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.
58 #define XFS_ERR_RETRY_FOREVER -1
61 * Although retry_timeout is in jiffies which is normally an unsigned long,
62 * we limit the retry timeout to 86400 seconds, or one day. So even a
63 * signed 32-bit long is sufficient for a HZ value up to 24855. Making it
64 * signed lets us store the special "-1" value, meaning retry forever.
66 struct xfs_error_cfg {
69 long retry_timeout; /* in jiffies, -1 = infinite */
72 typedef struct xfs_mount {
73 struct super_block *m_super;
74 xfs_tid_t m_tid; /* next unused tid for fs */
75 struct xfs_ail *m_ail; /* fs active log item list */
77 struct xfs_sb m_sb; /* copy of fs superblock */
78 spinlock_t m_sb_lock; /* sb counter lock */
79 struct percpu_counter m_icount; /* allocated inodes counter */
80 struct percpu_counter m_ifree; /* free inodes counter */
81 struct percpu_counter m_fdblocks; /* free block counter */
83 struct xfs_buf *m_sb_bp; /* buffer for superblock */
84 char *m_fsname; /* filesystem name */
85 int m_fsname_len; /* strlen of fs name */
86 char *m_rtname; /* realtime device name */
87 char *m_logname; /* external log device name */
88 int m_bsize; /* fs logical block size */
89 xfs_agnumber_t m_agfrotor; /* last ag where space found */
90 xfs_agnumber_t m_agirotor; /* last ag dir inode alloced */
91 spinlock_t m_agirotor_lock;/* .. and lock protecting it */
92 xfs_agnumber_t m_maxagi; /* highest inode alloc group */
93 uint m_readio_log; /* min read size log bytes */
94 uint m_readio_blocks; /* min read size blocks */
95 uint m_writeio_log; /* min write size log bytes */
96 uint m_writeio_blocks; /* min write size blocks */
97 struct xfs_da_geometry *m_dir_geo; /* directory block geometry */
98 struct xfs_da_geometry *m_attr_geo; /* attribute block geometry */
99 struct xlog *m_log; /* log specific stuff */
100 int m_logbufs; /* number of log buffers */
101 int m_logbsize; /* size of each log buffer */
102 uint m_rsumlevels; /* rt summary levels */
103 uint m_rsumsize; /* size of rt summary, bytes */
104 struct xfs_inode *m_rbmip; /* pointer to bitmap inode */
105 struct xfs_inode *m_rsumip; /* pointer to summary inode */
106 struct xfs_inode *m_rootip; /* pointer to root directory */
107 struct xfs_quotainfo *m_quotainfo; /* disk quota information */
108 xfs_buftarg_t *m_ddev_targp; /* saves taking the address */
109 xfs_buftarg_t *m_logdev_targp;/* ptr to log device */
110 xfs_buftarg_t *m_rtdev_targp; /* ptr to rt device */
111 __uint8_t m_blkbit_log; /* blocklog + NBBY */
112 __uint8_t m_blkbb_log; /* blocklog - BBSHIFT */
113 __uint8_t m_agno_log; /* log #ag's */
114 __uint8_t m_agino_log; /* #bits for agino in inum */
115 uint m_inode_cluster_size;/* min inode buf size */
116 uint m_blockmask; /* sb_blocksize-1 */
117 uint m_blockwsize; /* sb_blocksize in words */
118 uint m_blockwmask; /* blockwsize-1 */
119 uint m_alloc_mxr[2]; /* max alloc btree records */
120 uint m_alloc_mnr[2]; /* min alloc btree records */
121 uint m_bmap_dmxr[2]; /* max bmap btree records */
122 uint m_bmap_dmnr[2]; /* min bmap btree records */
123 uint m_inobt_mxr[2]; /* max inobt btree records */
124 uint m_inobt_mnr[2]; /* min inobt btree records */
125 uint m_rmap_mxr[2]; /* max rmap btree records */
126 uint m_rmap_mnr[2]; /* min rmap btree records */
127 uint m_ag_maxlevels; /* XFS_AG_MAXLEVELS */
128 uint m_bm_maxlevels[2]; /* XFS_BM_MAXLEVELS */
129 uint m_in_maxlevels; /* max inobt btree levels. */
130 uint m_rmap_maxlevels; /* max rmap btree levels */
131 xfs_extlen_t m_ag_prealloc_blocks; /* reserved ag blocks */
132 uint m_alloc_set_aside; /* space we can't use */
133 uint m_ag_max_usable; /* max space per AG */
134 struct radix_tree_root m_perag_tree; /* per-ag accounting info */
135 spinlock_t m_perag_lock; /* lock for m_perag_tree */
136 struct mutex m_growlock; /* growfs mutex */
137 int m_fixedfsid[2]; /* unchanged for life of FS */
138 uint m_dmevmask; /* DMI events for this FS */
139 __uint64_t m_flags; /* global mount flags */
140 int m_ialloc_inos; /* inodes in inode allocation */
141 int m_ialloc_blks; /* blocks in inode allocation */
142 int m_ialloc_min_blks;/* min blocks in sparse inode
144 int m_inoalign_mask;/* mask sb_inoalignmt if used */
145 uint m_qflags; /* quota status flags */
146 struct xfs_trans_resv m_resv; /* precomputed res values */
147 __uint64_t m_maxicount; /* maximum inode count */
148 __uint64_t m_resblks; /* total reserved blocks */
149 __uint64_t m_resblks_avail;/* available reserved blocks */
150 __uint64_t m_resblks_save; /* reserved blks @ remount,ro */
151 int m_dalign; /* stripe unit */
152 int m_swidth; /* stripe width */
153 int m_sinoalign; /* stripe unit inode alignment */
154 __uint8_t m_sectbb_log; /* sectlog - BBSHIFT */
155 const struct xfs_nameops *m_dirnameops; /* vector of dir name ops */
156 const struct xfs_dir_ops *m_dir_inode_ops; /* vector of dir inode ops */
157 const struct xfs_dir_ops *m_nondir_inode_ops; /* !dir inode ops */
158 uint m_chsize; /* size of next field */
159 atomic_t m_active_trans; /* number trans frozen */
160 struct xfs_mru_cache *m_filestream; /* per-mount filestream data */
161 struct delayed_work m_reclaim_work; /* background inode reclaim */
162 struct delayed_work m_eofblocks_work; /* background eof blocks
164 bool m_update_sb; /* sb needs update in mount */
165 int64_t m_low_space[XFS_LOWSP_MAX];
166 /* low free space thresholds */
167 struct xfs_kobj m_kobj;
168 struct xfs_kobj m_error_kobj;
169 struct xfs_kobj m_error_meta_kobj;
170 struct xfs_error_cfg m_error_cfg[XFS_ERR_CLASS_MAX][XFS_ERR_ERRNO_MAX];
171 struct xstats m_stats; /* per-fs stats */
173 struct workqueue_struct *m_buf_workqueue;
174 struct workqueue_struct *m_data_workqueue;
175 struct workqueue_struct *m_unwritten_workqueue;
176 struct workqueue_struct *m_cil_workqueue;
177 struct workqueue_struct *m_reclaim_workqueue;
178 struct workqueue_struct *m_log_workqueue;
179 struct workqueue_struct *m_eofblocks_workqueue;
182 * Generation of the filesysyem layout. This is incremented by each
183 * growfs, and used by the pNFS server to ensure the client updates
184 * its view of the block device once it gets a layout that might
185 * reference the newly added blocks. Does not need to be persistent
186 * as long as we only allow file system size increments, but if we
187 * ever support shrinks it would have to be persisted in addition
188 * to various other kinds of pain inflicted on the pNFS server.
190 __uint32_t m_generation;
195 * DEBUG mode instrumentation to test and/or trigger delayed allocation
196 * block killing in the event of failed writes. When enabled, all
197 * buffered writes are forced to fail. All delalloc blocks in the range
198 * of the write (including pre-existing delalloc blocks!) are tossed as
199 * part of the write failure error handling sequence.
208 #define XFS_MOUNT_WSYNC (1ULL << 0) /* for nfs - all metadata ops
209 must be synchronous except
210 for space allocations */
211 #define XFS_MOUNT_UNMOUNTING (1ULL << 1) /* filesystem is unmounting */
212 #define XFS_MOUNT_WAS_CLEAN (1ULL << 3)
213 #define XFS_MOUNT_FS_SHUTDOWN (1ULL << 4) /* atomic stop of all filesystem
214 operations, typically for
215 disk errors in metadata */
216 #define XFS_MOUNT_DISCARD (1ULL << 5) /* discard unused blocks */
217 #define XFS_MOUNT_NOALIGN (1ULL << 7) /* turn off stripe alignment
219 #define XFS_MOUNT_ATTR2 (1ULL << 8) /* allow use of attr2 format */
220 #define XFS_MOUNT_GRPID (1ULL << 9) /* group-ID assigned from directory */
221 #define XFS_MOUNT_NORECOVERY (1ULL << 10) /* no recovery - dirty fs */
222 #define XFS_MOUNT_DFLT_IOSIZE (1ULL << 12) /* set default i/o size */
223 #define XFS_MOUNT_SMALL_INUMS (1ULL << 14) /* user wants 32bit inodes */
224 #define XFS_MOUNT_32BITINODES (1ULL << 15) /* inode32 allocator active */
225 #define XFS_MOUNT_NOUUID (1ULL << 16) /* ignore uuid during mount */
226 #define XFS_MOUNT_BARRIER (1ULL << 17)
227 #define XFS_MOUNT_IKEEP (1ULL << 18) /* keep empty inode clusters*/
228 #define XFS_MOUNT_SWALLOC (1ULL << 19) /* turn on stripe width
230 #define XFS_MOUNT_RDONLY (1ULL << 20) /* read-only fs */
231 #define XFS_MOUNT_DIRSYNC (1ULL << 21) /* synchronous directory ops */
232 #define XFS_MOUNT_COMPAT_IOSIZE (1ULL << 22) /* don't report large preferred
233 * I/O size in stat() */
234 #define XFS_MOUNT_FILESTREAMS (1ULL << 24) /* enable the filestreams
236 #define XFS_MOUNT_NOATTR2 (1ULL << 25) /* disable use of attr2 format */
238 #define XFS_MOUNT_DAX (1ULL << 62) /* TEST ONLY! */
242 * Default minimum read and write sizes.
244 #define XFS_READIO_LOG_LARGE 16
245 #define XFS_WRITEIO_LOG_LARGE 16
248 * Max and min values for mount-option defined I/O
249 * preallocation sizes.
251 #define XFS_MAX_IO_LOG 30 /* 1G */
252 #define XFS_MIN_IO_LOG PAGE_SHIFT
255 * Synchronous read and write sizes. This should be
256 * better for NFSv2 wsync filesystems.
258 #define XFS_WSYNC_READIO_LOG 15 /* 32k */
259 #define XFS_WSYNC_WRITEIO_LOG 14 /* 16k */
262 * Allow large block sizes to be reported to userspace programs if the
263 * "largeio" mount option is used.
265 * If compatibility mode is specified, simply return the basic unit of caching
266 * so that we don't get inefficient read/modify/write I/O from user apps.
269 * If the underlying volume is a stripe, then return the stripe width in bytes
270 * as the recommended I/O size. It is not a stripe and we've set a default
271 * buffered I/O size, return that, otherwise return the compat default.
273 static inline unsigned long
274 xfs_preferred_iosize(xfs_mount_t *mp)
276 if (mp->m_flags & XFS_MOUNT_COMPAT_IOSIZE)
278 return (mp->m_swidth ?
279 (mp->m_swidth << mp->m_sb.sb_blocklog) :
280 ((mp->m_flags & XFS_MOUNT_DFLT_IOSIZE) ?
281 (1 << (int)MAX(mp->m_readio_log, mp->m_writeio_log)) :
285 #define XFS_LAST_UNMOUNT_WAS_CLEAN(mp) \
286 ((mp)->m_flags & XFS_MOUNT_WAS_CLEAN)
287 #define XFS_FORCED_SHUTDOWN(mp) ((mp)->m_flags & XFS_MOUNT_FS_SHUTDOWN)
288 void xfs_do_force_shutdown(struct xfs_mount *mp, int flags, char *fname,
290 #define xfs_force_shutdown(m,f) \
291 xfs_do_force_shutdown(m, f, __FILE__, __LINE__)
293 #define SHUTDOWN_META_IO_ERROR 0x0001 /* write attempt to metadata failed */
294 #define SHUTDOWN_LOG_IO_ERROR 0x0002 /* write attempt to the log failed */
295 #define SHUTDOWN_FORCE_UMOUNT 0x0004 /* shutdown from a forced unmount */
296 #define SHUTDOWN_CORRUPT_INCORE 0x0008 /* corrupt in-memory data structures */
297 #define SHUTDOWN_REMOTE_REQ 0x0010 /* shutdown came from remote cell */
298 #define SHUTDOWN_DEVICE_REQ 0x0020 /* failed all paths to the device */
301 * Flags for xfs_mountfs
303 #define XFS_MFSI_QUIET 0x40 /* Be silent if mount errors found */
305 static inline xfs_agnumber_t
306 xfs_daddr_to_agno(struct xfs_mount *mp, xfs_daddr_t d)
308 xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
309 do_div(ld, mp->m_sb.sb_agblocks);
310 return (xfs_agnumber_t) ld;
313 static inline xfs_agblock_t
314 xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
316 xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
317 return (xfs_agblock_t) do_div(ld, mp->m_sb.sb_agblocks);
322 xfs_mp_fail_writes(struct xfs_mount *mp)
324 return mp->m_fail_writes;
328 xfs_mp_fail_writes(struct xfs_mount *mp)
334 /* per-AG block reservation data structures*/
335 enum xfs_ag_resv_type {
336 XFS_AG_RESV_NONE = 0,
337 XFS_AG_RESV_METADATA,
342 /* number of blocks originally reserved here */
343 xfs_extlen_t ar_orig_reserved;
344 /* number of blocks reserved here */
345 xfs_extlen_t ar_reserved;
346 /* number of blocks originally asked for */
347 xfs_extlen_t ar_asked;
351 * Per-ag incore structure, copies of information in agf and agi, to improve the
352 * performance of allocation group selection.
354 typedef struct xfs_perag {
355 struct xfs_mount *pag_mount; /* owner filesystem */
356 xfs_agnumber_t pag_agno; /* AG this structure belongs to */
357 atomic_t pag_ref; /* perag reference count */
358 char pagf_init; /* this agf's entry is initialized */
359 char pagi_init; /* this agi's entry is initialized */
360 char pagf_metadata; /* the agf is preferred to be metadata */
361 char pagi_inodeok; /* The agi is ok for inodes */
362 __uint8_t pagf_levels[XFS_BTNUM_AGF];
363 /* # of levels in bno & cnt btree */
364 __uint32_t pagf_flcount; /* count of blocks in freelist */
365 xfs_extlen_t pagf_freeblks; /* total free blocks */
366 xfs_extlen_t pagf_longest; /* longest free space */
367 __uint32_t pagf_btreeblks; /* # of blocks held in AGF btrees */
368 xfs_agino_t pagi_freecount; /* number of free inodes */
369 xfs_agino_t pagi_count; /* number of allocated inodes */
372 * Inode allocation search lookup optimisation.
373 * If the pagino matches, the search for new inodes
374 * doesn't need to search the near ones again straight away
376 xfs_agino_t pagl_pagino;
377 xfs_agino_t pagl_leftrec;
378 xfs_agino_t pagl_rightrec;
379 spinlock_t pagb_lock; /* lock for pagb_tree */
380 struct rb_root pagb_tree; /* ordered tree of busy extents */
382 atomic_t pagf_fstrms; /* # of filestreams active in this AG */
384 spinlock_t pag_ici_lock; /* incore inode cache lock */
385 struct radix_tree_root pag_ici_root; /* incore inode cache root */
386 int pag_ici_reclaimable; /* reclaimable inodes */
387 struct mutex pag_ici_reclaim_lock; /* serialisation point */
388 unsigned long pag_ici_reclaim_cursor; /* reclaim restart point */
390 /* buffer cache index */
391 spinlock_t pag_buf_lock; /* lock for pag_buf_tree */
392 struct rb_root pag_buf_tree; /* ordered tree of active buffers */
394 /* for rcu-safe freeing */
395 struct rcu_head rcu_head;
396 int pagb_count; /* pagb slots in use */
398 /* Blocks reserved for all kinds of metadata. */
399 struct xfs_ag_resv pag_meta_resv;
400 /* Blocks reserved for just AGFL-based metadata. */
401 struct xfs_ag_resv pag_agfl_resv;
404 static inline struct xfs_ag_resv *
406 struct xfs_perag *pag,
407 enum xfs_ag_resv_type type)
410 case XFS_AG_RESV_METADATA:
411 return &pag->pag_meta_resv;
412 case XFS_AG_RESV_AGFL:
413 return &pag->pag_agfl_resv;
419 extern void xfs_uuid_table_free(void);
420 extern int xfs_log_sbcount(xfs_mount_t *);
421 extern __uint64_t xfs_default_resblks(xfs_mount_t *mp);
422 extern int xfs_mountfs(xfs_mount_t *mp);
423 extern int xfs_initialize_perag(xfs_mount_t *mp, xfs_agnumber_t agcount,
424 xfs_agnumber_t *maxagi);
425 extern void xfs_unmountfs(xfs_mount_t *);
427 extern int xfs_mod_icount(struct xfs_mount *mp, int64_t delta);
428 extern int xfs_mod_ifree(struct xfs_mount *mp, int64_t delta);
429 extern int xfs_mod_fdblocks(struct xfs_mount *mp, int64_t delta,
431 extern int xfs_mod_frextents(struct xfs_mount *mp, int64_t delta);
433 extern struct xfs_buf *xfs_getsb(xfs_mount_t *, int);
434 extern int xfs_readsb(xfs_mount_t *, int);
435 extern void xfs_freesb(xfs_mount_t *);
436 extern bool xfs_fs_writable(struct xfs_mount *mp, int level);
437 extern int xfs_sb_validate_fsb_count(struct xfs_sb *, __uint64_t);
439 extern int xfs_dev_is_read_only(struct xfs_mount *, char *);
441 extern void xfs_set_low_space_thresholds(struct xfs_mount *);
443 int xfs_zero_extent(struct xfs_inode *ip, xfs_fsblock_t start_fsb,
444 xfs_off_t count_fsb);
446 struct xfs_error_cfg * xfs_error_get_cfg(struct xfs_mount *mp,
447 int error_class, int error);
449 #endif /* __XFS_MOUNT_H__ */