fs/xfs/xfs_mount.h

   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 /*
  41  * Error Configuration
  42  *
  43  * Error classes define the subsystem the configuration belongs to.
  44  * Error numbers define the errors that are configurable.
  45  */
  46 enum {
  47         XFS_ERR_METADATA,
  48         XFS_ERR_CLASS_MAX,
  49 };
  50 enum {
  51         XFS_ERR_DEFAULT,
  52         XFS_ERR_EIO,
  53         XFS_ERR_ENOSPC,
  54         XFS_ERR_ENODEV,
  55         XFS_ERR_ERRNO_MAX,
  56 };
  57
  58 #define XFS_ERR_RETRY_FOREVER   -1
  59
  60 /*
  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.
  65  */
  66 struct xfs_error_cfg {
  67         struct xfs_kobj kobj;
  68         int             max_retries;
  69         long            retry_timeout;  /* in jiffies, -1 = infinite */
  70 };
  71
  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 */
  76
  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 */
  82
  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_refc_mxr[2];  /* max refc btree records */
 128         uint                    m_refc_mnr[2];  /* min refc btree records */
 129         uint                    m_ag_maxlevels; /* XFS_AG_MAXLEVELS */
 130         uint                    m_bm_maxlevels[2]; /* XFS_BM_MAXLEVELS */
 131         uint                    m_in_maxlevels; /* max inobt btree levels. */
 132         uint                    m_rmap_maxlevels; /* max rmap btree levels */
 133         uint                    m_refc_maxlevels; /* max refcount btree level */
 134         xfs_extlen_t            m_ag_prealloc_blocks; /* reserved ag blocks */
 135         uint                    m_alloc_set_aside; /* space we can't use */
 136         uint                    m_ag_max_usable; /* max space per AG */
 137         struct radix_tree_root  m_perag_tree;   /* per-ag accounting info */
 138         spinlock_t              m_perag_lock;   /* lock for m_perag_tree */
 139         struct mutex            m_growlock;     /* growfs mutex */
 140         int                     m_fixedfsid[2]; /* unchanged for life of FS */
 141         uint                    m_dmevmask;     /* DMI events for this FS */
 142         __uint64_t              m_flags;        /* global mount flags */
 143         bool                    m_inotbt_nores; /* no per-AG finobt resv. */
 144         int                     m_ialloc_inos;  /* inodes in inode allocation */
 145         int                     m_ialloc_blks;  /* blocks in inode allocation */
 146         int                     m_ialloc_min_blks;/* min blocks in sparse inode
 147                                                    * allocation */
 148         int                     m_inoalign_mask;/* mask sb_inoalignmt if used */
 149         uint                    m_qflags;       /* quota status flags */
 150         struct xfs_trans_resv   m_resv;         /* precomputed res values */
 151         __uint64_t              m_maxicount;    /* maximum inode count */
 152         __uint64_t              m_resblks;      /* total reserved blocks */
 153         __uint64_t              m_resblks_avail;/* available reserved blocks */
 154         __uint64_t              m_resblks_save; /* reserved blks @ remount,ro */
 155         int                     m_dalign;       /* stripe unit */
 156         int                     m_swidth;       /* stripe width */
 157         int                     m_sinoalign;    /* stripe unit inode alignment */
 158         __uint8_t               m_sectbb_log;   /* sectlog - BBSHIFT */
 159         const struct xfs_nameops *m_dirnameops; /* vector of dir name ops */
 160         const struct xfs_dir_ops *m_dir_inode_ops; /* vector of dir inode ops */
 161         const struct xfs_dir_ops *m_nondir_inode_ops; /* !dir inode ops */
 162         uint                    m_chsize;       /* size of next field */
 163         atomic_t                m_active_trans; /* number trans frozen */
 164         struct xfs_mru_cache    *m_filestream;  /* per-mount filestream data */
 165         struct delayed_work     m_reclaim_work; /* background inode reclaim */
 166         struct delayed_work     m_eofblocks_work; /* background eof blocks
 167                                                      trimming */
 168         struct delayed_work     m_cowblocks_work; /* background cow blocks
 169                                                      trimming */
 170         bool                    m_update_sb;    /* sb needs update in mount */
 171         int64_t                 m_low_space[XFS_LOWSP_MAX];
 172                                                 /* low free space thresholds */
 173         struct xfs_kobj         m_kobj;
 174         struct xfs_kobj         m_error_kobj;
 175         struct xfs_kobj         m_error_meta_kobj;
 176         struct xfs_error_cfg    m_error_cfg[XFS_ERR_CLASS_MAX][XFS_ERR_ERRNO_MAX];
 177         struct xstats           m_stats;        /* per-fs stats */
 178
 179         struct workqueue_struct *m_buf_workqueue;
 180         struct workqueue_struct *m_data_workqueue;
 181         struct workqueue_struct *m_unwritten_workqueue;
 182         struct workqueue_struct *m_cil_workqueue;
 183         struct workqueue_struct *m_reclaim_workqueue;
 184         struct workqueue_struct *m_log_workqueue;
 185         struct workqueue_struct *m_eofblocks_workqueue;
 186         struct workqueue_struct *m_sync_workqueue;
 187
 188         /*
 189          * Generation of the filesysyem layout.  This is incremented by each
 190          * growfs, and used by the pNFS server to ensure the client updates
 191          * its view of the block device once it gets a layout that might
 192          * reference the newly added blocks.  Does not need to be persistent
 193          * as long as we only allow file system size increments, but if we
 194          * ever support shrinks it would have to be persisted in addition
 195          * to various other kinds of pain inflicted on the pNFS server.
 196          */
 197         __uint32_t              m_generation;
 198
 199         bool                    m_fail_unmount;
 200 #ifdef DEBUG
 201         /*
 202          * DEBUG mode instrumentation to test and/or trigger delayed allocation
 203          * block killing in the event of failed writes. When enabled, all
 204          * buffered writes are forced to fail. All delalloc blocks in the range
 205          * of the write (including pre-existing delalloc blocks!) are tossed as
 206          * part of the write failure error handling sequence.
 207          */
 208         bool                    m_fail_writes;
 209 #endif
 210 } xfs_mount_t;
 211
 212 /*
 213  * Flags for m_flags.
 214  */
 215 #define XFS_MOUNT_WSYNC         (1ULL << 0)     /* for nfs - all metadata ops
 216                                                    must be synchronous except
 217                                                    for space allocations */
 218 #define XFS_MOUNT_UNMOUNTING    (1ULL << 1)     /* filesystem is unmounting */
 219 #define XFS_MOUNT_WAS_CLEAN     (1ULL << 3)
 220 #define XFS_MOUNT_FS_SHUTDOWN   (1ULL << 4)     /* atomic stop of all filesystem
 221                                                    operations, typically for
 222                                                    disk errors in metadata */
 223 #define XFS_MOUNT_DISCARD       (1ULL << 5)     /* discard unused blocks */
 224 #define XFS_MOUNT_NOALIGN       (1ULL << 7)     /* turn off stripe alignment
 225                                                    allocations */
 226 #define XFS_MOUNT_ATTR2         (1ULL << 8)     /* allow use of attr2 format */
 227 #define XFS_MOUNT_GRPID         (1ULL << 9)     /* group-ID assigned from directory */
 228 #define XFS_MOUNT_NORECOVERY    (1ULL << 10)    /* no recovery - dirty fs */
 229 #define XFS_MOUNT_DFLT_IOSIZE   (1ULL << 12)    /* set default i/o size */
 230 #define XFS_MOUNT_SMALL_INUMS   (1ULL << 14)    /* user wants 32bit inodes */
 231 #define XFS_MOUNT_32BITINODES   (1ULL << 15)    /* inode32 allocator active */
 232 #define XFS_MOUNT_NOUUID        (1ULL << 16)    /* ignore uuid during mount */
 233 #define XFS_MOUNT_BARRIER       (1ULL << 17)
 234 #define XFS_MOUNT_IKEEP         (1ULL << 18)    /* keep empty inode clusters*/
 235 #define XFS_MOUNT_SWALLOC       (1ULL << 19)    /* turn on stripe width
 236                                                  * allocation */
 237 #define XFS_MOUNT_RDONLY        (1ULL << 20)    /* read-only fs */
 238 #define XFS_MOUNT_DIRSYNC       (1ULL << 21)    /* synchronous directory ops */
 239 #define XFS_MOUNT_COMPAT_IOSIZE (1ULL << 22)    /* don't report large preferred
 240                                                  * I/O size in stat() */
 241 #define XFS_MOUNT_FILESTREAMS   (1ULL << 24)    /* enable the filestreams
 242                                                    allocator */
 243 #define XFS_MOUNT_NOATTR2       (1ULL << 25)    /* disable use of attr2 format */
 244
 245 #define XFS_MOUNT_DAX           (1ULL << 62)    /* TEST ONLY! */
 246
 247
 248 /*
 249  * Default minimum read and write sizes.
 250  */
 251 #define XFS_READIO_LOG_LARGE    16
 252 #define XFS_WRITEIO_LOG_LARGE   16
 253
 254 /*
 255  * Max and min values for mount-option defined I/O
 256  * preallocation sizes.
 257  */
 258 #define XFS_MAX_IO_LOG          30      /* 1G */
 259 #define XFS_MIN_IO_LOG          PAGE_SHIFT
 260
 261 /*
 262  * Synchronous read and write sizes.  This should be
 263  * better for NFSv2 wsync filesystems.
 264  */
 265 #define XFS_WSYNC_READIO_LOG    15      /* 32k */
 266 #define XFS_WSYNC_WRITEIO_LOG   14      /* 16k */
 267
 268 /*
 269  * Allow large block sizes to be reported to userspace programs if the
 270  * "largeio" mount option is used.
 271  *
 272  * If compatibility mode is specified, simply return the basic unit of caching
 273  * so that we don't get inefficient read/modify/write I/O from user apps.
 274  * Otherwise....
 275  *
 276  * If the underlying volume is a stripe, then return the stripe width in bytes
 277  * as the recommended I/O size. It is not a stripe and we've set a default
 278  * buffered I/O size, return that, otherwise return the compat default.
 279  */
 280 static inline unsigned long
 281 xfs_preferred_iosize(xfs_mount_t *mp)
 282 {
 283         if (mp->m_flags & XFS_MOUNT_COMPAT_IOSIZE)
 284                 return PAGE_SIZE;
 285         return (mp->m_swidth ?
 286                 (mp->m_swidth << mp->m_sb.sb_blocklog) :
 287                 ((mp->m_flags & XFS_MOUNT_DFLT_IOSIZE) ?
 288                         (1 << (int)MAX(mp->m_readio_log, mp->m_writeio_log)) :
 289                         PAGE_SIZE));
 290 }
 291
 292 #define XFS_LAST_UNMOUNT_WAS_CLEAN(mp)  \
 293                                 ((mp)->m_flags & XFS_MOUNT_WAS_CLEAN)
 294 #define XFS_FORCED_SHUTDOWN(mp) ((mp)->m_flags & XFS_MOUNT_FS_SHUTDOWN)
 295 void xfs_do_force_shutdown(struct xfs_mount *mp, int flags, char *fname,
 296                 int lnnum);
 297 #define xfs_force_shutdown(m,f) \
 298         xfs_do_force_shutdown(m, f, __FILE__, __LINE__)
 299
 300 #define SHUTDOWN_META_IO_ERROR  0x0001  /* write attempt to metadata failed */
 301 #define SHUTDOWN_LOG_IO_ERROR   0x0002  /* write attempt to the log failed */
 302 #define SHUTDOWN_FORCE_UMOUNT   0x0004  /* shutdown from a forced unmount */
 303 #define SHUTDOWN_CORRUPT_INCORE 0x0008  /* corrupt in-memory data structures */
 304 #define SHUTDOWN_REMOTE_REQ     0x0010  /* shutdown came from remote cell */
 305 #define SHUTDOWN_DEVICE_REQ     0x0020  /* failed all paths to the device */
 306
 307 /*
 308  * Flags for xfs_mountfs
 309  */
 310 #define XFS_MFSI_QUIET          0x40    /* Be silent if mount errors found */
 311
 312 static inline xfs_agnumber_t
 313 xfs_daddr_to_agno(struct xfs_mount *mp, xfs_daddr_t d)
 314 {
 315         xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
 316         do_div(ld, mp->m_sb.sb_agblocks);
 317         return (xfs_agnumber_t) ld;
 318 }
 319
 320 static inline xfs_agblock_t
 321 xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
 322 {
 323         xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
 324         return (xfs_agblock_t) do_div(ld, mp->m_sb.sb_agblocks);
 325 }
 326
 327 #ifdef DEBUG
 328 static inline bool
 329 xfs_mp_fail_writes(struct xfs_mount *mp)
 330 {
 331         return mp->m_fail_writes;
 332 }
 333 #else
 334 static inline bool
 335 xfs_mp_fail_writes(struct xfs_mount *mp)
 336 {
 337         return 0;
 338 }
 339 #endif
 340
 341 /* per-AG block reservation data structures*/
 342 enum xfs_ag_resv_type {
 343         XFS_AG_RESV_NONE = 0,
 344         XFS_AG_RESV_METADATA,
 345         XFS_AG_RESV_AGFL,
 346 };
 347
 348 struct xfs_ag_resv {
 349         /* number of blocks originally reserved here */
 350         xfs_extlen_t                    ar_orig_reserved;
 351         /* number of blocks reserved here */
 352         xfs_extlen_t                    ar_reserved;
 353         /* number of blocks originally asked for */
 354         xfs_extlen_t                    ar_asked;
 355 };
 356
 357 /*
 358  * Per-ag incore structure, copies of information in agf and agi, to improve the
 359  * performance of allocation group selection.
 360  */
 361 typedef struct xfs_perag {
 362         struct xfs_mount *pag_mount;    /* owner filesystem */
 363         xfs_agnumber_t  pag_agno;       /* AG this structure belongs to */
 364         atomic_t        pag_ref;        /* perag reference count */
 365         char            pagf_init;      /* this agf's entry is initialized */
 366         char            pagi_init;      /* this agi's entry is initialized */
 367         char            pagf_metadata;  /* the agf is preferred to be metadata */
 368         char            pagi_inodeok;   /* The agi is ok for inodes */
 369         __uint8_t       pagf_levels[XFS_BTNUM_AGF];
 370                                         /* # of levels in bno & cnt btree */
 371         bool            pagf_agflreset; /* agfl requires reset before use */
 372         __uint32_t      pagf_flcount;   /* count of blocks in freelist */
 373         xfs_extlen_t    pagf_freeblks;  /* total free blocks */
 374         xfs_extlen_t    pagf_longest;   /* longest free space */
 375         __uint32_t      pagf_btreeblks; /* # of blocks held in AGF btrees */
 376         xfs_agino_t     pagi_freecount; /* number of free inodes */
 377         xfs_agino_t     pagi_count;     /* number of allocated inodes */
 378
 379         /*
 380          * Inode allocation search lookup optimisation.
 381          * If the pagino matches, the search for new inodes
 382          * doesn't need to search the near ones again straight away
 383          */
 384         xfs_agino_t     pagl_pagino;
 385         xfs_agino_t     pagl_leftrec;
 386         xfs_agino_t     pagl_rightrec;
 387         spinlock_t      pagb_lock;      /* lock for pagb_tree */
 388         struct rb_root  pagb_tree;      /* ordered tree of busy extents */
 389
 390         atomic_t        pagf_fstrms;    /* # of filestreams active in this AG */
 391
 392         spinlock_t      pag_ici_lock;   /* incore inode cache lock */
 393         struct radix_tree_root pag_ici_root;    /* incore inode cache root */
 394         int             pag_ici_reclaimable;    /* reclaimable inodes */
 395         struct mutex    pag_ici_reclaim_lock;   /* serialisation point */
 396         unsigned long   pag_ici_reclaim_cursor; /* reclaim restart point */
 397
 398         /* buffer cache index */
 399         spinlock_t      pag_buf_lock;   /* lock for pag_buf_tree */
 400         struct rb_root  pag_buf_tree;   /* ordered tree of active buffers */
 401
 402         /* for rcu-safe freeing */
 403         struct rcu_head rcu_head;
 404         int             pagb_count;     /* pagb slots in use */
 405
 406         /* Blocks reserved for all kinds of metadata. */
 407         struct xfs_ag_resv      pag_meta_resv;
 408         /* Blocks reserved for just AGFL-based metadata. */
 409         struct xfs_ag_resv      pag_agfl_resv;
 410
 411         /* reference count */
 412         __uint8_t               pagf_refcount_level;
 413 } xfs_perag_t;
 414
 415 static inline struct xfs_ag_resv *
 416 xfs_perag_resv(
 417         struct xfs_perag        *pag,
 418         enum xfs_ag_resv_type   type)
 419 {
 420         switch (type) {
 421         case XFS_AG_RESV_METADATA:
 422                 return &pag->pag_meta_resv;
 423         case XFS_AG_RESV_AGFL:
 424                 return &pag->pag_agfl_resv;
 425         default:
 426                 return NULL;
 427         }
 428 }
 429
 430 extern void     xfs_uuid_table_free(void);
 431 extern int      xfs_log_sbcount(xfs_mount_t *);
 432 extern __uint64_t xfs_default_resblks(xfs_mount_t *mp);
 433 extern int      xfs_mountfs(xfs_mount_t *mp);
 434 extern int      xfs_initialize_perag(xfs_mount_t *mp, xfs_agnumber_t agcount,
 435                                      xfs_agnumber_t *maxagi);
 436 extern void     xfs_unmountfs(xfs_mount_t *);
 437
 438 extern int      xfs_mod_icount(struct xfs_mount *mp, int64_t delta);
 439 extern int      xfs_mod_ifree(struct xfs_mount *mp, int64_t delta);
 440 extern int      xfs_mod_fdblocks(struct xfs_mount *mp, int64_t delta,
 441                                  bool reserved);
 442 extern int      xfs_mod_frextents(struct xfs_mount *mp, int64_t delta);
 443
 444 extern struct xfs_buf *xfs_getsb(xfs_mount_t *, int);
 445 extern int      xfs_readsb(xfs_mount_t *, int);
 446 extern void     xfs_freesb(xfs_mount_t *);
 447 extern bool     xfs_fs_writable(struct xfs_mount *mp, int level);
 448 extern int      xfs_sb_validate_fsb_count(struct xfs_sb *, __uint64_t);
 449
 450 extern int      xfs_dev_is_read_only(struct xfs_mount *, char *);
 451
 452 extern void     xfs_set_low_space_thresholds(struct xfs_mount *);
 453
 454 int     xfs_zero_extent(struct xfs_inode *ip, xfs_fsblock_t start_fsb,
 455                         xfs_off_t count_fsb);
 456
 457 struct xfs_error_cfg * xfs_error_get_cfg(struct xfs_mount *mp,
 458                 int error_class, int error);
 459
 460 #endif  /* __XFS_MOUNT_H__ */