xref: /linux/fs/xfs/xfs_mount.h (revision c148bc7535650fbfa95a1f571b9ffa2ab478ea33)
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 
233 	/*
234 	 * Bitsets of per-fs metadata that have been checked and/or are sick.
235 	 * Callers must hold m_sb_lock to access these two fields.
236 	 */
237 	uint8_t			m_fs_checked;
238 	uint8_t			m_fs_sick;
239 	/*
240 	 * Bitsets of rt metadata that have been checked and/or are sick.
241 	 * Callers must hold m_sb_lock to access this field.
242 	 */
243 	uint8_t			m_rt_checked;
244 	uint8_t			m_rt_sick;
245 
246 	/*
247 	 * End of read-mostly variables. Frequently written variables and locks
248 	 * should be placed below this comment from now on. The first variable
249 	 * here is marked as cacheline aligned so they it is separated from
250 	 * the read-mostly variables.
251 	 */
252 
253 	spinlock_t ____cacheline_aligned m_sb_lock; /* sb counter lock */
254 	struct percpu_counter	m_icount;	/* allocated inodes counter */
255 	struct percpu_counter	m_ifree;	/* free inodes counter */
256 
257 	struct xfs_freecounter	m_free[XC_FREE_NR];
258 
259 	/*
260 	 * Count of data device blocks reserved for delayed allocations,
261 	 * including indlen blocks.  Does not include allocated CoW staging
262 	 * extents or anything related to the rt device.
263 	 */
264 	struct percpu_counter	m_delalloc_blks;
265 
266 	/*
267 	 * RT version of the above.
268 	 */
269 	struct percpu_counter	m_delalloc_rtextents;
270 
271 	/*
272 	 * Global count of allocation btree blocks in use across all AGs. Only
273 	 * used when perag reservation is enabled. Helps prevent block
274 	 * reservation from attempting to reserve allocation btree blocks.
275 	 */
276 	atomic64_t		m_allocbt_blks;
277 
278 	struct xfs_groups	m_groups[XG_TYPE_MAX];
279 	struct delayed_work	m_reclaim_work;	/* background inode reclaim */
280 	struct xfs_zone_info	*m_zone_info;	/* zone allocator information */
281 	struct dentry		*m_debugfs;	/* debugfs parent */
282 	struct xfs_kobj		m_kobj;
283 	struct xfs_kobj		m_error_kobj;
284 	struct xfs_kobj		m_error_meta_kobj;
285 	struct xfs_error_cfg	m_error_cfg[XFS_ERR_CLASS_MAX][XFS_ERR_ERRNO_MAX];
286 	struct xstats		m_stats;	/* per-fs stats */
287 #ifdef CONFIG_XFS_ONLINE_SCRUB_STATS
288 	struct xchk_stats	*m_scrub_stats;
289 #endif
290 	struct xfs_kobj		m_zoned_kobj;
291 	xfs_agnumber_t		m_agfrotor;	/* last ag where space found */
292 	atomic_t		m_agirotor;	/* last ag dir inode alloced */
293 	atomic_t		m_rtgrotor;	/* last rtgroup rtpicked */
294 
295 	struct mutex		m_metafile_resv_lock;
296 	uint64_t		m_metafile_resv_target;
297 	uint64_t		m_metafile_resv_used;
298 	uint64_t		m_metafile_resv_avail;
299 
300 	/* Memory shrinker to throttle and reprioritize inodegc */
301 	struct shrinker		*m_inodegc_shrinker;
302 	/*
303 	 * Workqueue item so that we can coalesce multiple inode flush attempts
304 	 * into a single flush.
305 	 */
306 	struct work_struct	m_flush_inodes_work;
307 
308 	/*
309 	 * Generation of the filesysyem layout.  This is incremented by each
310 	 * growfs, and used by the pNFS server to ensure the client updates
311 	 * its view of the block device once it gets a layout that might
312 	 * reference the newly added blocks.  Does not need to be persistent
313 	 * as long as we only allow file system size increments, but if we
314 	 * ever support shrinks it would have to be persisted in addition
315 	 * to various other kinds of pain inflicted on the pNFS server.
316 	 */
317 	uint32_t		m_generation;
318 	struct mutex		m_growlock;	/* growfs mutex */
319 
320 #ifdef DEBUG
321 	/*
322 	 * Frequency with which errors are injected.  Replaces xfs_etest; the
323 	 * value stored in here is the inverse of the frequency with which the
324 	 * error triggers.  1 = always, 2 = half the time, etc.
325 	 */
326 	unsigned int		*m_errortag;
327 	struct xfs_kobj		m_errortag_kobj;
328 #endif
329 
330 	/* cpus that have inodes queued for inactivation */
331 	struct cpumask		m_inodegc_cpumask;
332 
333 	/* Hook to feed dirent updates to an active online repair. */
334 	struct xfs_hooks	m_dir_update_hooks;
335 } xfs_mount_t;
336 
337 #define M_IGEO(mp)		(&(mp)->m_ino_geo)
338 
339 /*
340  * Flags for m_features.
341  *
342  * These are all the active features in the filesystem, regardless of how
343  * they are configured.
344  */
345 #define XFS_FEAT_ATTR		(1ULL << 0)	/* xattrs present in fs */
346 #define XFS_FEAT_NLINK		(1ULL << 1)	/* 32 bit link counts */
347 #define XFS_FEAT_QUOTA		(1ULL << 2)	/* quota active */
348 #define XFS_FEAT_ALIGN		(1ULL << 3)	/* inode alignment */
349 #define XFS_FEAT_DALIGN		(1ULL << 4)	/* data alignment */
350 #define XFS_FEAT_LOGV2		(1ULL << 5)	/* version 2 logs */
351 #define XFS_FEAT_SECTOR		(1ULL << 6)	/* sector size > 512 bytes */
352 #define XFS_FEAT_EXTFLG		(1ULL << 7)	/* unwritten extents */
353 #define XFS_FEAT_ASCIICI	(1ULL << 8)	/* ASCII only case-insens. */
354 #define XFS_FEAT_LAZYSBCOUNT	(1ULL << 9)	/* Superblk counters */
355 #define XFS_FEAT_ATTR2		(1ULL << 10)	/* dynamic attr fork */
356 #define XFS_FEAT_PARENT		(1ULL << 11)	/* parent pointers */
357 #define XFS_FEAT_PROJID32	(1ULL << 12)	/* 32 bit project id */
358 #define XFS_FEAT_CRC		(1ULL << 13)	/* metadata CRCs */
359 #define XFS_FEAT_V3INODES	(1ULL << 14)	/* Version 3 inodes */
360 #define XFS_FEAT_PQUOTINO	(1ULL << 15)	/* non-shared proj/grp quotas */
361 #define XFS_FEAT_FTYPE		(1ULL << 16)	/* inode type in dir */
362 #define XFS_FEAT_FINOBT		(1ULL << 17)	/* free inode btree */
363 #define XFS_FEAT_RMAPBT		(1ULL << 18)	/* reverse map btree */
364 #define XFS_FEAT_REFLINK	(1ULL << 19)	/* reflinked files */
365 #define XFS_FEAT_SPINODES	(1ULL << 20)	/* sparse inode chunks */
366 #define XFS_FEAT_META_UUID	(1ULL << 21)	/* metadata UUID */
367 #define XFS_FEAT_REALTIME	(1ULL << 22)	/* realtime device present */
368 #define XFS_FEAT_INOBTCNT	(1ULL << 23)	/* inobt block counts */
369 #define XFS_FEAT_BIGTIME	(1ULL << 24)	/* large timestamps */
370 #define XFS_FEAT_NEEDSREPAIR	(1ULL << 25)	/* needs xfs_repair */
371 #define XFS_FEAT_NREXT64	(1ULL << 26)	/* large extent counters */
372 #define XFS_FEAT_EXCHANGE_RANGE	(1ULL << 27)	/* exchange range */
373 #define XFS_FEAT_METADIR	(1ULL << 28)	/* metadata directory tree */
374 #define XFS_FEAT_ZONED		(1ULL << 29)	/* zoned RT device */
375 
376 /* Mount features */
377 #define XFS_FEAT_NOLIFETIME	(1ULL << 47)	/* disable lifetime hints */
378 #define XFS_FEAT_NOATTR2	(1ULL << 48)	/* disable attr2 creation */
379 #define XFS_FEAT_NOALIGN	(1ULL << 49)	/* ignore alignment */
380 #define XFS_FEAT_ALLOCSIZE	(1ULL << 50)	/* user specified allocation size */
381 #define XFS_FEAT_LARGE_IOSIZE	(1ULL << 51)	/* report large preferred
382 						 * I/O size in stat() */
383 #define XFS_FEAT_WSYNC		(1ULL << 52)	/* synchronous metadata ops */
384 #define XFS_FEAT_DIRSYNC	(1ULL << 53)	/* synchronous directory ops */
385 #define XFS_FEAT_DISCARD	(1ULL << 54)	/* discard unused blocks */
386 #define XFS_FEAT_GRPID		(1ULL << 55)	/* group-ID assigned from directory */
387 #define XFS_FEAT_SMALL_INUMS	(1ULL << 56)	/* user wants 32bit inodes */
388 #define XFS_FEAT_IKEEP		(1ULL << 57)	/* keep empty inode clusters*/
389 #define XFS_FEAT_SWALLOC	(1ULL << 58)	/* stripe width allocation */
390 #define XFS_FEAT_FILESTREAMS	(1ULL << 59)	/* use filestreams allocator */
391 #define XFS_FEAT_DAX_ALWAYS	(1ULL << 60)	/* DAX always enabled */
392 #define XFS_FEAT_DAX_NEVER	(1ULL << 61)	/* DAX never enabled */
393 #define XFS_FEAT_NORECOVERY	(1ULL << 62)	/* no recovery - dirty fs */
394 #define XFS_FEAT_NOUUID		(1ULL << 63)	/* ignore uuid during mount */
395 
396 #define __XFS_HAS_FEAT(name, NAME) \
397 static inline bool xfs_has_ ## name (const struct xfs_mount *mp) \
398 { \
399 	return mp->m_features & XFS_FEAT_ ## NAME; \
400 }
401 
402 /* Some features can be added dynamically so they need a set wrapper, too. */
403 #define __XFS_ADD_FEAT(name, NAME) \
404 	__XFS_HAS_FEAT(name, NAME); \
405 static inline void xfs_add_ ## name (struct xfs_mount *mp) \
406 { \
407 	mp->m_features |= XFS_FEAT_ ## NAME; \
408 	xfs_sb_version_add ## name(&mp->m_sb); \
409 }
410 
411 /* Superblock features */
__XFS_ADD_FEAT(attr,ATTR)412 __XFS_ADD_FEAT(attr, ATTR)
413 __XFS_HAS_FEAT(nlink, NLINK)
414 __XFS_ADD_FEAT(quota, QUOTA)
415 __XFS_HAS_FEAT(dalign, DALIGN)
416 __XFS_HAS_FEAT(sector, SECTOR)
417 __XFS_HAS_FEAT(asciici, ASCIICI)
418 __XFS_HAS_FEAT(parent, PARENT)
419 __XFS_HAS_FEAT(ftype, FTYPE)
420 __XFS_HAS_FEAT(finobt, FINOBT)
421 __XFS_HAS_FEAT(rmapbt, RMAPBT)
422 __XFS_HAS_FEAT(reflink, REFLINK)
423 __XFS_HAS_FEAT(sparseinodes, SPINODES)
424 __XFS_HAS_FEAT(metauuid, META_UUID)
425 __XFS_HAS_FEAT(realtime, REALTIME)
426 __XFS_HAS_FEAT(inobtcounts, INOBTCNT)
427 __XFS_HAS_FEAT(bigtime, BIGTIME)
428 __XFS_HAS_FEAT(needsrepair, NEEDSREPAIR)
429 __XFS_HAS_FEAT(large_extent_counts, NREXT64)
430 __XFS_HAS_FEAT(exchange_range, EXCHANGE_RANGE)
431 __XFS_HAS_FEAT(metadir, METADIR)
432 __XFS_HAS_FEAT(zoned, ZONED)
433 __XFS_HAS_FEAT(nolifetime, NOLIFETIME)
434 
435 static inline bool xfs_has_rtgroups(const struct xfs_mount *mp)
436 {
437 	/* all metadir file systems also allow rtgroups */
438 	return xfs_has_metadir(mp);
439 }
440 
xfs_has_rtsb(const struct xfs_mount * mp)441 static inline bool xfs_has_rtsb(const struct xfs_mount *mp)
442 {
443 	/* all rtgroups filesystems with an rt section have an rtsb */
444 	return xfs_has_rtgroups(mp) &&
445 		xfs_has_realtime(mp) &&
446 		!xfs_has_zoned(mp);
447 }
448 
xfs_has_rtrmapbt(const struct xfs_mount * mp)449 static inline bool xfs_has_rtrmapbt(const struct xfs_mount *mp)
450 {
451 	return xfs_has_rtgroups(mp) && xfs_has_realtime(mp) &&
452 	       xfs_has_rmapbt(mp);
453 }
454 
xfs_has_rtreflink(const struct xfs_mount * mp)455 static inline bool xfs_has_rtreflink(const struct xfs_mount *mp)
456 {
457 	return xfs_has_metadir(mp) && xfs_has_realtime(mp) &&
458 	       xfs_has_reflink(mp);
459 }
460 
xfs_has_nonzoned(const struct xfs_mount * mp)461 static inline bool xfs_has_nonzoned(const struct xfs_mount *mp)
462 {
463 	return !xfs_has_zoned(mp);
464 }
465 
466 /*
467  * Some features are always on for v5 file systems, allow the compiler to
468  * eliminiate dead code when building without v4 support.
469  */
470 #define __XFS_HAS_V4_FEAT(name, NAME) \
471 static inline bool xfs_has_ ## name (struct xfs_mount *mp) \
472 { \
473 	return !IS_ENABLED(CONFIG_XFS_SUPPORT_V4) || \
474 		(mp->m_features & XFS_FEAT_ ## NAME); \
475 }
476 
477 #define __XFS_ADD_V4_FEAT(name, NAME) \
478 	__XFS_HAS_V4_FEAT(name, NAME); \
479 static inline void xfs_add_ ## name (struct xfs_mount *mp) \
480 { \
481 	if (IS_ENABLED(CONFIG_XFS_SUPPORT_V4)) { \
482 		mp->m_features |= XFS_FEAT_ ## NAME; \
483 		xfs_sb_version_add ## name(&mp->m_sb); \
484 	} \
485 }
486 
__XFS_HAS_V4_FEAT(align,ALIGN)487 __XFS_HAS_V4_FEAT(align, ALIGN)
488 __XFS_HAS_V4_FEAT(logv2, LOGV2)
489 __XFS_HAS_V4_FEAT(extflg, EXTFLG)
490 __XFS_HAS_V4_FEAT(lazysbcount, LAZYSBCOUNT)
491 __XFS_ADD_V4_FEAT(attr2, ATTR2)
492 __XFS_ADD_V4_FEAT(projid32, PROJID32)
493 __XFS_HAS_V4_FEAT(v3inodes, V3INODES)
494 __XFS_HAS_V4_FEAT(crc, CRC)
495 __XFS_HAS_V4_FEAT(pquotino, PQUOTINO)
496 
497 /*
498  * Mount features
499  *
500  * These do not change dynamically - features that can come and go, such as 32
501  * bit inodes and read-only state, are kept as operational state rather than
502  * features.
503  */
504 __XFS_HAS_FEAT(noattr2, NOATTR2)
505 __XFS_HAS_FEAT(noalign, NOALIGN)
506 __XFS_HAS_FEAT(allocsize, ALLOCSIZE)
507 __XFS_HAS_FEAT(large_iosize, LARGE_IOSIZE)
508 __XFS_HAS_FEAT(wsync, WSYNC)
509 __XFS_HAS_FEAT(dirsync, DIRSYNC)
510 __XFS_HAS_FEAT(discard, DISCARD)
511 __XFS_HAS_FEAT(grpid, GRPID)
512 __XFS_HAS_FEAT(small_inums, SMALL_INUMS)
513 __XFS_HAS_FEAT(ikeep, IKEEP)
514 __XFS_HAS_FEAT(swalloc, SWALLOC)
515 __XFS_HAS_FEAT(filestreams, FILESTREAMS)
516 __XFS_HAS_FEAT(dax_always, DAX_ALWAYS)
517 __XFS_HAS_FEAT(dax_never, DAX_NEVER)
518 __XFS_HAS_FEAT(norecovery, NORECOVERY)
519 __XFS_HAS_FEAT(nouuid, NOUUID)
520 
521 /*
522  * Operational mount state flags
523  *
524  * Use these with atomic bit ops only!
525  */
526 #define XFS_OPSTATE_UNMOUNTING		0	/* filesystem is unmounting */
527 #define XFS_OPSTATE_CLEAN		1	/* mount was clean */
528 #define XFS_OPSTATE_SHUTDOWN		2	/* stop all fs operations */
529 #define XFS_OPSTATE_INODE32		3	/* inode32 allocator active */
530 #define XFS_OPSTATE_READONLY		4	/* read-only fs */
531 
532 /*
533  * If set, inactivation worker threads will be scheduled to process queued
534  * inodegc work.  If not, queued inodes remain in memory waiting to be
535  * processed.
536  */
537 #define XFS_OPSTATE_INODEGC_ENABLED	5
538 /*
539  * If set, background speculative prealloc gc worker threads will be scheduled
540  * to process queued blockgc work.  If not, inodes retain their preallocations
541  * until explicitly deleted.
542  */
543 #define XFS_OPSTATE_BLOCKGC_ENABLED	6
544 
545 /* Kernel has logged a warning about pNFS being used on this fs. */
546 #define XFS_OPSTATE_WARNED_PNFS		7
547 /* Kernel has logged a warning about online fsck being used on this fs. */
548 #define XFS_OPSTATE_WARNED_SCRUB	8
549 /* Kernel has logged a warning about shrink being used on this fs. */
550 #define XFS_OPSTATE_WARNED_SHRINK	9
551 /* Kernel has logged a warning about logged xattr updates being used. */
552 #define XFS_OPSTATE_WARNED_LARP		10
553 /* Mount time quotacheck is running */
554 #define XFS_OPSTATE_QUOTACHECK_RUNNING	11
555 /* Do we want to clear log incompat flags? */
556 #define XFS_OPSTATE_UNSET_LOG_INCOMPAT	12
557 /* Filesystem can use logged extended attributes */
558 #define XFS_OPSTATE_USE_LARP		13
559 /* Kernel has logged a warning about blocksize > pagesize on this fs. */
560 #define XFS_OPSTATE_WARNED_LBS		14
561 /* Kernel has logged a warning about exchange-range being used on this fs. */
562 #define XFS_OPSTATE_WARNED_EXCHRANGE	15
563 /* Kernel has logged a warning about parent pointers being used on this fs. */
564 #define XFS_OPSTATE_WARNED_PPTR		16
565 /* Kernel has logged a warning about metadata dirs being used on this fs. */
566 #define XFS_OPSTATE_WARNED_METADIR	17
567 /* Filesystem should use qflags to determine quotaon status */
568 #define XFS_OPSTATE_RESUMING_QUOTAON	18
569 /* Kernel has logged a warning about zoned RT device being used on this fs. */
570 #define XFS_OPSTATE_WARNED_ZONED	19
571 /* (Zoned) GC is in progress */
572 #define XFS_OPSTATE_ZONEGC_RUNNING	20
573 
574 #define __XFS_IS_OPSTATE(name, NAME) \
575 static inline bool xfs_is_ ## name (struct xfs_mount *mp) \
576 { \
577 	return test_bit(XFS_OPSTATE_ ## NAME, &mp->m_opstate); \
578 } \
579 static inline bool xfs_clear_ ## name (struct xfs_mount *mp) \
580 { \
581 	return test_and_clear_bit(XFS_OPSTATE_ ## NAME, &mp->m_opstate); \
582 } \
583 static inline bool xfs_set_ ## name (struct xfs_mount *mp) \
584 { \
585 	return test_and_set_bit(XFS_OPSTATE_ ## NAME, &mp->m_opstate); \
586 }
587 
588 __XFS_IS_OPSTATE(unmounting, UNMOUNTING)
589 __XFS_IS_OPSTATE(clean, CLEAN)
590 __XFS_IS_OPSTATE(shutdown, SHUTDOWN)
591 __XFS_IS_OPSTATE(inode32, INODE32)
592 __XFS_IS_OPSTATE(readonly, READONLY)
593 __XFS_IS_OPSTATE(inodegc_enabled, INODEGC_ENABLED)
594 __XFS_IS_OPSTATE(blockgc_enabled, BLOCKGC_ENABLED)
595 #ifdef CONFIG_XFS_QUOTA
596 __XFS_IS_OPSTATE(quotacheck_running, QUOTACHECK_RUNNING)
597 __XFS_IS_OPSTATE(resuming_quotaon, RESUMING_QUOTAON)
598 #else
599 static inline bool xfs_is_quotacheck_running(struct xfs_mount *mp)
600 {
601 	return false;
602 }
603 static inline bool xfs_is_resuming_quotaon(struct xfs_mount *mp)
604 {
605 	return false;
606 }
607 static inline void xfs_set_resuming_quotaon(struct xfs_mount *m)
608 {
609 }
610 static inline bool xfs_clear_resuming_quotaon(struct xfs_mount *mp)
611 {
612 	return false;
613 }
614 #endif /* CONFIG_XFS_QUOTA */
615 __XFS_IS_OPSTATE(done_with_log_incompat, UNSET_LOG_INCOMPAT)
616 __XFS_IS_OPSTATE(using_logged_xattrs, USE_LARP)
617 __XFS_IS_OPSTATE(zonegc_running, ZONEGC_RUNNING)
618 
619 static inline bool
620 xfs_should_warn(struct xfs_mount *mp, long nr)
621 {
622 	return !test_and_set_bit(nr, &mp->m_opstate);
623 }
624 
625 #define XFS_OPSTATE_STRINGS \
626 	{ (1UL << XFS_OPSTATE_UNMOUNTING),		"unmounting" }, \
627 	{ (1UL << XFS_OPSTATE_CLEAN),			"clean" }, \
628 	{ (1UL << XFS_OPSTATE_SHUTDOWN),		"shutdown" }, \
629 	{ (1UL << XFS_OPSTATE_INODE32),			"inode32" }, \
630 	{ (1UL << XFS_OPSTATE_READONLY),		"read_only" }, \
631 	{ (1UL << XFS_OPSTATE_INODEGC_ENABLED),		"inodegc" }, \
632 	{ (1UL << XFS_OPSTATE_BLOCKGC_ENABLED),		"blockgc" }, \
633 	{ (1UL << XFS_OPSTATE_WARNED_SCRUB),		"wscrub" }, \
634 	{ (1UL << XFS_OPSTATE_WARNED_SHRINK),		"wshrink" }, \
635 	{ (1UL << XFS_OPSTATE_WARNED_LARP),		"wlarp" }, \
636 	{ (1UL << XFS_OPSTATE_QUOTACHECK_RUNNING),	"quotacheck" }, \
637 	{ (1UL << XFS_OPSTATE_UNSET_LOG_INCOMPAT),	"unset_log_incompat" }, \
638 	{ (1UL << XFS_OPSTATE_USE_LARP),		"logged_xattrs" }
639 
640 /*
641  * Max and min values for mount-option defined I/O
642  * preallocation sizes.
643  */
644 #define XFS_MAX_IO_LOG		30	/* 1G */
645 #define XFS_MIN_IO_LOG		PAGE_SHIFT
646 
647 void xfs_do_force_shutdown(struct xfs_mount *mp, uint32_t flags, char *fname,
648 		int lnnum);
649 #define xfs_force_shutdown(m,f)	\
650 	xfs_do_force_shutdown(m, f, __FILE__, __LINE__)
651 
652 #define SHUTDOWN_META_IO_ERROR	(1u << 0) /* write attempt to metadata failed */
653 #define SHUTDOWN_LOG_IO_ERROR	(1u << 1) /* write attempt to the log failed */
654 #define SHUTDOWN_FORCE_UMOUNT	(1u << 2) /* shutdown from a forced unmount */
655 #define SHUTDOWN_CORRUPT_INCORE	(1u << 3) /* corrupt in-memory structures */
656 #define SHUTDOWN_CORRUPT_ONDISK	(1u << 4)  /* corrupt metadata on device */
657 #define SHUTDOWN_DEVICE_REMOVED	(1u << 5) /* device removed underneath us */
658 
659 #define XFS_SHUTDOWN_STRINGS \
660 	{ SHUTDOWN_META_IO_ERROR,	"metadata_io" }, \
661 	{ SHUTDOWN_LOG_IO_ERROR,	"log_io" }, \
662 	{ SHUTDOWN_FORCE_UMOUNT,	"force_umount" }, \
663 	{ SHUTDOWN_CORRUPT_INCORE,	"corruption" }, \
664 	{ SHUTDOWN_DEVICE_REMOVED,	"device_removed" }
665 
666 /*
667  * Flags for xfs_mountfs
668  */
669 #define XFS_MFSI_QUIET		0x40	/* Be silent if mount errors found */
670 
671 static inline xfs_agnumber_t
xfs_daddr_to_agno(struct xfs_mount * mp,xfs_daddr_t d)672 xfs_daddr_to_agno(struct xfs_mount *mp, xfs_daddr_t d)
673 {
674 	xfs_rfsblock_t ld = XFS_BB_TO_FSBT(mp, d);
675 	do_div(ld, mp->m_sb.sb_agblocks);
676 	return (xfs_agnumber_t) ld;
677 }
678 
679 static inline xfs_agblock_t
xfs_daddr_to_agbno(struct xfs_mount * mp,xfs_daddr_t d)680 xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
681 {
682 	xfs_rfsblock_t ld = XFS_BB_TO_FSBT(mp, d);
683 	return (xfs_agblock_t) do_div(ld, mp->m_sb.sb_agblocks);
684 }
685 
686 extern void	xfs_uuid_table_free(void);
687 uint64_t	xfs_default_resblks(struct xfs_mount *mp,
688 			enum xfs_free_counter ctr);
689 extern int	xfs_mountfs(xfs_mount_t *mp);
690 extern void	xfs_unmountfs(xfs_mount_t *);
691 
692 /*
693  * Deltas for the block count can vary from 1 to very large, but lock contention
694  * only occurs on frequent small block count updates such as in the delayed
695  * allocation path for buffered writes (page a time updates). Hence we set
696  * a large batch count (1024) to minimise global counter updates except when
697  * we get near to ENOSPC and we have to be very accurate with our updates.
698  */
699 #define XFS_FDBLOCKS_BATCH	1024
700 
701 uint64_t xfs_freecounter_unavailable(struct xfs_mount *mp,
702 		enum xfs_free_counter ctr);
703 
704 /*
705  * Sum up the freecount, but never return negative values.
706  */
xfs_sum_freecounter(struct xfs_mount * mp,enum xfs_free_counter ctr)707 static inline s64 xfs_sum_freecounter(struct xfs_mount *mp,
708 		enum xfs_free_counter ctr)
709 {
710 	return percpu_counter_sum_positive(&mp->m_free[ctr].count);
711 }
712 
713 /*
714  * Same as above, but does return negative values.  Mostly useful for
715  * special cases like repair and tracing.
716  */
xfs_sum_freecounter_raw(struct xfs_mount * mp,enum xfs_free_counter ctr)717 static inline s64 xfs_sum_freecounter_raw(struct xfs_mount *mp,
718 		enum xfs_free_counter ctr)
719 {
720 	return percpu_counter_sum(&mp->m_free[ctr].count);
721 }
722 
723 /*
724  * This just provides and estimate without the cpu-local updates, use
725  * xfs_sum_freecounter for the exact value.
726  */
xfs_estimate_freecounter(struct xfs_mount * mp,enum xfs_free_counter ctr)727 static inline s64 xfs_estimate_freecounter(struct xfs_mount *mp,
728 		enum xfs_free_counter ctr)
729 {
730 	return percpu_counter_read_positive(&mp->m_free[ctr].count);
731 }
732 
xfs_compare_freecounter(struct xfs_mount * mp,enum xfs_free_counter ctr,s64 rhs,s32 batch)733 static inline int xfs_compare_freecounter(struct xfs_mount *mp,
734 		enum xfs_free_counter ctr, s64 rhs, s32 batch)
735 {
736 	return __percpu_counter_compare(&mp->m_free[ctr].count, rhs, batch);
737 }
738 
xfs_set_freecounter(struct xfs_mount * mp,enum xfs_free_counter ctr,uint64_t val)739 static inline void xfs_set_freecounter(struct xfs_mount *mp,
740 		enum xfs_free_counter ctr, uint64_t val)
741 {
742 	percpu_counter_set(&mp->m_free[ctr].count, val);
743 }
744 
745 int xfs_dec_freecounter(struct xfs_mount *mp, enum xfs_free_counter ctr,
746 		uint64_t delta, bool rsvd);
747 void xfs_add_freecounter(struct xfs_mount *mp, enum xfs_free_counter ctr,
748 		uint64_t delta);
749 
xfs_dec_fdblocks(struct xfs_mount * mp,uint64_t delta,bool reserved)750 static inline int xfs_dec_fdblocks(struct xfs_mount *mp, uint64_t delta,
751 		bool reserved)
752 {
753 	return xfs_dec_freecounter(mp, XC_FREE_BLOCKS, delta, reserved);
754 }
755 
xfs_add_fdblocks(struct xfs_mount * mp,uint64_t delta)756 static inline void xfs_add_fdblocks(struct xfs_mount *mp, uint64_t delta)
757 {
758 	xfs_add_freecounter(mp, XC_FREE_BLOCKS, delta);
759 }
760 
xfs_dec_frextents(struct xfs_mount * mp,uint64_t delta)761 static inline int xfs_dec_frextents(struct xfs_mount *mp, uint64_t delta)
762 {
763 	return xfs_dec_freecounter(mp, XC_FREE_RTEXTENTS, delta, false);
764 }
765 
xfs_add_frextents(struct xfs_mount * mp,uint64_t delta)766 static inline void xfs_add_frextents(struct xfs_mount *mp, uint64_t delta)
767 {
768 	xfs_add_freecounter(mp, XC_FREE_RTEXTENTS, delta);
769 }
770 
771 extern int	xfs_readsb(xfs_mount_t *, int);
772 extern void	xfs_freesb(xfs_mount_t *);
773 extern bool	xfs_fs_writable(struct xfs_mount *mp, int level);
774 extern int	xfs_sb_validate_fsb_count(struct xfs_sb *, uint64_t);
775 
776 extern int	xfs_dev_is_read_only(struct xfs_mount *, char *);
777 
778 extern void	xfs_set_low_space_thresholds(struct xfs_mount *);
779 
780 int	xfs_zero_extent(struct xfs_inode *ip, xfs_fsblock_t start_fsb,
781 			xfs_off_t count_fsb);
782 
783 struct xfs_error_cfg * xfs_error_get_cfg(struct xfs_mount *mp,
784 		int error_class, int error);
785 void xfs_force_summary_recalc(struct xfs_mount *mp);
786 int xfs_add_incompat_log_feature(struct xfs_mount *mp, uint32_t feature);
787 bool xfs_clear_incompat_log_features(struct xfs_mount *mp);
788 void xfs_mod_delalloc(struct xfs_inode *ip, int64_t data_delta,
789 		int64_t ind_delta);
xfs_mod_sb_delalloc(struct xfs_mount * mp,int64_t delta)790 static inline void xfs_mod_sb_delalloc(struct xfs_mount *mp, int64_t delta)
791 {
792 	percpu_counter_add(&mp->m_delalloc_blks, delta);
793 }
794 
795 #endif	/* __XFS_MOUNT_H__ */
796