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 typedef struct xfs_trans_reservations {
22 	uint	tr_write;	/* extent alloc trans */
23 	uint	tr_itruncate;	/* truncate trans */
24 	uint	tr_rename;	/* rename trans */
25 	uint	tr_link;	/* link trans */
26 	uint	tr_remove;	/* unlink trans */
27 	uint	tr_symlink;	/* symlink trans */
28 	uint	tr_create;	/* create trans */
29 	uint	tr_mkdir;	/* mkdir trans */
30 	uint	tr_ifree;	/* inode free trans */
31 	uint	tr_ichange;	/* inode update trans */
32 	uint	tr_growdata;	/* fs data section grow trans */
33 	uint	tr_swrite;	/* sync write inode trans */
34 	uint	tr_addafork;	/* cvt inode to attributed trans */
35 	uint	tr_writeid;	/* write setuid/setgid file */
36 	uint	tr_attrinval;	/* attr fork buffer invalidation */
37 	uint	tr_attrset;	/* set/create an attribute */
38 	uint	tr_attrrm;	/* remove an attribute */
39 	uint	tr_clearagi;	/* clear bad agi unlinked ino bucket */
40 	uint	tr_growrtalloc;	/* grow realtime allocations */
41 	uint	tr_growrtzero;	/* grow realtime zeroing */
42 	uint	tr_growrtfree;	/* grow realtime freeing */
43 } xfs_trans_reservations_t;
44 
45 #ifndef __KERNEL__
46 
47 #define xfs_daddr_to_agno(mp,d) \
48 	((xfs_agnumber_t)(XFS_BB_TO_FSBT(mp, d) / (mp)->m_sb.sb_agblocks))
49 #define xfs_daddr_to_agbno(mp,d) \
50 	((xfs_agblock_t)(XFS_BB_TO_FSBT(mp, d) % (mp)->m_sb.sb_agblocks))
51 
52 #else /* __KERNEL__ */
53 
54 #include "xfs_sync.h"
55 
56 struct log;
57 struct xfs_mount_args;
58 struct xfs_inode;
59 struct xfs_bmbt_irec;
60 struct xfs_bmap_free;
61 struct xfs_extdelta;
62 struct xfs_swapext;
63 struct xfs_mru_cache;
64 struct xfs_nameops;
65 struct xfs_ail;
66 struct xfs_quotainfo;
67 
68 #ifdef HAVE_PERCPU_SB
69 
70 /*
71  * Valid per-cpu incore superblock counters. Note that if you add new counters,
72  * you may need to define new counter disabled bit field descriptors as there
73  * are more possible fields in the superblock that can fit in a bitfield on a
74  * 32 bit platform. The XFS_SBS_* values for the current current counters just
75  * fit.
76  */
77 typedef struct xfs_icsb_cnts {
78 	uint64_t	icsb_fdblocks;
79 	uint64_t	icsb_ifree;
80 	uint64_t	icsb_icount;
81 	unsigned long	icsb_flags;
82 } xfs_icsb_cnts_t;
83 
84 #define XFS_ICSB_FLAG_LOCK	(1 << 0)	/* counter lock bit */
85 
86 #define XFS_ICSB_LAZY_COUNT	(1 << 1)	/* accuracy not needed */
87 
88 extern int	xfs_icsb_init_counters(struct xfs_mount *);
89 extern void	xfs_icsb_reinit_counters(struct xfs_mount *);
90 extern void	xfs_icsb_destroy_counters(struct xfs_mount *);
91 extern void	xfs_icsb_sync_counters(struct xfs_mount *, int);
92 extern void	xfs_icsb_sync_counters_locked(struct xfs_mount *, int);
93 extern int	xfs_icsb_modify_counters(struct xfs_mount *, xfs_sb_field_t,
94 						int64_t, int);
95 
96 #else
97 #define xfs_icsb_init_counters(mp)		(0)
98 #define xfs_icsb_destroy_counters(mp)		do { } while (0)
99 #define xfs_icsb_reinit_counters(mp)		do { } while (0)
100 #define xfs_icsb_sync_counters(mp, flags)	do { } while (0)
101 #define xfs_icsb_sync_counters_locked(mp, flags) do { } while (0)
102 #define xfs_icsb_modify_counters(mp, field, delta, rsvd) \
103 	xfs_mod_incore_sb(mp, field, delta, rsvd)
104 #endif
105 
106 /* dynamic preallocation free space thresholds, 5% down to 1% */
107 enum {
108 	XFS_LOWSP_1_PCNT = 0,
109 	XFS_LOWSP_2_PCNT,
110 	XFS_LOWSP_3_PCNT,
111 	XFS_LOWSP_4_PCNT,
112 	XFS_LOWSP_5_PCNT,
113 	XFS_LOWSP_MAX,
114 };
115 
116 typedef struct xfs_mount {
117 	struct super_block	*m_super;
118 	xfs_tid_t		m_tid;		/* next unused tid for fs */
119 	struct xfs_ail		*m_ail;		/* fs active log item list */
120 	xfs_sb_t		m_sb;		/* copy of fs superblock */
121 	spinlock_t		m_sb_lock;	/* sb counter lock */
122 	struct xfs_buf		*m_sb_bp;	/* buffer for superblock */
123 	char			*m_fsname;	/* filesystem name */
124 	int			m_fsname_len;	/* strlen of fs name */
125 	char			*m_rtname;	/* realtime device name */
126 	char			*m_logname;	/* external log device name */
127 	int			m_bsize;	/* fs logical block size */
128 	xfs_agnumber_t		m_agfrotor;	/* last ag where space found */
129 	xfs_agnumber_t		m_agirotor;	/* last ag dir inode alloced */
130 	spinlock_t		m_agirotor_lock;/* .. and lock protecting it */
131 	xfs_agnumber_t		m_maxagi;	/* highest inode alloc group */
132 	uint			m_readio_log;	/* min read size log bytes */
133 	uint			m_readio_blocks; /* min read size blocks */
134 	uint			m_writeio_log;	/* min write size log bytes */
135 	uint			m_writeio_blocks; /* min write size blocks */
136 	struct log		*m_log;		/* log specific stuff */
137 	int			m_logbufs;	/* number of log buffers */
138 	int			m_logbsize;	/* size of each log buffer */
139 	uint			m_rsumlevels;	/* rt summary levels */
140 	uint			m_rsumsize;	/* size of rt summary, bytes */
141 	struct xfs_inode	*m_rbmip;	/* pointer to bitmap inode */
142 	struct xfs_inode	*m_rsumip;	/* pointer to summary inode */
143 	struct xfs_inode	*m_rootip;	/* pointer to root directory */
144 	struct xfs_quotainfo	*m_quotainfo;	/* disk quota information */
145 	xfs_buftarg_t		*m_ddev_targp;	/* saves taking the address */
146 	xfs_buftarg_t		*m_logdev_targp;/* ptr to log device */
147 	xfs_buftarg_t		*m_rtdev_targp;	/* ptr to rt device */
148 	__uint8_t		m_blkbit_log;	/* blocklog + NBBY */
149 	__uint8_t		m_blkbb_log;	/* blocklog - BBSHIFT */
150 	__uint8_t		m_agno_log;	/* log #ag's */
151 	__uint8_t		m_agino_log;	/* #bits for agino in inum */
152 	__uint16_t		m_inode_cluster_size;/* min inode buf size */
153 	uint			m_blockmask;	/* sb_blocksize-1 */
154 	uint			m_blockwsize;	/* sb_blocksize in words */
155 	uint			m_blockwmask;	/* blockwsize-1 */
156 	uint			m_alloc_mxr[2];	/* max alloc btree records */
157 	uint			m_alloc_mnr[2];	/* min alloc btree records */
158 	uint			m_bmap_dmxr[2];	/* max bmap btree records */
159 	uint			m_bmap_dmnr[2];	/* min bmap btree records */
160 	uint			m_inobt_mxr[2];	/* max inobt btree records */
161 	uint			m_inobt_mnr[2];	/* min inobt btree records */
162 	uint			m_ag_maxlevels;	/* XFS_AG_MAXLEVELS */
163 	uint			m_bm_maxlevels[2]; /* XFS_BM_MAXLEVELS */
164 	uint			m_in_maxlevels;	/* max inobt btree levels. */
165 	struct radix_tree_root	m_perag_tree;	/* per-ag accounting info */
166 	spinlock_t		m_perag_lock;	/* lock for m_perag_tree */
167 	struct mutex		m_growlock;	/* growfs mutex */
168 	int			m_fixedfsid[2];	/* unchanged for life of FS */
169 	uint			m_dmevmask;	/* DMI events for this FS */
170 	__uint64_t		m_flags;	/* global mount flags */
171 	uint			m_dir_node_ents; /* #entries in a dir danode */
172 	uint			m_attr_node_ents; /* #entries in attr danode */
173 	int			m_ialloc_inos;	/* inodes in inode allocation */
174 	int			m_ialloc_blks;	/* blocks in inode allocation */
175 	int			m_inoalign_mask;/* mask sb_inoalignmt if used */
176 	uint			m_qflags;	/* quota status flags */
177 	xfs_trans_reservations_t m_reservations;/* precomputed res values */
178 	__uint64_t		m_maxicount;	/* maximum inode count */
179 	__uint64_t		m_maxioffset;	/* maximum inode offset */
180 	__uint64_t		m_resblks;	/* total reserved blocks */
181 	__uint64_t		m_resblks_avail;/* available reserved blocks */
182 	__uint64_t		m_resblks_save;	/* reserved blks @ remount,ro */
183 	int			m_dalign;	/* stripe unit */
184 	int			m_swidth;	/* stripe width */
185 	int			m_sinoalign;	/* stripe unit inode alignment */
186 	int			m_attr_magicpct;/* 37% of the blocksize */
187 	int			m_dir_magicpct;	/* 37% of the dir blocksize */
188 	__uint8_t		m_sectbb_log;	/* sectlog - BBSHIFT */
189 	const struct xfs_nameops *m_dirnameops;	/* vector of dir name ops */
190 	int			m_dirblksize;	/* directory block sz--bytes */
191 	int			m_dirblkfsbs;	/* directory block sz--fsbs */
192 	xfs_dablk_t		m_dirdatablk;	/* blockno of dir data v2 */
193 	xfs_dablk_t		m_dirleafblk;	/* blockno of dir non-data v2 */
194 	xfs_dablk_t		m_dirfreeblk;	/* blockno of dirfreeindex v2 */
195 	uint			m_chsize;	/* size of next field */
196 	struct xfs_chash	*m_chash;	/* fs private inode per-cluster
197 						 * hash table */
198 	atomic_t		m_active_trans;	/* number trans frozen */
199 #ifdef HAVE_PERCPU_SB
200 	xfs_icsb_cnts_t __percpu *m_sb_cnts;	/* per-cpu superblock counters */
201 	unsigned long		m_icsb_counters; /* disabled per-cpu counters */
202 	struct notifier_block	m_icsb_notifier; /* hotplug cpu notifier */
203 	struct mutex		m_icsb_mutex;	/* balancer sync lock */
204 #endif
205 	struct xfs_mru_cache	*m_filestream;  /* per-mount filestream data */
206 	struct delayed_work	m_sync_work;	/* background sync work */
207 	struct delayed_work	m_reclaim_work;	/* background inode reclaim */
208 	struct work_struct	m_flush_work;	/* background inode flush */
209 	__int64_t		m_update_flags;	/* sb flags we need to update
210 						   on the next remount,rw */
211 	struct shrinker		m_inode_shrink;	/* inode reclaim shrinker */
212 	int64_t			m_low_space[XFS_LOWSP_MAX];
213 						/* low free space thresholds */
214 
215 	struct workqueue_struct	*m_data_workqueue;
216 	struct workqueue_struct	*m_unwritten_workqueue;
217 } xfs_mount_t;
218 
219 /*
220  * Flags for m_flags.
221  */
222 #define XFS_MOUNT_WSYNC		(1ULL << 0)	/* for nfs - all metadata ops
223 						   must be synchronous except
224 						   for space allocations */
225 #define XFS_MOUNT_WAS_CLEAN	(1ULL << 3)
226 #define XFS_MOUNT_FS_SHUTDOWN	(1ULL << 4)	/* atomic stop of all filesystem
227 						   operations, typically for
228 						   disk errors in metadata */
229 #define XFS_MOUNT_DISCARD	(1ULL << 5)	/* discard unused blocks */
230 #define XFS_MOUNT_RETERR	(1ULL << 6)     /* return alignment errors to
231 						   user */
232 #define XFS_MOUNT_NOALIGN	(1ULL << 7)	/* turn off stripe alignment
233 						   allocations */
234 #define XFS_MOUNT_ATTR2		(1ULL << 8)	/* allow use of attr2 format */
235 #define XFS_MOUNT_GRPID		(1ULL << 9)	/* group-ID assigned from directory */
236 #define XFS_MOUNT_NORECOVERY	(1ULL << 10)	/* no recovery - dirty fs */
237 #define XFS_MOUNT_DFLT_IOSIZE	(1ULL << 12)	/* set default i/o size */
238 #define XFS_MOUNT_32BITINODES	(1ULL << 14)	/* do not create inodes above
239 						 * 32 bits in size */
240 #define XFS_MOUNT_SMALL_INUMS	(1ULL << 15)	/* users wants 32bit inodes */
241 #define XFS_MOUNT_NOUUID	(1ULL << 16)	/* ignore uuid during mount */
242 #define XFS_MOUNT_BARRIER	(1ULL << 17)
243 #define XFS_MOUNT_IKEEP		(1ULL << 18)	/* keep empty inode clusters*/
244 #define XFS_MOUNT_SWALLOC	(1ULL << 19)	/* turn on stripe width
245 						 * allocation */
246 #define XFS_MOUNT_RDONLY	(1ULL << 20)	/* read-only fs */
247 #define XFS_MOUNT_DIRSYNC	(1ULL << 21)	/* synchronous directory ops */
248 #define XFS_MOUNT_COMPAT_IOSIZE	(1ULL << 22)	/* don't report large preferred
249 						 * I/O size in stat() */
250 #define XFS_MOUNT_FILESTREAMS	(1ULL << 24)	/* enable the filestreams
251 						   allocator */
252 #define XFS_MOUNT_NOATTR2	(1ULL << 25)	/* disable use of attr2 format */
253 
254 
255 /*
256  * Default minimum read and write sizes.
257  */
258 #define XFS_READIO_LOG_LARGE	16
259 #define XFS_WRITEIO_LOG_LARGE	16
260 
261 /*
262  * Max and min values for mount-option defined I/O
263  * preallocation sizes.
264  */
265 #define XFS_MAX_IO_LOG		30	/* 1G */
266 #define XFS_MIN_IO_LOG		PAGE_SHIFT
267 
268 /*
269  * Synchronous read and write sizes.  This should be
270  * better for NFSv2 wsync filesystems.
271  */
272 #define	XFS_WSYNC_READIO_LOG	15	/* 32k */
273 #define	XFS_WSYNC_WRITEIO_LOG	14	/* 16k */
274 
275 /*
276  * Allow large block sizes to be reported to userspace programs if the
277  * "largeio" mount option is used.
278  *
279  * If compatibility mode is specified, simply return the basic unit of caching
280  * so that we don't get inefficient read/modify/write I/O from user apps.
281  * Otherwise....
282  *
283  * If the underlying volume is a stripe, then return the stripe width in bytes
284  * as the recommended I/O size. It is not a stripe and we've set a default
285  * buffered I/O size, return that, otherwise return the compat default.
286  */
287 static inline unsigned long
xfs_preferred_iosize(xfs_mount_t * mp)288 xfs_preferred_iosize(xfs_mount_t *mp)
289 {
290 	if (mp->m_flags & XFS_MOUNT_COMPAT_IOSIZE)
291 		return PAGE_CACHE_SIZE;
292 	return (mp->m_swidth ?
293 		(mp->m_swidth << mp->m_sb.sb_blocklog) :
294 		((mp->m_flags & XFS_MOUNT_DFLT_IOSIZE) ?
295 			(1 << (int)MAX(mp->m_readio_log, mp->m_writeio_log)) :
296 			PAGE_CACHE_SIZE));
297 }
298 
299 #define XFS_MAXIOFFSET(mp)	((mp)->m_maxioffset)
300 
301 #define XFS_LAST_UNMOUNT_WAS_CLEAN(mp)	\
302 				((mp)->m_flags & XFS_MOUNT_WAS_CLEAN)
303 #define XFS_FORCED_SHUTDOWN(mp)	((mp)->m_flags & XFS_MOUNT_FS_SHUTDOWN)
304 void xfs_do_force_shutdown(struct xfs_mount *mp, int flags, char *fname,
305 		int lnnum);
306 #define xfs_force_shutdown(m,f)	\
307 	xfs_do_force_shutdown(m, f, __FILE__, __LINE__)
308 
309 #define SHUTDOWN_META_IO_ERROR	0x0001	/* write attempt to metadata failed */
310 #define SHUTDOWN_LOG_IO_ERROR	0x0002	/* write attempt to the log failed */
311 #define SHUTDOWN_FORCE_UMOUNT	0x0004	/* shutdown from a forced unmount */
312 #define SHUTDOWN_CORRUPT_INCORE	0x0008	/* corrupt in-memory data structures */
313 #define SHUTDOWN_REMOTE_REQ	0x0010	/* shutdown came from remote cell */
314 #define SHUTDOWN_DEVICE_REQ	0x0020	/* failed all paths to the device */
315 
316 #define xfs_test_for_freeze(mp)		((mp)->m_super->s_frozen)
317 #define xfs_wait_for_freeze(mp,l)	vfs_check_frozen((mp)->m_super, (l))
318 
319 /*
320  * Flags for xfs_mountfs
321  */
322 #define XFS_MFSI_QUIET		0x40	/* Be silent if mount errors found */
323 
324 static inline xfs_agnumber_t
xfs_daddr_to_agno(struct xfs_mount * mp,xfs_daddr_t d)325 xfs_daddr_to_agno(struct xfs_mount *mp, xfs_daddr_t d)
326 {
327 	xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
328 	do_div(ld, mp->m_sb.sb_agblocks);
329 	return (xfs_agnumber_t) ld;
330 }
331 
332 static inline xfs_agblock_t
xfs_daddr_to_agbno(struct xfs_mount * mp,xfs_daddr_t d)333 xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
334 {
335 	xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
336 	return (xfs_agblock_t) do_div(ld, mp->m_sb.sb_agblocks);
337 }
338 
339 /*
340  * perag get/put wrappers for ref counting
341  */
342 struct xfs_perag *xfs_perag_get(struct xfs_mount *mp, xfs_agnumber_t agno);
343 struct xfs_perag *xfs_perag_get_tag(struct xfs_mount *mp, xfs_agnumber_t agno,
344 					int tag);
345 void	xfs_perag_put(struct xfs_perag *pag);
346 
347 /*
348  * Per-cpu superblock locking functions
349  */
350 #ifdef HAVE_PERCPU_SB
351 static inline void
xfs_icsb_lock(xfs_mount_t * mp)352 xfs_icsb_lock(xfs_mount_t *mp)
353 {
354 	mutex_lock(&mp->m_icsb_mutex);
355 }
356 
357 static inline void
xfs_icsb_unlock(xfs_mount_t * mp)358 xfs_icsb_unlock(xfs_mount_t *mp)
359 {
360 	mutex_unlock(&mp->m_icsb_mutex);
361 }
362 #else
363 #define xfs_icsb_lock(mp)
364 #define xfs_icsb_unlock(mp)
365 #endif
366 
367 /*
368  * This structure is for use by the xfs_mod_incore_sb_batch() routine.
369  * xfs_growfs can specify a few fields which are more than int limit
370  */
371 typedef struct xfs_mod_sb {
372 	xfs_sb_field_t	msb_field;	/* Field to modify, see below */
373 	int64_t		msb_delta;	/* Change to make to specified field */
374 } xfs_mod_sb_t;
375 
376 extern int	xfs_log_sbcount(xfs_mount_t *);
377 extern __uint64_t xfs_default_resblks(xfs_mount_t *mp);
378 extern int	xfs_mountfs(xfs_mount_t *mp);
379 
380 extern void	xfs_unmountfs(xfs_mount_t *);
381 extern int	xfs_unmountfs_writesb(xfs_mount_t *);
382 extern int	xfs_mod_incore_sb(xfs_mount_t *, xfs_sb_field_t, int64_t, int);
383 extern int	xfs_mod_incore_sb_batch(xfs_mount_t *, xfs_mod_sb_t *,
384 			uint, int);
385 extern int	xfs_mount_log_sb(xfs_mount_t *, __int64_t);
386 extern struct xfs_buf *xfs_getsb(xfs_mount_t *, int);
387 extern int	xfs_readsb(xfs_mount_t *, int);
388 extern void	xfs_freesb(xfs_mount_t *);
389 extern int	xfs_fs_writable(xfs_mount_t *);
390 extern int	xfs_sb_validate_fsb_count(struct xfs_sb *, __uint64_t);
391 
392 extern int	xfs_dev_is_read_only(struct xfs_mount *, char *);
393 
394 extern void	xfs_set_low_space_thresholds(struct xfs_mount *);
395 
396 #endif	/* __KERNEL__ */
397 
398 extern void	xfs_mod_sb(struct xfs_trans *, __int64_t);
399 extern int	xfs_initialize_perag(struct xfs_mount *, xfs_agnumber_t,
400 					xfs_agnumber_t *);
401 extern void	xfs_sb_from_disk(struct xfs_mount *, struct xfs_dsb *);
402 extern void	xfs_sb_to_disk(struct xfs_dsb *, struct xfs_sb *, __int64_t);
403 
404 #endif	/* __XFS_MOUNT_H__ */
405