1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
4 * All Rights Reserved.
5 */
6 #ifndef __XFS_MOUNT_H__
7 #define __XFS_MOUNT_H__
8
9 struct xlog;
10 struct xfs_inode;
11 struct xfs_mru_cache;
12 struct xfs_ail;
13 struct xfs_quotainfo;
14 struct xfs_da_geometry;
15 struct xfs_perag;
16
17 /* dynamic preallocation free space thresholds, 5% down to 1% */
18 enum {
19 XFS_LOWSP_1_PCNT = 0,
20 XFS_LOWSP_2_PCNT,
21 XFS_LOWSP_3_PCNT,
22 XFS_LOWSP_4_PCNT,
23 XFS_LOWSP_5_PCNT,
24 XFS_LOWSP_MAX,
25 };
26
27 /*
28 * Error Configuration
29 *
30 * Error classes define the subsystem the configuration belongs to.
31 * Error numbers define the errors that are configurable.
32 */
33 enum {
34 XFS_ERR_METADATA,
35 XFS_ERR_CLASS_MAX,
36 };
37 enum {
38 XFS_ERR_DEFAULT,
39 XFS_ERR_EIO,
40 XFS_ERR_ENOSPC,
41 XFS_ERR_ENODEV,
42 XFS_ERR_ERRNO_MAX,
43 };
44
45 #define XFS_ERR_RETRY_FOREVER -1
46
47 /*
48 * Although retry_timeout is in jiffies which is normally an unsigned long,
49 * we limit the retry timeout to 86400 seconds, or one day. So even a
50 * signed 32-bit long is sufficient for a HZ value up to 24855. Making it
51 * signed lets us store the special "-1" value, meaning retry forever.
52 */
53 struct xfs_error_cfg {
54 struct xfs_kobj kobj;
55 int max_retries;
56 long retry_timeout; /* in jiffies, -1 = infinite */
57 };
58
59 /*
60 * Per-cpu deferred inode inactivation GC lists.
61 */
62 struct xfs_inodegc {
63 struct xfs_mount *mp;
64 struct llist_head list;
65 struct delayed_work work;
66 int error;
67
68 /* approximate count of inodes in the list */
69 unsigned int items;
70 unsigned int shrinker_hits;
71 unsigned int cpu;
72 };
73
74 /*
75 * Container for each type of groups, used to look up individual groups and
76 * describes the geometry.
77 */
78 struct xfs_groups {
79 struct xarray xa;
80
81 /*
82 * Maximum capacity of the group in FSBs.
83 *
84 * Each group is laid out densely in the daddr space. For the
85 * degenerate case of a pre-rtgroups filesystem, the incore rtgroup
86 * pretends to have a zero-block and zero-blklog rtgroup.
87 */
88 uint32_t blocks;
89
90 /*
91 * Log(2) of the logical size of each group.
92 *
93 * Compared to the blocks field above this is rounded up to the next
94 * power of two, and thus lays out the xfs_fsblock_t/xfs_rtblock_t
95 * space sparsely with a hole from blocks to (1 << blklog) at the end
96 * of each group.
97 */
98 uint8_t blklog;
99
100 /*
101 * Zoned devices can have gaps beyond the usable capacity of a zone and
102 * the end in the LBA/daddr address space. In other words, the hardware
103 * equivalent to the RT groups already takes care of the power of 2
104 * alignment for us. In this case the sparse FSB/RTB address space maps
105 * 1:1 to the device address space.
106 */
107 bool has_daddr_gaps;
108
109 /*
110 * Mask to extract the group-relative block number from a FSB.
111 * For a pre-rtgroups filesystem we pretend to have one very large
112 * rtgroup, so this mask must be 64-bit.
113 */
114 uint64_t blkmask;
115
116 /*
117 * Start of the first group in the device. This is used to support a
118 * RT device following the data device on the same block device for
119 * SMR hard drives.
120 */
121 xfs_fsblock_t start_fsb;
122 };
123
124 struct xfs_freecounter {
125 /* free blocks for general use: */
126 struct percpu_counter count;
127
128 /* total reserved blocks: */
129 uint64_t res_total;
130
131 /* available reserved blocks: */
132 uint64_t res_avail;
133
134 /* reserved blks @ remount,ro: */
135 uint64_t res_saved;
136 };
137
138 /*
139 * The struct xfsmount layout is optimised to separate read-mostly variables
140 * from variables that are frequently modified. We put the read-mostly variables
141 * first, then place all the other variables at the end.
142 *
143 * Typically, read-mostly variables are those that are set at mount time and
144 * never changed again, or only change rarely as a result of things like sysfs
145 * knobs being tweaked.
146 */
147 typedef struct xfs_mount {
148 struct xfs_sb m_sb; /* copy of fs superblock */
149 struct super_block *m_super;
150 struct xfs_ail *m_ail; /* fs active log item list */
151 struct xfs_buf *m_sb_bp; /* buffer for superblock */
152 struct xfs_buf *m_rtsb_bp; /* realtime superblock */
153 char *m_rtname; /* realtime device name */
154 char *m_logname; /* external log device name */
155 struct xfs_da_geometry *m_dir_geo; /* directory block geometry */
156 struct xfs_da_geometry *m_attr_geo; /* attribute block geometry */
157 struct xlog *m_log; /* log specific stuff */
158 struct xfs_inode *m_rootip; /* pointer to root directory */
159 struct xfs_inode *m_metadirip; /* ptr to metadata directory */
160 struct xfs_inode *m_rtdirip; /* ptr to realtime metadir */
161 struct xfs_quotainfo *m_quotainfo; /* disk quota information */
162 struct xfs_buftarg *m_ddev_targp; /* data device */
163 struct xfs_buftarg *m_logdev_targp;/* log device */
164 struct xfs_buftarg *m_rtdev_targp; /* rt device */
165 void __percpu *m_inodegc; /* percpu inodegc structures */
166 struct xfs_mru_cache *m_filestream; /* per-mount filestream data */
167 struct workqueue_struct *m_buf_workqueue;
168 struct workqueue_struct *m_unwritten_workqueue;
169 struct workqueue_struct *m_reclaim_workqueue;
170 struct workqueue_struct *m_sync_workqueue;
171 struct workqueue_struct *m_blockgc_wq;
172 struct workqueue_struct *m_inodegc_wq;
173
174 int m_bsize; /* fs logical block size */
175 uint8_t m_blkbit_log; /* blocklog + NBBY */
176 uint8_t m_blkbb_log; /* blocklog - BBSHIFT */
177 uint8_t m_agno_log; /* log #ag's */
178 uint8_t m_sectbb_log; /* sectlog - BBSHIFT */
179 int8_t m_rtxblklog; /* log2 of rextsize, if possible */
180
181 uint m_blockmask; /* sb_blocksize-1 */
182 uint m_blockwsize; /* sb_blocksize in words */
183 /* number of rt extents per rt bitmap block if rtgroups enabled */
184 unsigned int m_rtx_per_rbmblock;
185 uint m_alloc_mxr[2]; /* max alloc btree records */
186 uint m_alloc_mnr[2]; /* min alloc btree records */
187 uint m_bmap_dmxr[2]; /* max bmap btree records */
188 uint m_bmap_dmnr[2]; /* min bmap btree records */
189 uint m_rmap_mxr[2]; /* max rmap btree records */
190 uint m_rmap_mnr[2]; /* min rmap btree records */
191 uint m_rtrmap_mxr[2]; /* max rtrmap btree records */
192 uint m_rtrmap_mnr[2]; /* min rtrmap btree records */
193 uint m_refc_mxr[2]; /* max refc btree records */
194 uint m_refc_mnr[2]; /* min refc btree records */
195 uint m_rtrefc_mxr[2]; /* max rtrefc btree records */
196 uint m_rtrefc_mnr[2]; /* min rtrefc btree records */
197 uint m_alloc_maxlevels; /* max alloc btree levels */
198 uint m_bm_maxlevels[2]; /* max bmap btree levels */
199 uint m_rmap_maxlevels; /* max rmap btree levels */
200 uint m_rtrmap_maxlevels; /* max rtrmap btree level */
201 uint m_refc_maxlevels; /* max refcount btree level */
202 uint m_rtrefc_maxlevels; /* max rtrefc btree level */
203 unsigned int m_agbtree_maxlevels; /* max level of all AG btrees */
204 unsigned int m_rtbtree_maxlevels; /* max level of all rt btrees */
205 xfs_extlen_t m_ag_prealloc_blocks; /* reserved ag blocks */
206 uint m_alloc_set_aside; /* space we can't use */
207 uint m_ag_max_usable; /* max space per AG */
208 int m_dalign; /* stripe unit */
209 int m_swidth; /* stripe width */
210 xfs_agnumber_t m_maxagi; /* highest inode alloc group */
211 uint m_allocsize_log;/* min write size log bytes */
212 uint m_allocsize_blocks; /* min write size blocks */
213 int m_logbufs; /* number of log buffers */
214 int m_logbsize; /* size of each log buffer */
215 unsigned int m_rsumlevels; /* rt summary levels */
216 xfs_filblks_t m_rsumblocks; /* size of rt summary, FSBs */
217 int m_fixedfsid[2]; /* unchanged for life of FS */
218 uint m_qflags; /* quota status flags */
219 uint64_t m_features; /* active filesystem features */
220 uint64_t m_low_space[XFS_LOWSP_MAX];
221 uint64_t m_low_rtexts[XFS_LOWSP_MAX];
222 uint64_t m_rtxblkmask; /* rt extent block mask */
223 struct xfs_ino_geometry m_ino_geo; /* inode geometry */
224 struct xfs_trans_resv m_resv; /* precomputed res values */
225 /* low free space thresholds */
226 unsigned long m_opstate; /* dynamic state flags */
227 bool m_always_cow;
228 bool m_fail_unmount;
229 bool m_finobt_nores; /* no per-AG finobt resv. */
230 bool m_update_sb; /* sb needs update in mount */
231 unsigned int m_max_open_zones;
232 unsigned int m_zonegc_low_space;
233
234 /*
235 * Bitsets of per-fs metadata that have been checked and/or are sick.
236 * Callers must hold m_sb_lock to access these two fields.
237 */
238 uint8_t m_fs_checked;
239 uint8_t m_fs_sick;
240 /*
241 * Bitsets of rt metadata that have been checked and/or are sick.
242 * Callers must hold m_sb_lock to access this field.
243 */
244 uint8_t m_rt_checked;
245 uint8_t m_rt_sick;
246
247 /*
248 * End of read-mostly variables. Frequently written variables and locks
249 * should be placed below this comment from now on. The first variable
250 * here is marked as cacheline aligned so they it is separated from
251 * the read-mostly variables.
252 */
253
254 spinlock_t ____cacheline_aligned m_sb_lock; /* sb counter lock */
255 struct percpu_counter m_icount; /* allocated inodes counter */
256 struct percpu_counter m_ifree; /* free inodes counter */
257
258 struct xfs_freecounter m_free[XC_FREE_NR];
259
260 /*
261 * Count of data device blocks reserved for delayed allocations,
262 * including indlen blocks. Does not include allocated CoW staging
263 * extents or anything related to the rt device.
264 */
265 struct percpu_counter m_delalloc_blks;
266
267 /*
268 * RT version of the above.
269 */
270 struct percpu_counter m_delalloc_rtextents;
271
272 /*
273 * Global count of allocation btree blocks in use across all AGs. Only
274 * used when perag reservation is enabled. Helps prevent block
275 * reservation from attempting to reserve allocation btree blocks.
276 */
277 atomic64_t m_allocbt_blks;
278
279 struct xfs_groups m_groups[XG_TYPE_MAX];
280 struct delayed_work m_reclaim_work; /* background inode reclaim */
281 struct xfs_zone_info *m_zone_info; /* zone allocator information */
282 struct dentry *m_debugfs; /* debugfs parent */
283 struct xfs_kobj m_kobj;
284 struct xfs_kobj m_error_kobj;
285 struct xfs_kobj m_error_meta_kobj;
286 struct xfs_error_cfg m_error_cfg[XFS_ERR_CLASS_MAX][XFS_ERR_ERRNO_MAX];
287 struct xstats m_stats; /* per-fs stats */
288 #ifdef CONFIG_XFS_ONLINE_SCRUB_STATS
289 struct xchk_stats *m_scrub_stats;
290 #endif
291 struct xfs_kobj m_zoned_kobj;
292 xfs_agnumber_t m_agfrotor; /* last ag where space found */
293 atomic_t m_agirotor; /* last ag dir inode alloced */
294 atomic_t m_rtgrotor; /* last rtgroup rtpicked */
295
296 struct mutex m_metafile_resv_lock;
297 uint64_t m_metafile_resv_target;
298 uint64_t m_metafile_resv_used;
299 uint64_t m_metafile_resv_avail;
300
301 /* Memory shrinker to throttle and reprioritize inodegc */
302 struct shrinker *m_inodegc_shrinker;
303 /*
304 * Workqueue item so that we can coalesce multiple inode flush attempts
305 * into a single flush.
306 */
307 struct work_struct m_flush_inodes_work;
308
309 /*
310 * Generation of the filesysyem layout. This is incremented by each
311 * growfs, and used by the pNFS server to ensure the client updates
312 * its view of the block device once it gets a layout that might
313 * reference the newly added blocks. Does not need to be persistent
314 * as long as we only allow file system size increments, but if we
315 * ever support shrinks it would have to be persisted in addition
316 * to various other kinds of pain inflicted on the pNFS server.
317 */
318 uint32_t m_generation;
319 struct mutex m_growlock; /* growfs mutex */
320
321 #ifdef DEBUG
322 /*
323 * Frequency with which errors are injected. Replaces xfs_etest; the
324 * value stored in here is the inverse of the frequency with which the
325 * error triggers. 1 = always, 2 = half the time, etc.
326 */
327 unsigned int *m_errortag;
328 struct xfs_kobj m_errortag_kobj;
329 #endif
330
331 /* cpus that have inodes queued for inactivation */
332 struct cpumask m_inodegc_cpumask;
333
334 /* Hook to feed dirent updates to an active online repair. */
335 struct xfs_hooks m_dir_update_hooks;
336 } xfs_mount_t;
337
338 #define M_IGEO(mp) (&(mp)->m_ino_geo)
339
340 /*
341 * Flags for m_features.
342 *
343 * These are all the active features in the filesystem, regardless of how
344 * they are configured.
345 */
346 #define XFS_FEAT_ATTR (1ULL << 0) /* xattrs present in fs */
347 #define XFS_FEAT_NLINK (1ULL << 1) /* 32 bit link counts */
348 #define XFS_FEAT_QUOTA (1ULL << 2) /* quota active */
349 #define XFS_FEAT_ALIGN (1ULL << 3) /* inode alignment */
350 #define XFS_FEAT_DALIGN (1ULL << 4) /* data alignment */
351 #define XFS_FEAT_LOGV2 (1ULL << 5) /* version 2 logs */
352 #define XFS_FEAT_SECTOR (1ULL << 6) /* sector size > 512 bytes */
353 #define XFS_FEAT_EXTFLG (1ULL << 7) /* unwritten extents */
354 #define XFS_FEAT_ASCIICI (1ULL << 8) /* ASCII only case-insens. */
355 #define XFS_FEAT_LAZYSBCOUNT (1ULL << 9) /* Superblk counters */
356 #define XFS_FEAT_ATTR2 (1ULL << 10) /* dynamic attr fork */
357 #define XFS_FEAT_PARENT (1ULL << 11) /* parent pointers */
358 #define XFS_FEAT_PROJID32 (1ULL << 12) /* 32 bit project id */
359 #define XFS_FEAT_CRC (1ULL << 13) /* metadata CRCs */
360 #define XFS_FEAT_V3INODES (1ULL << 14) /* Version 3 inodes */
361 #define XFS_FEAT_PQUOTINO (1ULL << 15) /* non-shared proj/grp quotas */
362 #define XFS_FEAT_FTYPE (1ULL << 16) /* inode type in dir */
363 #define XFS_FEAT_FINOBT (1ULL << 17) /* free inode btree */
364 #define XFS_FEAT_RMAPBT (1ULL << 18) /* reverse map btree */
365 #define XFS_FEAT_REFLINK (1ULL << 19) /* reflinked files */
366 #define XFS_FEAT_SPINODES (1ULL << 20) /* sparse inode chunks */
367 #define XFS_FEAT_META_UUID (1ULL << 21) /* metadata UUID */
368 #define XFS_FEAT_REALTIME (1ULL << 22) /* realtime device present */
369 #define XFS_FEAT_INOBTCNT (1ULL << 23) /* inobt block counts */
370 #define XFS_FEAT_BIGTIME (1ULL << 24) /* large timestamps */
371 #define XFS_FEAT_NEEDSREPAIR (1ULL << 25) /* needs xfs_repair */
372 #define XFS_FEAT_NREXT64 (1ULL << 26) /* large extent counters */
373 #define XFS_FEAT_EXCHANGE_RANGE (1ULL << 27) /* exchange range */
374 #define XFS_FEAT_METADIR (1ULL << 28) /* metadata directory tree */
375 #define XFS_FEAT_ZONED (1ULL << 29) /* zoned RT device */
376
377 /* Mount features */
378 #define XFS_FEAT_NOLIFETIME (1ULL << 47) /* disable lifetime hints */
379 #define XFS_FEAT_NOATTR2 (1ULL << 48) /* disable attr2 creation */
380 #define XFS_FEAT_NOALIGN (1ULL << 49) /* ignore alignment */
381 #define XFS_FEAT_ALLOCSIZE (1ULL << 50) /* user specified allocation size */
382 #define XFS_FEAT_LARGE_IOSIZE (1ULL << 51) /* report large preferred
383 * I/O size in stat() */
384 #define XFS_FEAT_WSYNC (1ULL << 52) /* synchronous metadata ops */
385 #define XFS_FEAT_DIRSYNC (1ULL << 53) /* synchronous directory ops */
386 #define XFS_FEAT_DISCARD (1ULL << 54) /* discard unused blocks */
387 #define XFS_FEAT_GRPID (1ULL << 55) /* group-ID assigned from directory */
388 #define XFS_FEAT_SMALL_INUMS (1ULL << 56) /* user wants 32bit inodes */
389 #define XFS_FEAT_IKEEP (1ULL << 57) /* keep empty inode clusters*/
390 #define XFS_FEAT_SWALLOC (1ULL << 58) /* stripe width allocation */
391 #define XFS_FEAT_FILESTREAMS (1ULL << 59) /* use filestreams allocator */
392 #define XFS_FEAT_DAX_ALWAYS (1ULL << 60) /* DAX always enabled */
393 #define XFS_FEAT_DAX_NEVER (1ULL << 61) /* DAX never enabled */
394 #define XFS_FEAT_NORECOVERY (1ULL << 62) /* no recovery - dirty fs */
395 #define XFS_FEAT_NOUUID (1ULL << 63) /* ignore uuid during mount */
396
397 #define __XFS_HAS_FEAT(name, NAME) \
398 static inline bool xfs_has_ ## name (const struct xfs_mount *mp) \
399 { \
400 return mp->m_features & XFS_FEAT_ ## NAME; \
401 }
402
403 /* Some features can be added dynamically so they need a set wrapper, too. */
404 #define __XFS_ADD_FEAT(name, NAME) \
405 __XFS_HAS_FEAT(name, NAME); \
406 static inline void xfs_add_ ## name (struct xfs_mount *mp) \
407 { \
408 mp->m_features |= XFS_FEAT_ ## NAME; \
409 xfs_sb_version_add ## name(&mp->m_sb); \
410 }
411
412 /* Superblock features */
__XFS_ADD_FEAT(attr,ATTR)413 __XFS_ADD_FEAT(attr, ATTR)
414 __XFS_HAS_FEAT(nlink, NLINK)
415 __XFS_ADD_FEAT(quota, QUOTA)
416 __XFS_HAS_FEAT(dalign, DALIGN)
417 __XFS_HAS_FEAT(sector, SECTOR)
418 __XFS_HAS_FEAT(asciici, ASCIICI)
419 __XFS_HAS_FEAT(parent, PARENT)
420 __XFS_HAS_FEAT(ftype, FTYPE)
421 __XFS_HAS_FEAT(finobt, FINOBT)
422 __XFS_HAS_FEAT(rmapbt, RMAPBT)
423 __XFS_HAS_FEAT(reflink, REFLINK)
424 __XFS_HAS_FEAT(sparseinodes, SPINODES)
425 __XFS_HAS_FEAT(metauuid, META_UUID)
426 __XFS_HAS_FEAT(realtime, REALTIME)
427 __XFS_HAS_FEAT(inobtcounts, INOBTCNT)
428 __XFS_HAS_FEAT(bigtime, BIGTIME)
429 __XFS_HAS_FEAT(needsrepair, NEEDSREPAIR)
430 __XFS_HAS_FEAT(large_extent_counts, NREXT64)
431 __XFS_HAS_FEAT(exchange_range, EXCHANGE_RANGE)
432 __XFS_HAS_FEAT(metadir, METADIR)
433 __XFS_HAS_FEAT(zoned, ZONED)
434 __XFS_HAS_FEAT(nolifetime, NOLIFETIME)
435
436 static inline bool xfs_has_rtgroups(const struct xfs_mount *mp)
437 {
438 /* all metadir file systems also allow rtgroups */
439 return xfs_has_metadir(mp);
440 }
441
xfs_has_rtsb(const struct xfs_mount * mp)442 static inline bool xfs_has_rtsb(const struct xfs_mount *mp)
443 {
444 /* all rtgroups filesystems with an rt section have an rtsb */
445 return xfs_has_rtgroups(mp) &&
446 xfs_has_realtime(mp) &&
447 !xfs_has_zoned(mp);
448 }
449
xfs_has_rtrmapbt(const struct xfs_mount * mp)450 static inline bool xfs_has_rtrmapbt(const struct xfs_mount *mp)
451 {
452 return xfs_has_rtgroups(mp) && xfs_has_realtime(mp) &&
453 xfs_has_rmapbt(mp);
454 }
455
xfs_has_rtreflink(const struct xfs_mount * mp)456 static inline bool xfs_has_rtreflink(const struct xfs_mount *mp)
457 {
458 return xfs_has_metadir(mp) && xfs_has_realtime(mp) &&
459 xfs_has_reflink(mp);
460 }
461
xfs_has_nonzoned(const struct xfs_mount * mp)462 static inline bool xfs_has_nonzoned(const struct xfs_mount *mp)
463 {
464 return !xfs_has_zoned(mp);
465 }
466
467 /*
468 * Some features are always on for v5 file systems, allow the compiler to
469 * eliminiate dead code when building without v4 support.
470 */
471 #define __XFS_HAS_V4_FEAT(name, NAME) \
472 static inline bool xfs_has_ ## name (struct xfs_mount *mp) \
473 { \
474 return !IS_ENABLED(CONFIG_XFS_SUPPORT_V4) || \
475 (mp->m_features & XFS_FEAT_ ## NAME); \
476 }
477
478 #define __XFS_ADD_V4_FEAT(name, NAME) \
479 __XFS_HAS_V4_FEAT(name, NAME); \
480 static inline void xfs_add_ ## name (struct xfs_mount *mp) \
481 { \
482 if (IS_ENABLED(CONFIG_XFS_SUPPORT_V4)) { \
483 mp->m_features |= XFS_FEAT_ ## NAME; \
484 xfs_sb_version_add ## name(&mp->m_sb); \
485 } \
486 }
487
__XFS_HAS_V4_FEAT(align,ALIGN)488 __XFS_HAS_V4_FEAT(align, ALIGN)
489 __XFS_HAS_V4_FEAT(logv2, LOGV2)
490 __XFS_HAS_V4_FEAT(extflg, EXTFLG)
491 __XFS_HAS_V4_FEAT(lazysbcount, LAZYSBCOUNT)
492 __XFS_ADD_V4_FEAT(attr2, ATTR2)
493 __XFS_ADD_V4_FEAT(projid32, PROJID32)
494 __XFS_HAS_V4_FEAT(v3inodes, V3INODES)
495 __XFS_HAS_V4_FEAT(crc, CRC)
496 __XFS_HAS_V4_FEAT(pquotino, PQUOTINO)
497
498 /*
499 * Mount features
500 *
501 * These do not change dynamically - features that can come and go, such as 32
502 * bit inodes and read-only state, are kept as operational state rather than
503 * features.
504 */
505 __XFS_HAS_FEAT(noattr2, NOATTR2)
506 __XFS_HAS_FEAT(noalign, NOALIGN)
507 __XFS_HAS_FEAT(allocsize, ALLOCSIZE)
508 __XFS_HAS_FEAT(large_iosize, LARGE_IOSIZE)
509 __XFS_HAS_FEAT(wsync, WSYNC)
510 __XFS_HAS_FEAT(dirsync, DIRSYNC)
511 __XFS_HAS_FEAT(discard, DISCARD)
512 __XFS_HAS_FEAT(grpid, GRPID)
513 __XFS_HAS_FEAT(small_inums, SMALL_INUMS)
514 __XFS_HAS_FEAT(ikeep, IKEEP)
515 __XFS_HAS_FEAT(swalloc, SWALLOC)
516 __XFS_HAS_FEAT(filestreams, FILESTREAMS)
517 __XFS_HAS_FEAT(dax_always, DAX_ALWAYS)
518 __XFS_HAS_FEAT(dax_never, DAX_NEVER)
519 __XFS_HAS_FEAT(norecovery, NORECOVERY)
520 __XFS_HAS_FEAT(nouuid, NOUUID)
521
522 /*
523 * Operational mount state flags
524 *
525 * Use these with atomic bit ops only!
526 */
527 #define XFS_OPSTATE_UNMOUNTING 0 /* filesystem is unmounting */
528 #define XFS_OPSTATE_CLEAN 1 /* mount was clean */
529 #define XFS_OPSTATE_SHUTDOWN 2 /* stop all fs operations */
530 #define XFS_OPSTATE_INODE32 3 /* inode32 allocator active */
531 #define XFS_OPSTATE_READONLY 4 /* read-only fs */
532
533 /*
534 * If set, inactivation worker threads will be scheduled to process queued
535 * inodegc work. If not, queued inodes remain in memory waiting to be
536 * processed.
537 */
538 #define XFS_OPSTATE_INODEGC_ENABLED 5
539 /*
540 * If set, background speculative prealloc gc worker threads will be scheduled
541 * to process queued blockgc work. If not, inodes retain their preallocations
542 * until explicitly deleted.
543 */
544 #define XFS_OPSTATE_BLOCKGC_ENABLED 6
545
546 /* Kernel has logged a warning about pNFS being used on this fs. */
547 #define XFS_OPSTATE_WARNED_PNFS 7
548 /* Kernel has logged a warning about online fsck being used on this fs. */
549 #define XFS_OPSTATE_WARNED_SCRUB 8
550 /* Kernel has logged a warning about shrink being used on this fs. */
551 #define XFS_OPSTATE_WARNED_SHRINK 9
552 /* Kernel has logged a warning about logged xattr updates being used. */
553 #define XFS_OPSTATE_WARNED_LARP 10
554 /* Mount time quotacheck is running */
555 #define XFS_OPSTATE_QUOTACHECK_RUNNING 11
556 /* Do we want to clear log incompat flags? */
557 #define XFS_OPSTATE_UNSET_LOG_INCOMPAT 12
558 /* Filesystem can use logged extended attributes */
559 #define XFS_OPSTATE_USE_LARP 13
560 /* Kernel has logged a warning about blocksize > pagesize on this fs. */
561 #define XFS_OPSTATE_WARNED_LBS 14
562 /* Kernel has logged a warning about exchange-range being used on this fs. */
563 #define XFS_OPSTATE_WARNED_EXCHRANGE 15
564 /* Kernel has logged a warning about parent pointers being used on this fs. */
565 #define XFS_OPSTATE_WARNED_PPTR 16
566 /* Kernel has logged a warning about metadata dirs being used on this fs. */
567 #define XFS_OPSTATE_WARNED_METADIR 17
568 /* Filesystem should use qflags to determine quotaon status */
569 #define XFS_OPSTATE_RESUMING_QUOTAON 18
570 /* Kernel has logged a warning about zoned RT device being used on this fs. */
571 #define XFS_OPSTATE_WARNED_ZONED 19
572 /* (Zoned) GC is in progress */
573 #define XFS_OPSTATE_ZONEGC_RUNNING 20
574
575 #define __XFS_IS_OPSTATE(name, NAME) \
576 static inline bool xfs_is_ ## name (struct xfs_mount *mp) \
577 { \
578 return test_bit(XFS_OPSTATE_ ## NAME, &mp->m_opstate); \
579 } \
580 static inline bool xfs_clear_ ## name (struct xfs_mount *mp) \
581 { \
582 return test_and_clear_bit(XFS_OPSTATE_ ## NAME, &mp->m_opstate); \
583 } \
584 static inline bool xfs_set_ ## name (struct xfs_mount *mp) \
585 { \
586 return test_and_set_bit(XFS_OPSTATE_ ## NAME, &mp->m_opstate); \
587 }
588
589 __XFS_IS_OPSTATE(unmounting, UNMOUNTING)
590 __XFS_IS_OPSTATE(clean, CLEAN)
591 __XFS_IS_OPSTATE(shutdown, SHUTDOWN)
592 __XFS_IS_OPSTATE(inode32, INODE32)
593 __XFS_IS_OPSTATE(readonly, READONLY)
594 __XFS_IS_OPSTATE(inodegc_enabled, INODEGC_ENABLED)
595 __XFS_IS_OPSTATE(blockgc_enabled, BLOCKGC_ENABLED)
596 #ifdef CONFIG_XFS_QUOTA
597 __XFS_IS_OPSTATE(quotacheck_running, QUOTACHECK_RUNNING)
598 __XFS_IS_OPSTATE(resuming_quotaon, RESUMING_QUOTAON)
599 #else
600 static inline bool xfs_is_quotacheck_running(struct xfs_mount *mp)
601 {
602 return false;
603 }
604 static inline bool xfs_is_resuming_quotaon(struct xfs_mount *mp)
605 {
606 return false;
607 }
608 static inline void xfs_set_resuming_quotaon(struct xfs_mount *m)
609 {
610 }
611 static inline bool xfs_clear_resuming_quotaon(struct xfs_mount *mp)
612 {
613 return false;
614 }
615 #endif /* CONFIG_XFS_QUOTA */
616 __XFS_IS_OPSTATE(done_with_log_incompat, UNSET_LOG_INCOMPAT)
617 __XFS_IS_OPSTATE(using_logged_xattrs, USE_LARP)
618 __XFS_IS_OPSTATE(zonegc_running, ZONEGC_RUNNING)
619
620 static inline bool
621 xfs_should_warn(struct xfs_mount *mp, long nr)
622 {
623 return !test_and_set_bit(nr, &mp->m_opstate);
624 }
625
626 #define XFS_OPSTATE_STRINGS \
627 { (1UL << XFS_OPSTATE_UNMOUNTING), "unmounting" }, \
628 { (1UL << XFS_OPSTATE_CLEAN), "clean" }, \
629 { (1UL << XFS_OPSTATE_SHUTDOWN), "shutdown" }, \
630 { (1UL << XFS_OPSTATE_INODE32), "inode32" }, \
631 { (1UL << XFS_OPSTATE_READONLY), "read_only" }, \
632 { (1UL << XFS_OPSTATE_INODEGC_ENABLED), "inodegc" }, \
633 { (1UL << XFS_OPSTATE_BLOCKGC_ENABLED), "blockgc" }, \
634 { (1UL << XFS_OPSTATE_WARNED_SCRUB), "wscrub" }, \
635 { (1UL << XFS_OPSTATE_WARNED_SHRINK), "wshrink" }, \
636 { (1UL << XFS_OPSTATE_WARNED_LARP), "wlarp" }, \
637 { (1UL << XFS_OPSTATE_QUOTACHECK_RUNNING), "quotacheck" }, \
638 { (1UL << XFS_OPSTATE_UNSET_LOG_INCOMPAT), "unset_log_incompat" }, \
639 { (1UL << XFS_OPSTATE_USE_LARP), "logged_xattrs" }
640
641 /*
642 * Max and min values for mount-option defined I/O
643 * preallocation sizes.
644 */
645 #define XFS_MAX_IO_LOG 30 /* 1G */
646 #define XFS_MIN_IO_LOG PAGE_SHIFT
647
648 void xfs_do_force_shutdown(struct xfs_mount *mp, uint32_t flags, char *fname,
649 int lnnum);
650 #define xfs_force_shutdown(m,f) \
651 xfs_do_force_shutdown(m, f, __FILE__, __LINE__)
652
653 #define SHUTDOWN_META_IO_ERROR (1u << 0) /* write attempt to metadata failed */
654 #define SHUTDOWN_LOG_IO_ERROR (1u << 1) /* write attempt to the log failed */
655 #define SHUTDOWN_FORCE_UMOUNT (1u << 2) /* shutdown from a forced unmount */
656 #define SHUTDOWN_CORRUPT_INCORE (1u << 3) /* corrupt in-memory structures */
657 #define SHUTDOWN_CORRUPT_ONDISK (1u << 4) /* corrupt metadata on device */
658 #define SHUTDOWN_DEVICE_REMOVED (1u << 5) /* device removed underneath us */
659
660 #define XFS_SHUTDOWN_STRINGS \
661 { SHUTDOWN_META_IO_ERROR, "metadata_io" }, \
662 { SHUTDOWN_LOG_IO_ERROR, "log_io" }, \
663 { SHUTDOWN_FORCE_UMOUNT, "force_umount" }, \
664 { SHUTDOWN_CORRUPT_INCORE, "corruption" }, \
665 { SHUTDOWN_DEVICE_REMOVED, "device_removed" }
666
667 /*
668 * Flags for xfs_mountfs
669 */
670 #define XFS_MFSI_QUIET 0x40 /* Be silent if mount errors found */
671
672 static inline xfs_agnumber_t
xfs_daddr_to_agno(struct xfs_mount * mp,xfs_daddr_t d)673 xfs_daddr_to_agno(struct xfs_mount *mp, xfs_daddr_t d)
674 {
675 xfs_rfsblock_t ld = XFS_BB_TO_FSBT(mp, d);
676 do_div(ld, mp->m_sb.sb_agblocks);
677 return (xfs_agnumber_t) ld;
678 }
679
680 static inline xfs_agblock_t
xfs_daddr_to_agbno(struct xfs_mount * mp,xfs_daddr_t d)681 xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
682 {
683 xfs_rfsblock_t ld = XFS_BB_TO_FSBT(mp, d);
684 return (xfs_agblock_t) do_div(ld, mp->m_sb.sb_agblocks);
685 }
686
687 extern void xfs_uuid_table_free(void);
688 uint64_t xfs_default_resblks(struct xfs_mount *mp,
689 enum xfs_free_counter ctr);
690 extern int xfs_mountfs(xfs_mount_t *mp);
691 extern void xfs_unmountfs(xfs_mount_t *);
692
693 /*
694 * Deltas for the block count can vary from 1 to very large, but lock contention
695 * only occurs on frequent small block count updates such as in the delayed
696 * allocation path for buffered writes (page a time updates). Hence we set
697 * a large batch count (1024) to minimise global counter updates except when
698 * we get near to ENOSPC and we have to be very accurate with our updates.
699 */
700 #define XFS_FDBLOCKS_BATCH 1024
701
702 uint64_t xfs_freecounter_unavailable(struct xfs_mount *mp,
703 enum xfs_free_counter ctr);
704
705 /*
706 * Sum up the freecount, but never return negative values.
707 */
xfs_sum_freecounter(struct xfs_mount * mp,enum xfs_free_counter ctr)708 static inline s64 xfs_sum_freecounter(struct xfs_mount *mp,
709 enum xfs_free_counter ctr)
710 {
711 return percpu_counter_sum_positive(&mp->m_free[ctr].count);
712 }
713
714 /*
715 * Same as above, but does return negative values. Mostly useful for
716 * special cases like repair and tracing.
717 */
xfs_sum_freecounter_raw(struct xfs_mount * mp,enum xfs_free_counter ctr)718 static inline s64 xfs_sum_freecounter_raw(struct xfs_mount *mp,
719 enum xfs_free_counter ctr)
720 {
721 return percpu_counter_sum(&mp->m_free[ctr].count);
722 }
723
724 /*
725 * This just provides and estimate without the cpu-local updates, use
726 * xfs_sum_freecounter for the exact value.
727 */
xfs_estimate_freecounter(struct xfs_mount * mp,enum xfs_free_counter ctr)728 static inline s64 xfs_estimate_freecounter(struct xfs_mount *mp,
729 enum xfs_free_counter ctr)
730 {
731 return percpu_counter_read_positive(&mp->m_free[ctr].count);
732 }
733
xfs_compare_freecounter(struct xfs_mount * mp,enum xfs_free_counter ctr,s64 rhs,s32 batch)734 static inline int xfs_compare_freecounter(struct xfs_mount *mp,
735 enum xfs_free_counter ctr, s64 rhs, s32 batch)
736 {
737 return __percpu_counter_compare(&mp->m_free[ctr].count, rhs, batch);
738 }
739
xfs_set_freecounter(struct xfs_mount * mp,enum xfs_free_counter ctr,uint64_t val)740 static inline void xfs_set_freecounter(struct xfs_mount *mp,
741 enum xfs_free_counter ctr, uint64_t val)
742 {
743 percpu_counter_set(&mp->m_free[ctr].count, val);
744 }
745
746 int xfs_dec_freecounter(struct xfs_mount *mp, enum xfs_free_counter ctr,
747 uint64_t delta, bool rsvd);
748 void xfs_add_freecounter(struct xfs_mount *mp, enum xfs_free_counter ctr,
749 uint64_t delta);
750
xfs_dec_fdblocks(struct xfs_mount * mp,uint64_t delta,bool reserved)751 static inline int xfs_dec_fdblocks(struct xfs_mount *mp, uint64_t delta,
752 bool reserved)
753 {
754 return xfs_dec_freecounter(mp, XC_FREE_BLOCKS, delta, reserved);
755 }
756
xfs_add_fdblocks(struct xfs_mount * mp,uint64_t delta)757 static inline void xfs_add_fdblocks(struct xfs_mount *mp, uint64_t delta)
758 {
759 xfs_add_freecounter(mp, XC_FREE_BLOCKS, delta);
760 }
761
xfs_dec_frextents(struct xfs_mount * mp,uint64_t delta)762 static inline int xfs_dec_frextents(struct xfs_mount *mp, uint64_t delta)
763 {
764 return xfs_dec_freecounter(mp, XC_FREE_RTEXTENTS, delta, false);
765 }
766
xfs_add_frextents(struct xfs_mount * mp,uint64_t delta)767 static inline void xfs_add_frextents(struct xfs_mount *mp, uint64_t delta)
768 {
769 xfs_add_freecounter(mp, XC_FREE_RTEXTENTS, delta);
770 }
771
772 extern int xfs_readsb(xfs_mount_t *, int);
773 extern void xfs_freesb(xfs_mount_t *);
774 extern bool xfs_fs_writable(struct xfs_mount *mp, int level);
775 extern int xfs_sb_validate_fsb_count(struct xfs_sb *, uint64_t);
776
777 extern int xfs_dev_is_read_only(struct xfs_mount *, char *);
778
779 extern void xfs_set_low_space_thresholds(struct xfs_mount *);
780
781 int xfs_zero_extent(struct xfs_inode *ip, xfs_fsblock_t start_fsb,
782 xfs_off_t count_fsb);
783
784 struct xfs_error_cfg * xfs_error_get_cfg(struct xfs_mount *mp,
785 int error_class, int error);
786 void xfs_force_summary_recalc(struct xfs_mount *mp);
787 int xfs_add_incompat_log_feature(struct xfs_mount *mp, uint32_t feature);
788 bool xfs_clear_incompat_log_features(struct xfs_mount *mp);
789 void xfs_mod_delalloc(struct xfs_inode *ip, int64_t data_delta,
790 int64_t ind_delta);
xfs_mod_sb_delalloc(struct xfs_mount * mp,int64_t delta)791 static inline void xfs_mod_sb_delalloc(struct xfs_mount *mp, int64_t delta)
792 {
793 percpu_counter_add(&mp->m_delalloc_blks, delta);
794 }
795
796 #endif /* __XFS_MOUNT_H__ */
797