* SPDX-License-Identifier: GPL-2.0-or-later */ |
ccd979bd MF |
2 | /* -*- mode: c; c-basic-offset: 8; -*- |
3 | * vim: noexpandtab sw=8 ts=8 sts=0: | |
4 | * | |
5 | * ocfs2.h | |
6 | * | |
7 | * Defines macros and structures used in OCFS2 | |
8 | * | |
9 | * Copyright (C) 2002, 2004 Oracle. All rights reserved. | |
ccd979bd MF |
10 | */ |
11 | ||
12 | #ifndef OCFS2_H | |
13 | #define OCFS2_H | |
14 | ||
15 | #include <linux/spinlock.h> | |
16 | #include <linux/sched.h> | |
17 | #include <linux/wait.h> | |
18 | #include <linux/list.h> | |
e3a767b6 | 19 | #include <linux/llist.h> |
ccd979bd MF |
20 | #include <linux/rbtree.h> |
21 | #include <linux/workqueue.h> | |
22 | #include <linux/kref.h> | |
c74ec2f7 | 23 | #include <linux/mutex.h> |
cb25797d | 24 | #include <linux/lockdep.h> |
7aee47b0 | 25 | #include <linux/jbd2.h> |
ccd979bd | 26 | |
8f2c9c1b JB |
27 | /* For union ocfs2_dlm_lksb */ |
28 | #include "stackglue.h" | |
ccd979bd MF |
29 | |
30 | #include "ocfs2_fs.h" | |
ccd979bd | 31 | #include "ocfs2_lockid.h" |
9df5778e | 32 | #include "ocfs2_ioctl.h" |
ccd979bd | 33 | |
73be192b JB |
34 | /* For struct ocfs2_blockcheck_stats */ |
35 | #include "blockcheck.h" | |
36 | ||
d02f00cc | 37 | #include "reservations.h" |
47460d65 | 38 | |
5f483c4a GH |
39 | #include "filecheck.h" |
40 | ||
47460d65 JB |
41 | /* Caching of metadata buffers */ |
42 | ||
ccd979bd MF |
43 | /* Most user visible OCFS2 inodes will have very few pieces of |
44 | * metadata, but larger files (including bitmaps, etc) must be taken | |
45 | * into account when designing an access scheme. We allow a small | |
46 | * amount of inlined blocks to be stored on an array and grow the | |
47 | * structure into a rb tree when necessary. */ | |
47460d65 JB |
48 | #define OCFS2_CACHE_INFO_MAX_ARRAY 2 |
49 | ||
50 | /* Flags for ocfs2_caching_info */ | |
51 | ||
52 | enum ocfs2_caching_info_flags { | |
53 | /* Indicates that the metadata cache is using the inline array */ | |
54 | OCFS2_CACHE_FL_INLINE = 1<<1, | |
55 | }; | |
ccd979bd | 56 | |
6e5a3d75 | 57 | struct ocfs2_caching_operations; |
ccd979bd | 58 | struct ocfs2_caching_info { |
47460d65 JB |
59 | /* |
60 | * The parent structure provides the locks, but because the | |
6e5a3d75 JB |
61 | * parent structure can differ, it provides locking operations |
62 | * to struct ocfs2_caching_info. | |
47460d65 | 63 | */ |
6e5a3d75 | 64 | const struct ocfs2_caching_operations *ci_ops; |
47460d65 | 65 | |
292dd27e JB |
66 | /* next two are protected by trans_inc_lock */ |
67 | /* which transaction were we created on? Zero if none. */ | |
68 | unsigned long ci_created_trans; | |
66fb345d JB |
69 | /* last transaction we were a part of. */ |
70 | unsigned long ci_last_trans; | |
71 | ||
72 | /* Cache structures */ | |
47460d65 | 73 | unsigned int ci_flags; |
ccd979bd MF |
74 | unsigned int ci_num_cached; |
75 | union { | |
47460d65 | 76 | sector_t ci_array[OCFS2_CACHE_INFO_MAX_ARRAY]; |
ccd979bd MF |
77 | struct rb_root ci_tree; |
78 | } ci_cache; | |
79 | }; | |
0cf2f763 JB |
80 | /* |
81 | * Need this prototype here instead of in uptodate.h because journal.h | |
82 | * uses it. | |
83 | */ | |
84 | struct super_block *ocfs2_metadata_cache_get_super(struct ocfs2_caching_info *ci); | |
ccd979bd MF |
85 | |
86 | /* this limits us to 256 nodes | |
87 | * if we need more, we can do a kmalloc for the map */ | |
88 | #define OCFS2_NODE_MAP_MAX_NODES 256 | |
89 | struct ocfs2_node_map { | |
90 | u16 num_nodes; | |
91 | unsigned long map[BITS_TO_LONGS(OCFS2_NODE_MAP_MAX_NODES)]; | |
92 | }; | |
93 | ||
94 | enum ocfs2_ast_action { | |
95 | OCFS2_AST_INVALID = 0, | |
96 | OCFS2_AST_ATTACH, | |
97 | OCFS2_AST_CONVERT, | |
98 | OCFS2_AST_DOWNCONVERT, | |
99 | }; | |
100 | ||
101 | /* actions for an unlockast function to take. */ | |
102 | enum ocfs2_unlock_action { | |
103 | OCFS2_UNLOCK_INVALID = 0, | |
104 | OCFS2_UNLOCK_CANCEL_CONVERT, | |
105 | OCFS2_UNLOCK_DROP_LOCK, | |
106 | }; | |
107 | ||
108 | /* ocfs2_lock_res->l_flags flags. */ | |
3b5da018 | 109 | #define OCFS2_LOCK_ATTACHED (0x00000001) /* we have initialized |
ccd979bd MF |
110 | * the lvb */ |
111 | #define OCFS2_LOCK_BUSY (0x00000002) /* we are currently in | |
112 | * dlm_lock */ | |
113 | #define OCFS2_LOCK_BLOCKED (0x00000004) /* blocked waiting to | |
114 | * downconvert*/ | |
115 | #define OCFS2_LOCK_LOCAL (0x00000008) /* newly created inode */ | |
116 | #define OCFS2_LOCK_NEEDS_REFRESH (0x00000010) | |
117 | #define OCFS2_LOCK_REFRESHING (0x00000020) | |
118 | #define OCFS2_LOCK_INITIALIZED (0x00000040) /* track initialization | |
119 | * for shutdown paths */ | |
120 | #define OCFS2_LOCK_FREEING (0x00000080) /* help dlmglue track | |
121 | * when to skip queueing | |
122 | * a lock because it's | |
123 | * about to be | |
124 | * dropped. */ | |
125 | #define OCFS2_LOCK_QUEUED (0x00000100) /* queued for downconvert */ | |
cf8e06f1 | 126 | #define OCFS2_LOCK_NOCACHE (0x00000200) /* don't use a holder count */ |
de551246 JB |
127 | #define OCFS2_LOCK_PENDING (0x00000400) /* This lockres is pending a |
128 | call to dlm_lock. Only | |
129 | exists with BUSY set. */ | |
a1912826 SM |
130 | #define OCFS2_LOCK_UPCONVERT_FINISHING (0x00000800) /* blocks the dc thread |
131 | * from downconverting | |
132 | * before the upconvert | |
133 | * has completed */ | |
ccd979bd | 134 | |
d1e78238 X |
135 | #define OCFS2_LOCK_NONBLOCK_FINISHED (0x00001000) /* NONBLOCK cluster |
136 | * lock has already | |
137 | * returned, do not block | |
138 | * dc thread from | |
139 | * downconverting */ | |
140 | ||
ccd979bd MF |
141 | struct ocfs2_lock_res_ops; |
142 | ||
143 | typedef void (*ocfs2_lock_callback)(int status, unsigned long data); | |
144 | ||
5bc970e8 SM |
145 | #ifdef CONFIG_OCFS2_FS_STATS |
146 | struct ocfs2_lock_stats { | |
147 | u64 ls_total; /* Total wait in NSEC */ | |
148 | u32 ls_gets; /* Num acquires */ | |
149 | u32 ls_fail; /* Num failed acquires */ | |
150 | ||
151 | /* Storing max wait in usecs saves 24 bytes per inode */ | |
152 | u32 ls_max; /* Max wait in USEC */ | |
153 | }; | |
154 | #endif | |
155 | ||
ccd979bd MF |
156 | struct ocfs2_lock_res { |
157 | void *l_priv; | |
158 | struct ocfs2_lock_res_ops *l_ops; | |
83fd9c7f | 159 | |
ccd979bd MF |
160 | |
161 | struct list_head l_blocked_list; | |
162 | struct list_head l_mask_waiters; | |
439a36b8 | 163 | struct list_head l_holders; |
ccd979bd | 164 | |
ccd979bd MF |
165 | unsigned long l_flags; |
166 | char l_name[OCFS2_LOCK_ID_MAX_LEN]; | |
ccd979bd MF |
167 | unsigned int l_ro_holders; |
168 | unsigned int l_ex_holders; | |
a2a2f552 MM |
169 | signed char l_level; |
170 | signed char l_requested; | |
171 | signed char l_blocking; | |
83fd9c7f GR |
172 | |
173 | /* Data packed - type enum ocfs2_lock_type */ | |
174 | unsigned char l_type; | |
ccd979bd MF |
175 | |
176 | /* used from AST/BAST funcs. */ | |
83fd9c7f GR |
177 | /* Data packed - enum type ocfs2_ast_action */ |
178 | unsigned char l_action; | |
179 | /* Data packed - enum type ocfs2_unlock_action */ | |
180 | unsigned char l_unlock_action; | |
de551246 | 181 | unsigned int l_pending_gen; |
ccd979bd | 182 | |
83fd9c7f GR |
183 | spinlock_t l_lock; |
184 | ||
185 | struct ocfs2_dlm_lksb l_lksb; | |
186 | ||
ccd979bd MF |
187 | wait_queue_head_t l_event; |
188 | ||
189 | struct list_head l_debug_list; | |
8ddb7b00 SM |
190 | |
191 | #ifdef CONFIG_OCFS2_FS_STATS | |
5bc970e8 SM |
192 | struct ocfs2_lock_stats l_lock_prmode; /* PR mode stats */ |
193 | u32 l_lock_refresh; /* Disk refreshes */ | |
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; | |
224 | struct dentry *d_locking_state; | |
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 */ | |
ccd979bd MF |
280 | }; |
281 | ||
8ed6b237 GR |
282 | #define OCFS2_OSB_SOFT_RO 0x0001 |
283 | #define OCFS2_OSB_HARD_RO 0x0002 | |
284 | #define OCFS2_OSB_ERROR_FS 0x0004 | |
285 | #define OCFS2_DEFAULT_ATIME_QUANTUM 60 | |
ccd979bd MF |
286 | |
287 | struct ocfs2_journal; | |
d85b20e4 | 288 | struct ocfs2_slot_info; |
553abd04 | 289 | struct ocfs2_recovery_map; |
9140db04 | 290 | struct ocfs2_replay_map; |
2205363d | 291 | struct ocfs2_quota_recovery; |
ccd979bd MF |
292 | struct ocfs2_super |
293 | { | |
ccd979bd MF |
294 | struct task_struct *commit_task; |
295 | struct super_block *sb; | |
296 | struct inode *root_inode; | |
297 | struct inode *sys_root_inode; | |
b4d693fc TM |
298 | struct inode *global_system_inodes[NUM_GLOBAL_SYSTEM_INODES]; |
299 | struct inode **local_system_inodes; | |
ccd979bd MF |
300 | |
301 | struct ocfs2_slot_info *slot_info; | |
302 | ||
539d8264 SM |
303 | u32 *slot_recovery_generations; |
304 | ||
ccd979bd | 305 | spinlock_t node_map_lock; |
ccd979bd | 306 | |
ccd979bd MF |
307 | u64 root_blkno; |
308 | u64 system_dir_blkno; | |
309 | u64 bitmap_blkno; | |
310 | u32 bitmap_cpg; | |
ccd979bd | 311 | char *uuid_str; |
cf1d6c76 | 312 | u32 uuid_hash; |
ccd979bd MF |
313 | u8 *vol_label; |
314 | u64 first_cluster_group_blkno; | |
315 | u32 fs_generation; | |
316 | ||
317 | u32 s_feature_compat; | |
318 | u32 s_feature_incompat; | |
319 | u32 s_feature_ro_compat; | |
320 | ||
4d0ddb2c TM |
321 | /* Protects s_next_generation, osb_flags and s_inode_steal_slot. |
322 | * Could protect more on osb as it's very short lived. | |
323 | */ | |
ccd979bd MF |
324 | spinlock_t osb_lock; |
325 | u32 s_next_generation; | |
326 | unsigned long osb_flags; | |
4d0ddb2c | 327 | s16 s_inode_steal_slot; |
b89c5428 | 328 | s16 s_meta_steal_slot; |
4d0ddb2c | 329 | atomic_t s_num_inodes_stolen; |
b89c5428 | 330 | atomic_t s_num_meta_stolen; |
ccd979bd MF |
331 | |
332 | unsigned long s_mount_opt; | |
7f1a37e3 | 333 | unsigned int s_atime_quantum; |
ccd979bd | 334 | |
fc881fa0 | 335 | unsigned int max_slots; |
19fdb624 | 336 | unsigned int node_num; |
fc881fa0 JB |
337 | int slot_num; |
338 | int preferred_slot; | |
ccd979bd MF |
339 | int s_sectsize_bits; |
340 | int s_clustersize; | |
341 | int s_clustersize_bits; | |
fdd77704 | 342 | unsigned int s_xattr_inline_size; |
ccd979bd MF |
343 | |
344 | atomic_t vol_state; | |
c74ec2f7 | 345 | struct mutex recovery_lock; |
553abd04 | 346 | struct ocfs2_recovery_map *recovery_map; |
9140db04 | 347 | struct ocfs2_replay_map *replay_map; |
ccd979bd MF |
348 | struct task_struct *recovery_thread_task; |
349 | int disable_recovery; | |
350 | wait_queue_head_t checkpoint_event; | |
ccd979bd | 351 | struct ocfs2_journal *journal; |
d147b3d6 | 352 | unsigned long osb_commit_interval; |
ccd979bd | 353 | |
9c7af40b MF |
354 | struct delayed_work la_enable_wq; |
355 | ||
356 | /* | |
357 | * Must hold local alloc i_mutex and osb->osb_lock to change | |
358 | * local_alloc_bits. Reads can be done under either lock. | |
359 | */ | |
ebcee4b5 | 360 | unsigned int local_alloc_bits; |
9c7af40b | 361 | unsigned int local_alloc_default_bits; |
6b82021b MF |
362 | /* osb_clusters_at_boot can become stale! Do not trust it to |
363 | * be up to date. */ | |
364 | unsigned int osb_clusters_at_boot; | |
9c7af40b MF |
365 | |
366 | enum ocfs2_local_alloc_state local_alloc_state; /* protected | |
367 | * by osb_lock */ | |
368 | ||
ccd979bd | 369 | struct buffer_head *local_alloc_bh; |
9c7af40b | 370 | |
883d4cae | 371 | u64 la_last_gd; |
ccd979bd | 372 | |
d02f00cc MF |
373 | struct ocfs2_reservation_map osb_la_resmap; |
374 | ||
375 | unsigned int osb_resv_level; | |
83f92318 | 376 | unsigned int osb_dir_resv_level; |
d02f00cc | 377 | |
964f14a0 | 378 | /* Next two fields are for local node slot recovery during |
ccd979bd | 379 | * mount. */ |
ccd979bd | 380 | struct ocfs2_dinode *local_alloc_copy; |
2205363d | 381 | struct ocfs2_quota_recovery *quota_rec; |
ccd979bd | 382 | |
73be192b | 383 | struct ocfs2_blockcheck_stats osb_ecc_stats; |
ccd979bd MF |
384 | struct ocfs2_alloc_stats alloc_stats; |
385 | char dev_str[20]; /* "major,minor" of the device */ | |
386 | ||
98f486f2 SM |
387 | u8 osb_stackflags; |
388 | ||
b61817e1 | 389 | char osb_cluster_stack[OCFS2_STACK_LABEL_LEN + 1]; |
c74a3bdd | 390 | char osb_cluster_name[OCFS2_CLUSTER_NAME_LEN + 1]; |
4670c46d | 391 | struct ocfs2_cluster_connection *cconn; |
ccd979bd MF |
392 | struct ocfs2_lock_res osb_super_lockres; |
393 | struct ocfs2_lock_res osb_rename_lockres; | |
6ca497a8 | 394 | struct ocfs2_lock_res osb_nfs_sync_lockres; |
4882abeb | 395 | struct ocfs2_lock_res osb_trim_fs_lockres; |
5500ab4e | 396 | struct mutex obs_trim_fs_mutex; |
ccd979bd MF |
397 | struct ocfs2_dlm_debug *osb_dlm_debug; |
398 | ||
399 | struct dentry *osb_debug_root; | |
50397507 | 400 | struct dentry *osb_ctxt; |
ccd979bd MF |
401 | |
402 | wait_queue_head_t recovery_event; | |
403 | ||
34d024f8 MF |
404 | spinlock_t dc_task_lock; |
405 | struct task_struct *dc_task; | |
406 | wait_queue_head_t dc_event; | |
407 | unsigned long dc_wake_sequence; | |
408 | unsigned long dc_work_sequence; | |
ccd979bd | 409 | |
7ec373cf MF |
410 | /* |
411 | * Any thread can add locks to the list, but the downconvert | |
412 | * thread is the only one allowed to remove locks. Any change | |
413 | * to this rule requires updating | |
414 | * ocfs2_downconvert_thread_do_work(). | |
415 | */ | |
ccd979bd MF |
416 | struct list_head blocked_lock_list; |
417 | unsigned long blocked_lock_count; | |
418 | ||
e3a767b6 JK |
419 | /* List of dquot structures to drop last reference to */ |
420 | struct llist_head dquot_drop_list; | |
421 | struct work_struct dquot_drop_work; | |
422 | ||
ccd979bd MF |
423 | wait_queue_head_t osb_mount_event; |
424 | ||
425 | /* Truncate log info */ | |
426 | struct inode *osb_tl_inode; | |
427 | struct buffer_head *osb_tl_bh; | |
c4028958 | 428 | struct delayed_work osb_truncate_log_wq; |
a9e9acae | 429 | atomic_t osb_tl_disable; |
50308d81 TM |
430 | /* |
431 | * How many clusters in our truncate log. | |
432 | * It must be protected by osb_tl_inode->i_mutex. | |
433 | */ | |
434 | unsigned int truncated_clusters; | |
b4df6ed8 MF |
435 | |
436 | struct ocfs2_node_map osb_recovering_orphan_dirs; | |
437 | unsigned int *osb_orphan_wipes; | |
438 | wait_queue_head_t osb_wipe_event; | |
c8b9cf9a | 439 | |
83273932 SE |
440 | struct ocfs2_orphan_scan osb_orphan_scan; |
441 | ||
c8b9cf9a TM |
442 | /* used to protect metaecc calculation check of xattr. */ |
443 | spinlock_t osb_xattr_lock; | |
9b7895ef MF |
444 | |
445 | unsigned int osb_dx_mask; | |
446 | u32 osb_dx_seed[4]; | |
feb473a6 TM |
447 | |
448 | /* the group we used to allocate inodes. */ | |
449 | u64 osb_inode_alloc_group; | |
374a263e TM |
450 | |
451 | /* rb tree root for refcount lock. */ | |
452 | struct rb_root osb_rf_lock_tree; | |
453 | struct ocfs2_refcount_tree *osb_ref_tree_lru; | |
43b10a20 | 454 | |
455 | struct mutex system_file_mutex; | |
35ddf78e | 456 | |
457 | /* | |
458 | * OCFS2 needs to schedule several different types of work which | |
459 | * require cluster locking, disk I/O, recovery waits, etc. Since these | |
460 | * types of work tend to be heavy we avoid using the kernel events | |
461 | * workqueue and schedule on our own. | |
462 | */ | |
463 | struct workqueue_struct *ocfs2_wq; | |
5f483c4a GH |
464 | |
465 | /* sysfs directory per partition */ | |
466 | struct kset *osb_dev_kset; | |
467 | ||
468 | /* file check related stuff */ | |
469 | struct ocfs2_filecheck_sysfs_entry osb_fc_ent; | |
ccd979bd MF |
470 | }; |
471 | ||
472 | #define OCFS2_SB(sb) ((struct ocfs2_super *)(sb)->s_fs_info) | |
ccd979bd | 473 | |
13723d00 | 474 | /* Useful typedef for passing around journal access functions */ |
0cf2f763 JB |
475 | typedef int (*ocfs2_journal_access_func)(handle_t *handle, |
476 | struct ocfs2_caching_info *ci, | |
13723d00 JB |
477 | struct buffer_head *bh, int type); |
478 | ||
ccd979bd MF |
479 | static inline int ocfs2_should_order_data(struct inode *inode) |
480 | { | |
481 | if (!S_ISREG(inode->i_mode)) | |
482 | return 0; | |
483 | if (OCFS2_SB(inode->i_sb)->s_mount_opt & OCFS2_MOUNT_DATA_WRITEBACK) | |
484 | return 0; | |
485 | return 1; | |
486 | } | |
487 | ||
dcd0538f MF |
488 | static inline int ocfs2_sparse_alloc(struct ocfs2_super *osb) |
489 | { | |
490 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC) | |
491 | return 1; | |
492 | return 0; | |
493 | } | |
494 | ||
328d5752 MF |
495 | static inline int ocfs2_writes_unwritten_extents(struct ocfs2_super *osb) |
496 | { | |
497 | /* | |
498 | * Support for sparse files is a pre-requisite | |
499 | */ | |
500 | if (!ocfs2_sparse_alloc(osb)) | |
501 | return 0; | |
502 | ||
503 | if (osb->s_feature_ro_compat & OCFS2_FEATURE_RO_COMPAT_UNWRITTEN) | |
504 | return 1; | |
505 | return 0; | |
506 | } | |
507 | ||
160cc266 JQ |
508 | static inline int ocfs2_supports_append_dio(struct ocfs2_super *osb) |
509 | { | |
18d585f0 | 510 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_APPEND_DIO) |
160cc266 JQ |
511 | return 1; |
512 | return 0; | |
513 | } | |
514 | ||
515 | ||
15b1e36b MF |
516 | static inline int ocfs2_supports_inline_data(struct ocfs2_super *osb) |
517 | { | |
518 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_INLINE_DATA) | |
519 | return 1; | |
520 | return 0; | |
521 | } | |
522 | ||
8154da3d TY |
523 | static inline int ocfs2_supports_xattr(struct ocfs2_super *osb) |
524 | { | |
525 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_XATTR) | |
526 | return 1; | |
527 | return 0; | |
528 | } | |
529 | ||
70ad1ba7 JB |
530 | static inline int ocfs2_meta_ecc(struct ocfs2_super *osb) |
531 | { | |
532 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_META_ECC) | |
533 | return 1; | |
534 | return 0; | |
535 | } | |
536 | ||
9b7895ef MF |
537 | static inline int ocfs2_supports_indexed_dirs(struct ocfs2_super *osb) |
538 | { | |
539 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS) | |
540 | return 1; | |
541 | return 0; | |
542 | } | |
543 | ||
4711954e | 544 | static inline int ocfs2_supports_discontig_bg(struct ocfs2_super *osb) |
798db35f JB |
545 | { |
546 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG) | |
547 | return 1; | |
548 | return 0; | |
549 | } | |
550 | ||
198a1ca3 MF |
551 | static inline unsigned int ocfs2_link_max(struct ocfs2_super *osb) |
552 | { | |
553 | if (ocfs2_supports_indexed_dirs(osb)) | |
554 | return OCFS2_DX_LINK_MAX; | |
555 | return OCFS2_LINK_MAX; | |
556 | } | |
557 | ||
558 | static inline unsigned int ocfs2_read_links_count(struct ocfs2_dinode *di) | |
559 | { | |
560 | u32 nlink = le16_to_cpu(di->i_links_count); | |
561 | u32 hi = le16_to_cpu(di->i_links_count_hi); | |
562 | ||
563 | if (di->i_dyn_features & cpu_to_le16(OCFS2_INDEXED_DIR_FL)) | |
564 | nlink |= (hi << OCFS2_LINKS_HI_SHIFT); | |
565 | ||
566 | return nlink; | |
567 | } | |
568 | ||
569 | static inline void ocfs2_set_links_count(struct ocfs2_dinode *di, u32 nlink) | |
570 | { | |
571 | u16 lo, hi; | |
572 | ||
573 | lo = nlink; | |
574 | hi = nlink >> OCFS2_LINKS_HI_SHIFT; | |
575 | ||
576 | di->i_links_count = cpu_to_le16(lo); | |
577 | di->i_links_count_hi = cpu_to_le16(hi); | |
578 | } | |
579 | ||
580 | static inline void ocfs2_add_links_count(struct ocfs2_dinode *di, int n) | |
581 | { | |
582 | u32 links = ocfs2_read_links_count(di); | |
583 | ||
584 | links += n; | |
585 | ||
586 | ocfs2_set_links_count(di, links); | |
587 | } | |
588 | ||
1aa75fea TM |
589 | static inline int ocfs2_refcount_tree(struct ocfs2_super *osb) |
590 | { | |
591 | if (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE) | |
592 | return 1; | |
593 | return 0; | |
594 | } | |
595 | ||
ccd979bd MF |
596 | /* set / clear functions because cluster events can make these happen |
597 | * in parallel so we want the transitions to be atomic. this also | |
598 | * means that any future flags osb_flags must be protected by spinlock | |
599 | * too! */ | |
600 | static inline void ocfs2_set_osb_flag(struct ocfs2_super *osb, | |
601 | unsigned long flag) | |
602 | { | |
603 | spin_lock(&osb->osb_lock); | |
604 | osb->osb_flags |= flag; | |
605 | spin_unlock(&osb->osb_lock); | |
606 | } | |
607 | ||
608 | static inline void ocfs2_set_ro_flag(struct ocfs2_super *osb, | |
609 | int hard) | |
610 | { | |
611 | spin_lock(&osb->osb_lock); | |
612 | osb->osb_flags &= ~(OCFS2_OSB_SOFT_RO|OCFS2_OSB_HARD_RO); | |
613 | if (hard) | |
614 | osb->osb_flags |= OCFS2_OSB_HARD_RO; | |
615 | else | |
616 | osb->osb_flags |= OCFS2_OSB_SOFT_RO; | |
617 | spin_unlock(&osb->osb_lock); | |
618 | } | |
619 | ||
620 | static inline int ocfs2_is_hard_readonly(struct ocfs2_super *osb) | |
621 | { | |
622 | int ret; | |
623 | ||
624 | spin_lock(&osb->osb_lock); | |
625 | ret = osb->osb_flags & OCFS2_OSB_HARD_RO; | |
626 | spin_unlock(&osb->osb_lock); | |
627 | ||
628 | return ret; | |
629 | } | |
630 | ||
631 | static inline int ocfs2_is_soft_readonly(struct ocfs2_super *osb) | |
632 | { | |
633 | int ret; | |
634 | ||
635 | spin_lock(&osb->osb_lock); | |
636 | ret = osb->osb_flags & OCFS2_OSB_SOFT_RO; | |
637 | spin_unlock(&osb->osb_lock); | |
638 | ||
639 | return ret; | |
640 | } | |
641 | ||
98f486f2 | 642 | static inline int ocfs2_clusterinfo_valid(struct ocfs2_super *osb) |
b61817e1 JB |
643 | { |
644 | return (osb->s_feature_incompat & | |
98f486f2 SM |
645 | (OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK | |
646 | OCFS2_FEATURE_INCOMPAT_CLUSTERINFO)); | |
647 | } | |
648 | ||
649 | static inline int ocfs2_userspace_stack(struct ocfs2_super *osb) | |
650 | { | |
651 | if (ocfs2_clusterinfo_valid(osb) && | |
652 | memcmp(osb->osb_cluster_stack, OCFS2_CLASSIC_CLUSTER_STACK, | |
653 | OCFS2_STACK_LABEL_LEN)) | |
654 | return 1; | |
655 | return 0; | |
656 | } | |
657 | ||
658 | static inline int ocfs2_o2cb_stack(struct ocfs2_super *osb) | |
659 | { | |
660 | if (ocfs2_clusterinfo_valid(osb) && | |
661 | !memcmp(osb->osb_cluster_stack, OCFS2_CLASSIC_CLUSTER_STACK, | |
662 | OCFS2_STACK_LABEL_LEN)) | |
663 | return 1; | |
664 | return 0; | |
665 | } | |
666 | ||
667 | static inline int ocfs2_cluster_o2cb_global_heartbeat(struct ocfs2_super *osb) | |
668 | { | |
669 | return ocfs2_o2cb_stack(osb) && | |
670 | (osb->osb_stackflags & OCFS2_CLUSTER_O2CB_GLOBAL_HEARTBEAT); | |
b61817e1 JB |
671 | } |
672 | ||
c271c5c2 SM |
673 | static inline int ocfs2_mount_local(struct ocfs2_super *osb) |
674 | { | |
675 | return (osb->s_feature_incompat & OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT); | |
676 | } | |
677 | ||
386a2ef8 JB |
678 | static inline int ocfs2_uses_extended_slot_map(struct ocfs2_super *osb) |
679 | { | |
680 | return (osb->s_feature_incompat & | |
681 | OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP); | |
682 | } | |
683 | ||
684 | ||
ccd979bd MF |
685 | #define OCFS2_IS_VALID_DINODE(ptr) \ |
686 | (!strcmp((ptr)->i_signature, OCFS2_INODE_SIGNATURE)) | |
687 | ||
ccd979bd MF |
688 | #define OCFS2_IS_VALID_EXTENT_BLOCK(ptr) \ |
689 | (!strcmp((ptr)->h_signature, OCFS2_EXTENT_BLOCK_SIGNATURE)) | |
690 | ||
ccd979bd MF |
691 | #define OCFS2_IS_VALID_GROUP_DESC(ptr) \ |
692 | (!strcmp((ptr)->bg_signature, OCFS2_GROUP_DESC_SIGNATURE)) | |
693 | ||
ccd979bd | 694 | |
f6087fb7 JB |
695 | #define OCFS2_IS_VALID_XATTR_BLOCK(ptr) \ |
696 | (!strcmp((ptr)->xb_signature, OCFS2_XATTR_BLOCK_SIGNATURE)) | |
697 | ||
87d35a74 MF |
698 | #define OCFS2_IS_VALID_DIR_TRAILER(ptr) \ |
699 | (!strcmp((ptr)->db_signature, OCFS2_DIR_TRAILER_SIGNATURE)) | |
700 | ||
9b7895ef MF |
701 | #define OCFS2_IS_VALID_DX_ROOT(ptr) \ |
702 | (!strcmp((ptr)->dr_signature, OCFS2_DX_ROOT_SIGNATURE)) | |
703 | ||
704 | #define OCFS2_IS_VALID_DX_LEAF(ptr) \ | |
705 | (!strcmp((ptr)->dl_signature, OCFS2_DX_LEAF_SIGNATURE)) | |
706 | ||
f2c870e3 TM |
707 | #define OCFS2_IS_VALID_REFCOUNT_BLOCK(ptr) \ |
708 | (!strcmp((ptr)->rf_signature, OCFS2_REFCOUNT_BLOCK_SIGNATURE)) | |
709 | ||
ccd979bd MF |
710 | static inline unsigned long ino_from_blkno(struct super_block *sb, |
711 | u64 blkno) | |
712 | { | |
713 | return (unsigned long)(blkno & (u64)ULONG_MAX); | |
714 | } | |
715 | ||
716 | static inline u64 ocfs2_clusters_to_blocks(struct super_block *sb, | |
717 | u32 clusters) | |
718 | { | |
719 | int c_to_b_bits = OCFS2_SB(sb)->s_clustersize_bits - | |
720 | sb->s_blocksize_bits; | |
721 | ||
722 | return (u64)clusters << c_to_b_bits; | |
723 | } | |
724 | ||
ae1f0814 JQ |
725 | static inline u32 ocfs2_clusters_for_blocks(struct super_block *sb, |
726 | u64 blocks) | |
727 | { | |
728 | int b_to_c_bits = OCFS2_SB(sb)->s_clustersize_bits - | |
729 | sb->s_blocksize_bits; | |
730 | ||
731 | blocks += (1 << b_to_c_bits) - 1; | |
732 | return (u32)(blocks >> b_to_c_bits); | |
733 | } | |
734 | ||
ccd979bd MF |
735 | static inline u32 ocfs2_blocks_to_clusters(struct super_block *sb, |
736 | u64 blocks) | |
737 | { | |
738 | int b_to_c_bits = OCFS2_SB(sb)->s_clustersize_bits - | |
739 | sb->s_blocksize_bits; | |
740 | ||
741 | return (u32)(blocks >> b_to_c_bits); | |
742 | } | |
743 | ||
744 | static inline unsigned int ocfs2_clusters_for_bytes(struct super_block *sb, | |
745 | u64 bytes) | |
746 | { | |
747 | int cl_bits = OCFS2_SB(sb)->s_clustersize_bits; | |
748 | unsigned int clusters; | |
749 | ||
750 | bytes += OCFS2_SB(sb)->s_clustersize - 1; | |
751 | /* OCFS2 just cannot have enough clusters to overflow this */ | |
752 | clusters = (unsigned int)(bytes >> cl_bits); | |
753 | ||
754 | return clusters; | |
755 | } | |
756 | ||
24c40b32 JQ |
757 | static inline unsigned int ocfs2_bytes_to_clusters(struct super_block *sb, |
758 | u64 bytes) | |
759 | { | |
760 | int cl_bits = OCFS2_SB(sb)->s_clustersize_bits; | |
761 | unsigned int clusters; | |
762 | ||
763 | clusters = (unsigned int)(bytes >> cl_bits); | |
764 | return clusters; | |
765 | } | |
766 | ||
ccd979bd MF |
767 | static inline u64 ocfs2_blocks_for_bytes(struct super_block *sb, |
768 | u64 bytes) | |
769 | { | |
770 | bytes += sb->s_blocksize - 1; | |
771 | return bytes >> sb->s_blocksize_bits; | |
772 | } | |
773 | ||
774 | static inline u64 ocfs2_clusters_to_bytes(struct super_block *sb, | |
775 | u32 clusters) | |
776 | { | |
777 | return (u64)clusters << OCFS2_SB(sb)->s_clustersize_bits; | |
778 | } | |
779 | ||
1d46dc08 MF |
780 | static inline u64 ocfs2_block_to_cluster_start(struct super_block *sb, |
781 | u64 blocks) | |
782 | { | |
783 | int bits = OCFS2_SB(sb)->s_clustersize_bits - sb->s_blocksize_bits; | |
784 | unsigned int clusters; | |
785 | ||
786 | clusters = ocfs2_blocks_to_clusters(sb, blocks); | |
787 | return (u64)clusters << bits; | |
788 | } | |
789 | ||
ccd979bd MF |
790 | static inline u64 ocfs2_align_bytes_to_clusters(struct super_block *sb, |
791 | u64 bytes) | |
792 | { | |
793 | int cl_bits = OCFS2_SB(sb)->s_clustersize_bits; | |
794 | unsigned int clusters; | |
795 | ||
796 | clusters = ocfs2_clusters_for_bytes(sb, bytes); | |
797 | return (u64)clusters << cl_bits; | |
798 | } | |
799 | ||
800 | static inline u64 ocfs2_align_bytes_to_blocks(struct super_block *sb, | |
801 | u64 bytes) | |
802 | { | |
803 | u64 blocks; | |
804 | ||
805 | blocks = ocfs2_blocks_for_bytes(sb, bytes); | |
806 | return blocks << sb->s_blocksize_bits; | |
807 | } | |
808 | ||
809 | static inline unsigned long ocfs2_align_bytes_to_sectors(u64 bytes) | |
810 | { | |
811 | return (unsigned long)((bytes + 511) >> 9); | |
812 | } | |
813 | ||
9517bac6 MF |
814 | static inline unsigned int ocfs2_page_index_to_clusters(struct super_block *sb, |
815 | unsigned long pg_index) | |
816 | { | |
817 | u32 clusters = pg_index; | |
818 | unsigned int cbits = OCFS2_SB(sb)->s_clustersize_bits; | |
819 | ||
09cbfeaf KS |
820 | if (unlikely(PAGE_SHIFT > cbits)) |
821 | clusters = pg_index << (PAGE_SHIFT - cbits); | |
822 | else if (PAGE_SHIFT < cbits) | |
823 | clusters = pg_index >> (cbits - PAGE_SHIFT); | |
9517bac6 MF |
824 | |
825 | return clusters; | |
826 | } | |
827 | ||
828 | /* | |
829 | * Find the 1st page index which covers the given clusters. | |
830 | */ | |
7c08d70c | 831 | static inline pgoff_t ocfs2_align_clusters_to_page_index(struct super_block *sb, |
9517bac6 MF |
832 | u32 clusters) |
833 | { | |
834 | unsigned int cbits = OCFS2_SB(sb)->s_clustersize_bits; | |
7c08d70c | 835 | pgoff_t index = clusters; |
9517bac6 | 836 | |
09cbfeaf KS |
837 | if (PAGE_SHIFT > cbits) { |
838 | index = (pgoff_t)clusters >> (PAGE_SHIFT - cbits); | |
839 | } else if (PAGE_SHIFT < cbits) { | |
840 | index = (pgoff_t)clusters << (cbits - PAGE_SHIFT); | |
9517bac6 MF |
841 | } |
842 | ||
843 | return index; | |
844 | } | |
845 | ||
60b11392 MF |
846 | static inline unsigned int ocfs2_pages_per_cluster(struct super_block *sb) |
847 | { | |
848 | unsigned int cbits = OCFS2_SB(sb)->s_clustersize_bits; | |
849 | unsigned int pages_per_cluster = 1; | |
850 | ||
09cbfeaf KS |
851 | if (PAGE_SHIFT < cbits) |
852 | pages_per_cluster = 1 << (cbits - PAGE_SHIFT); | |
60b11392 MF |
853 | |
854 | return pages_per_cluster; | |
855 | } | |
856 | ||
ebcee4b5 MF |
857 | static inline unsigned int ocfs2_megabytes_to_clusters(struct super_block *sb, |
858 | unsigned int megs) | |
859 | { | |
860 | BUILD_BUG_ON(OCFS2_MAX_CLUSTERSIZE > 1048576); | |
861 | ||
862 | return megs << (20 - OCFS2_SB(sb)->s_clustersize_bits); | |
863 | } | |
864 | ||
73c8a800 MF |
865 | static inline unsigned int ocfs2_clusters_to_megabytes(struct super_block *sb, |
866 | unsigned int clusters) | |
867 | { | |
868 | return clusters >> (20 - OCFS2_SB(sb)->s_clustersize_bits); | |
869 | } | |
870 | ||
b4414eea MF |
871 | static inline void _ocfs2_set_bit(unsigned int bit, unsigned long *bitmap) |
872 | { | |
730e663b | 873 | __set_bit_le(bit, bitmap); |
b4414eea MF |
874 | } |
875 | #define ocfs2_set_bit(bit, addr) _ocfs2_set_bit((bit), (unsigned long *)(addr)) | |
876 | ||
877 | static inline void _ocfs2_clear_bit(unsigned int bit, unsigned long *bitmap) | |
878 | { | |
730e663b | 879 | __clear_bit_le(bit, bitmap); |
b4414eea MF |
880 | } |
881 | #define ocfs2_clear_bit(bit, addr) _ocfs2_clear_bit((bit), (unsigned long *)(addr)) | |
882 | ||
c4354d0d AM |
883 | #define ocfs2_test_bit test_bit_le |
884 | #define ocfs2_find_next_zero_bit find_next_zero_bit_le | |
885 | #define ocfs2_find_next_bit find_next_bit_le | |
93925579 AM |
886 | |
887 | static inline void *correct_addr_and_bit_unaligned(int *bit, void *addr) | |
888 | { | |
889 | #if BITS_PER_LONG == 64 | |
890 | *bit += ((unsigned long) addr & 7UL) << 3; | |
891 | addr = (void *) ((unsigned long) addr & ~7UL); | |
892 | #elif BITS_PER_LONG == 32 | |
893 | *bit += ((unsigned long) addr & 3UL) << 3; | |
894 | addr = (void *) ((unsigned long) addr & ~3UL); | |
895 | #else | |
896 | #error "how many bits you are?!" | |
897 | #endif | |
898 | return addr; | |
899 | } | |
900 | ||
901 | static inline void ocfs2_set_bit_unaligned(int bit, void *bitmap) | |
902 | { | |
903 | bitmap = correct_addr_and_bit_unaligned(&bit, bitmap); | |
904 | ocfs2_set_bit(bit, bitmap); | |
905 | } | |
906 | ||
907 | static inline void ocfs2_clear_bit_unaligned(int bit, void *bitmap) | |
908 | { | |
909 | bitmap = correct_addr_and_bit_unaligned(&bit, bitmap); | |
910 | ocfs2_clear_bit(bit, bitmap); | |
911 | } | |
912 | ||
913 | static inline int ocfs2_test_bit_unaligned(int bit, void *bitmap) | |
914 | { | |
915 | bitmap = correct_addr_and_bit_unaligned(&bit, bitmap); | |
916 | return ocfs2_test_bit(bit, bitmap); | |
917 | } | |
918 | ||
919 | static inline int ocfs2_find_next_zero_bit_unaligned(void *bitmap, int max, | |
920 | int start) | |
921 | { | |
922 | int fix = 0, ret, tmpmax; | |
923 | bitmap = correct_addr_and_bit_unaligned(&fix, bitmap); | |
924 | tmpmax = max + fix; | |
925 | start += fix; | |
926 | ||
927 | ret = ocfs2_find_next_zero_bit(bitmap, tmpmax, start) - fix; | |
928 | if (ret > max) | |
929 | return max; | |
930 | return ret; | |
931 | } | |
932 | ||
ccd979bd MF |
933 | #endif /* OCFS2_H */ |
934 |