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