xref: /openbmc/linux/fs/xfs/xfs_mount.h (revision 110e6f26)
1 /*
2  * Copyright (c) 2000-2005 Silicon Graphics, Inc.
3  * All Rights Reserved.
4  *
5  * This program is free software; you can redistribute it and/or
6  * modify it under the terms of the GNU General Public License as
7  * published by the Free Software Foundation.
8  *
9  * This program is distributed in the hope that it would be useful,
10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
12  * GNU General Public License for more details.
13  *
14  * You should have received a copy of the GNU General Public License
15  * along with this program; if not, write the Free Software Foundation,
16  * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
17  */
18 #ifndef __XFS_MOUNT_H__
19 #define	__XFS_MOUNT_H__
20 
21 struct xlog;
22 struct xfs_inode;
23 struct xfs_mru_cache;
24 struct xfs_nameops;
25 struct xfs_ail;
26 struct xfs_quotainfo;
27 struct xfs_dir_ops;
28 struct xfs_da_geometry;
29 
30 /* dynamic preallocation free space thresholds, 5% down to 1% */
31 enum {
32 	XFS_LOWSP_1_PCNT = 0,
33 	XFS_LOWSP_2_PCNT,
34 	XFS_LOWSP_3_PCNT,
35 	XFS_LOWSP_4_PCNT,
36 	XFS_LOWSP_5_PCNT,
37 	XFS_LOWSP_MAX,
38 };
39 
40 typedef struct xfs_mount {
41 	struct super_block	*m_super;
42 	xfs_tid_t		m_tid;		/* next unused tid for fs */
43 	struct xfs_ail		*m_ail;		/* fs active log item list */
44 
45 	struct xfs_sb		m_sb;		/* copy of fs superblock */
46 	spinlock_t		m_sb_lock;	/* sb counter lock */
47 	struct percpu_counter	m_icount;	/* allocated inodes counter */
48 	struct percpu_counter	m_ifree;	/* free inodes counter */
49 	struct percpu_counter	m_fdblocks;	/* free block counter */
50 
51 	struct xfs_buf		*m_sb_bp;	/* buffer for superblock */
52 	char			*m_fsname;	/* filesystem name */
53 	int			m_fsname_len;	/* strlen of fs name */
54 	char			*m_rtname;	/* realtime device name */
55 	char			*m_logname;	/* external log device name */
56 	int			m_bsize;	/* fs logical block size */
57 	xfs_agnumber_t		m_agfrotor;	/* last ag where space found */
58 	xfs_agnumber_t		m_agirotor;	/* last ag dir inode alloced */
59 	spinlock_t		m_agirotor_lock;/* .. and lock protecting it */
60 	xfs_agnumber_t		m_maxagi;	/* highest inode alloc group */
61 	uint			m_readio_log;	/* min read size log bytes */
62 	uint			m_readio_blocks; /* min read size blocks */
63 	uint			m_writeio_log;	/* min write size log bytes */
64 	uint			m_writeio_blocks; /* min write size blocks */
65 	struct xfs_da_geometry	*m_dir_geo;	/* directory block geometry */
66 	struct xfs_da_geometry	*m_attr_geo;	/* attribute block geometry */
67 	struct xlog		*m_log;		/* log specific stuff */
68 	int			m_logbufs;	/* number of log buffers */
69 	int			m_logbsize;	/* size of each log buffer */
70 	uint			m_rsumlevels;	/* rt summary levels */
71 	uint			m_rsumsize;	/* size of rt summary, bytes */
72 	struct xfs_inode	*m_rbmip;	/* pointer to bitmap inode */
73 	struct xfs_inode	*m_rsumip;	/* pointer to summary inode */
74 	struct xfs_inode	*m_rootip;	/* pointer to root directory */
75 	struct xfs_quotainfo	*m_quotainfo;	/* disk quota information */
76 	xfs_buftarg_t		*m_ddev_targp;	/* saves taking the address */
77 	xfs_buftarg_t		*m_logdev_targp;/* ptr to log device */
78 	xfs_buftarg_t		*m_rtdev_targp;	/* ptr to rt device */
79 	__uint8_t		m_blkbit_log;	/* blocklog + NBBY */
80 	__uint8_t		m_blkbb_log;	/* blocklog - BBSHIFT */
81 	__uint8_t		m_agno_log;	/* log #ag's */
82 	__uint8_t		m_agino_log;	/* #bits for agino in inum */
83 	uint			m_inode_cluster_size;/* min inode buf size */
84 	uint			m_blockmask;	/* sb_blocksize-1 */
85 	uint			m_blockwsize;	/* sb_blocksize in words */
86 	uint			m_blockwmask;	/* blockwsize-1 */
87 	uint			m_alloc_mxr[2];	/* max alloc btree records */
88 	uint			m_alloc_mnr[2];	/* min alloc btree records */
89 	uint			m_bmap_dmxr[2];	/* max bmap btree records */
90 	uint			m_bmap_dmnr[2];	/* min bmap btree records */
91 	uint			m_inobt_mxr[2];	/* max inobt btree records */
92 	uint			m_inobt_mnr[2];	/* min inobt btree records */
93 	uint			m_ag_maxlevels;	/* XFS_AG_MAXLEVELS */
94 	uint			m_bm_maxlevels[2]; /* XFS_BM_MAXLEVELS */
95 	uint			m_in_maxlevels;	/* max inobt btree levels. */
96 	struct radix_tree_root	m_perag_tree;	/* per-ag accounting info */
97 	spinlock_t		m_perag_lock;	/* lock for m_perag_tree */
98 	struct mutex		m_growlock;	/* growfs mutex */
99 	int			m_fixedfsid[2];	/* unchanged for life of FS */
100 	uint			m_dmevmask;	/* DMI events for this FS */
101 	__uint64_t		m_flags;	/* global mount flags */
102 	int			m_ialloc_inos;	/* inodes in inode allocation */
103 	int			m_ialloc_blks;	/* blocks in inode allocation */
104 	int			m_ialloc_min_blks;/* min blocks in sparse inode
105 						   * allocation */
106 	int			m_inoalign_mask;/* mask sb_inoalignmt if used */
107 	uint			m_qflags;	/* quota status flags */
108 	struct xfs_trans_resv	m_resv;		/* precomputed res values */
109 	__uint64_t		m_maxicount;	/* maximum inode count */
110 	__uint64_t		m_resblks;	/* total reserved blocks */
111 	__uint64_t		m_resblks_avail;/* available reserved blocks */
112 	__uint64_t		m_resblks_save;	/* reserved blks @ remount,ro */
113 	int			m_dalign;	/* stripe unit */
114 	int			m_swidth;	/* stripe width */
115 	int			m_sinoalign;	/* stripe unit inode alignment */
116 	__uint8_t		m_sectbb_log;	/* sectlog - BBSHIFT */
117 	const struct xfs_nameops *m_dirnameops;	/* vector of dir name ops */
118 	const struct xfs_dir_ops *m_dir_inode_ops; /* vector of dir inode ops */
119 	const struct xfs_dir_ops *m_nondir_inode_ops; /* !dir inode ops */
120 	uint			m_chsize;	/* size of next field */
121 	atomic_t		m_active_trans;	/* number trans frozen */
122 	struct xfs_mru_cache	*m_filestream;  /* per-mount filestream data */
123 	struct delayed_work	m_reclaim_work;	/* background inode reclaim */
124 	struct delayed_work	m_eofblocks_work; /* background eof blocks
125 						     trimming */
126 	bool			m_update_sb;	/* sb needs update in mount */
127 	int64_t			m_low_space[XFS_LOWSP_MAX];
128 						/* low free space thresholds */
129 	struct xfs_kobj		m_kobj;
130 	struct xstats		m_stats;	/* per-fs stats */
131 
132 	struct workqueue_struct *m_buf_workqueue;
133 	struct workqueue_struct	*m_data_workqueue;
134 	struct workqueue_struct	*m_unwritten_workqueue;
135 	struct workqueue_struct	*m_cil_workqueue;
136 	struct workqueue_struct	*m_reclaim_workqueue;
137 	struct workqueue_struct	*m_log_workqueue;
138 	struct workqueue_struct *m_eofblocks_workqueue;
139 
140 	/*
141 	 * Generation of the filesysyem layout.  This is incremented by each
142 	 * growfs, and used by the pNFS server to ensure the client updates
143 	 * its view of the block device once it gets a layout that might
144 	 * reference the newly added blocks.  Does not need to be persistent
145 	 * as long as we only allow file system size increments, but if we
146 	 * ever support shrinks it would have to be persisted in addition
147 	 * to various other kinds of pain inflicted on the pNFS server.
148 	 */
149 	__uint32_t		m_generation;
150 
151 #ifdef DEBUG
152 	/*
153 	 * DEBUG mode instrumentation to test and/or trigger delayed allocation
154 	 * block killing in the event of failed writes. When enabled, all
155 	 * buffered writes are forced to fail. All delalloc blocks in the range
156 	 * of the write (including pre-existing delalloc blocks!) are tossed as
157 	 * part of the write failure error handling sequence.
158 	 */
159 	bool			m_fail_writes;
160 #endif
161 } xfs_mount_t;
162 
163 /*
164  * Flags for m_flags.
165  */
166 #define XFS_MOUNT_WSYNC		(1ULL << 0)	/* for nfs - all metadata ops
167 						   must be synchronous except
168 						   for space allocations */
169 #define XFS_MOUNT_WAS_CLEAN	(1ULL << 3)
170 #define XFS_MOUNT_FS_SHUTDOWN	(1ULL << 4)	/* atomic stop of all filesystem
171 						   operations, typically for
172 						   disk errors in metadata */
173 #define XFS_MOUNT_DISCARD	(1ULL << 5)	/* discard unused blocks */
174 #define XFS_MOUNT_NOALIGN	(1ULL << 7)	/* turn off stripe alignment
175 						   allocations */
176 #define XFS_MOUNT_ATTR2		(1ULL << 8)	/* allow use of attr2 format */
177 #define XFS_MOUNT_GRPID		(1ULL << 9)	/* group-ID assigned from directory */
178 #define XFS_MOUNT_NORECOVERY	(1ULL << 10)	/* no recovery - dirty fs */
179 #define XFS_MOUNT_DFLT_IOSIZE	(1ULL << 12)	/* set default i/o size */
180 #define XFS_MOUNT_SMALL_INUMS	(1ULL << 14)	/* user wants 32bit inodes */
181 #define XFS_MOUNT_32BITINODES	(1ULL << 15)	/* inode32 allocator active */
182 #define XFS_MOUNT_NOUUID	(1ULL << 16)	/* ignore uuid during mount */
183 #define XFS_MOUNT_BARRIER	(1ULL << 17)
184 #define XFS_MOUNT_IKEEP		(1ULL << 18)	/* keep empty inode clusters*/
185 #define XFS_MOUNT_SWALLOC	(1ULL << 19)	/* turn on stripe width
186 						 * allocation */
187 #define XFS_MOUNT_RDONLY	(1ULL << 20)	/* read-only fs */
188 #define XFS_MOUNT_DIRSYNC	(1ULL << 21)	/* synchronous directory ops */
189 #define XFS_MOUNT_COMPAT_IOSIZE	(1ULL << 22)	/* don't report large preferred
190 						 * I/O size in stat() */
191 #define XFS_MOUNT_FILESTREAMS	(1ULL << 24)	/* enable the filestreams
192 						   allocator */
193 #define XFS_MOUNT_NOATTR2	(1ULL << 25)	/* disable use of attr2 format */
194 
195 #define XFS_MOUNT_DAX		(1ULL << 62)	/* TEST ONLY! */
196 
197 
198 /*
199  * Default minimum read and write sizes.
200  */
201 #define XFS_READIO_LOG_LARGE	16
202 #define XFS_WRITEIO_LOG_LARGE	16
203 
204 /*
205  * Max and min values for mount-option defined I/O
206  * preallocation sizes.
207  */
208 #define XFS_MAX_IO_LOG		30	/* 1G */
209 #define XFS_MIN_IO_LOG		PAGE_SHIFT
210 
211 /*
212  * Synchronous read and write sizes.  This should be
213  * better for NFSv2 wsync filesystems.
214  */
215 #define	XFS_WSYNC_READIO_LOG	15	/* 32k */
216 #define	XFS_WSYNC_WRITEIO_LOG	14	/* 16k */
217 
218 /*
219  * Allow large block sizes to be reported to userspace programs if the
220  * "largeio" mount option is used.
221  *
222  * If compatibility mode is specified, simply return the basic unit of caching
223  * so that we don't get inefficient read/modify/write I/O from user apps.
224  * Otherwise....
225  *
226  * If the underlying volume is a stripe, then return the stripe width in bytes
227  * as the recommended I/O size. It is not a stripe and we've set a default
228  * buffered I/O size, return that, otherwise return the compat default.
229  */
230 static inline unsigned long
231 xfs_preferred_iosize(xfs_mount_t *mp)
232 {
233 	if (mp->m_flags & XFS_MOUNT_COMPAT_IOSIZE)
234 		return PAGE_SIZE;
235 	return (mp->m_swidth ?
236 		(mp->m_swidth << mp->m_sb.sb_blocklog) :
237 		((mp->m_flags & XFS_MOUNT_DFLT_IOSIZE) ?
238 			(1 << (int)MAX(mp->m_readio_log, mp->m_writeio_log)) :
239 			PAGE_SIZE));
240 }
241 
242 #define XFS_LAST_UNMOUNT_WAS_CLEAN(mp)	\
243 				((mp)->m_flags & XFS_MOUNT_WAS_CLEAN)
244 #define XFS_FORCED_SHUTDOWN(mp)	((mp)->m_flags & XFS_MOUNT_FS_SHUTDOWN)
245 void xfs_do_force_shutdown(struct xfs_mount *mp, int flags, char *fname,
246 		int lnnum);
247 #define xfs_force_shutdown(m,f)	\
248 	xfs_do_force_shutdown(m, f, __FILE__, __LINE__)
249 
250 #define SHUTDOWN_META_IO_ERROR	0x0001	/* write attempt to metadata failed */
251 #define SHUTDOWN_LOG_IO_ERROR	0x0002	/* write attempt to the log failed */
252 #define SHUTDOWN_FORCE_UMOUNT	0x0004	/* shutdown from a forced unmount */
253 #define SHUTDOWN_CORRUPT_INCORE	0x0008	/* corrupt in-memory data structures */
254 #define SHUTDOWN_REMOTE_REQ	0x0010	/* shutdown came from remote cell */
255 #define SHUTDOWN_DEVICE_REQ	0x0020	/* failed all paths to the device */
256 
257 /*
258  * Flags for xfs_mountfs
259  */
260 #define XFS_MFSI_QUIET		0x40	/* Be silent if mount errors found */
261 
262 static inline xfs_agnumber_t
263 xfs_daddr_to_agno(struct xfs_mount *mp, xfs_daddr_t d)
264 {
265 	xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
266 	do_div(ld, mp->m_sb.sb_agblocks);
267 	return (xfs_agnumber_t) ld;
268 }
269 
270 static inline xfs_agblock_t
271 xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
272 {
273 	xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
274 	return (xfs_agblock_t) do_div(ld, mp->m_sb.sb_agblocks);
275 }
276 
277 #ifdef DEBUG
278 static inline bool
279 xfs_mp_fail_writes(struct xfs_mount *mp)
280 {
281 	return mp->m_fail_writes;
282 }
283 #else
284 static inline bool
285 xfs_mp_fail_writes(struct xfs_mount *mp)
286 {
287 	return 0;
288 }
289 #endif
290 
291 /*
292  * Per-ag incore structure, copies of information in agf and agi, to improve the
293  * performance of allocation group selection.
294  */
295 typedef struct xfs_perag {
296 	struct xfs_mount *pag_mount;	/* owner filesystem */
297 	xfs_agnumber_t	pag_agno;	/* AG this structure belongs to */
298 	atomic_t	pag_ref;	/* perag reference count */
299 	char		pagf_init;	/* this agf's entry is initialized */
300 	char		pagi_init;	/* this agi's entry is initialized */
301 	char		pagf_metadata;	/* the agf is preferred to be metadata */
302 	char		pagi_inodeok;	/* The agi is ok for inodes */
303 	__uint8_t	pagf_levels[XFS_BTNUM_AGF];
304 					/* # of levels in bno & cnt btree */
305 	__uint32_t	pagf_flcount;	/* count of blocks in freelist */
306 	xfs_extlen_t	pagf_freeblks;	/* total free blocks */
307 	xfs_extlen_t	pagf_longest;	/* longest free space */
308 	__uint32_t	pagf_btreeblks;	/* # of blocks held in AGF btrees */
309 	xfs_agino_t	pagi_freecount;	/* number of free inodes */
310 	xfs_agino_t	pagi_count;	/* number of allocated inodes */
311 
312 	/*
313 	 * Inode allocation search lookup optimisation.
314 	 * If the pagino matches, the search for new inodes
315 	 * doesn't need to search the near ones again straight away
316 	 */
317 	xfs_agino_t	pagl_pagino;
318 	xfs_agino_t	pagl_leftrec;
319 	xfs_agino_t	pagl_rightrec;
320 	spinlock_t	pagb_lock;	/* lock for pagb_tree */
321 	struct rb_root	pagb_tree;	/* ordered tree of busy extents */
322 
323 	atomic_t        pagf_fstrms;    /* # of filestreams active in this AG */
324 
325 	spinlock_t	pag_ici_lock;	/* incore inode cache lock */
326 	struct radix_tree_root pag_ici_root;	/* incore inode cache root */
327 	int		pag_ici_reclaimable;	/* reclaimable inodes */
328 	struct mutex	pag_ici_reclaim_lock;	/* serialisation point */
329 	unsigned long	pag_ici_reclaim_cursor;	/* reclaim restart point */
330 
331 	/* buffer cache index */
332 	spinlock_t	pag_buf_lock;	/* lock for pag_buf_tree */
333 	struct rb_root	pag_buf_tree;	/* ordered tree of active buffers */
334 
335 	/* for rcu-safe freeing */
336 	struct rcu_head	rcu_head;
337 	int		pagb_count;	/* pagb slots in use */
338 } xfs_perag_t;
339 
340 extern void	xfs_uuid_table_free(void);
341 extern int	xfs_log_sbcount(xfs_mount_t *);
342 extern __uint64_t xfs_default_resblks(xfs_mount_t *mp);
343 extern int	xfs_mountfs(xfs_mount_t *mp);
344 extern int	xfs_initialize_perag(xfs_mount_t *mp, xfs_agnumber_t agcount,
345 				     xfs_agnumber_t *maxagi);
346 extern void	xfs_unmountfs(xfs_mount_t *);
347 
348 extern int	xfs_mod_icount(struct xfs_mount *mp, int64_t delta);
349 extern int	xfs_mod_ifree(struct xfs_mount *mp, int64_t delta);
350 extern int	xfs_mod_fdblocks(struct xfs_mount *mp, int64_t delta,
351 				 bool reserved);
352 extern int	xfs_mod_frextents(struct xfs_mount *mp, int64_t delta);
353 
354 extern struct xfs_buf *xfs_getsb(xfs_mount_t *, int);
355 extern int	xfs_readsb(xfs_mount_t *, int);
356 extern void	xfs_freesb(xfs_mount_t *);
357 extern bool	xfs_fs_writable(struct xfs_mount *mp, int level);
358 extern int	xfs_sb_validate_fsb_count(struct xfs_sb *, __uint64_t);
359 
360 extern int	xfs_dev_is_read_only(struct xfs_mount *, char *);
361 
362 extern void	xfs_set_low_space_thresholds(struct xfs_mount *);
363 
364 int	xfs_zero_extent(struct xfs_inode *ip, xfs_fsblock_t start_fsb,
365 			xfs_off_t count_fsb);
366 
367 #endif	/* __XFS_MOUNT_H__ */
368