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 struct xfs_error_cfg {
  61         struct xfs_kobj kobj;
  62         int             max_retries;
  63         unsigned long   retry_timeout;  /* in jiffies, 0 = no timeout */
  64 };
  65
  66 typedef struct xfs_mount {
  67         struct super_block      *m_super;
  68         xfs_tid_t               m_tid;          /* next unused tid for fs */
  69         struct xfs_ail          *m_ail;         /* fs active log item list */
  70
  71         struct xfs_sb           m_sb;           /* copy of fs superblock */
  72         spinlock_t              m_sb_lock;      /* sb counter lock */
  73         struct percpu_counter   m_icount;       /* allocated inodes counter */
  74         struct percpu_counter   m_ifree;        /* free inodes counter */
  75         struct percpu_counter   m_fdblocks;     /* free block counter */
  76
  77         struct xfs_buf          *m_sb_bp;       /* buffer for superblock */
  78         char                    *m_fsname;      /* filesystem name */
  79         int                     m_fsname_len;   /* strlen of fs name */
  80         char                    *m_rtname;      /* realtime device name */
  81         char                    *m_logname;     /* external log device name */
  82         int                     m_bsize;        /* fs logical block size */
  83         xfs_agnumber_t          m_agfrotor;     /* last ag where space found */
  84         xfs_agnumber_t          m_agirotor;     /* last ag dir inode alloced */
  85         spinlock_t              m_agirotor_lock;/* .. and lock protecting it */
  86         xfs_agnumber_t          m_maxagi;       /* highest inode alloc group */
  87         uint                    m_readio_log;   /* min read size log bytes */
  88         uint                    m_readio_blocks; /* min read size blocks */
  89         uint                    m_writeio_log;  /* min write size log bytes */
  90         uint                    m_writeio_blocks; /* min write size blocks */
  91         struct xfs_da_geometry  *m_dir_geo;     /* directory block geometry */
  92         struct xfs_da_geometry  *m_attr_geo;    /* attribute block geometry */
  93         struct xlog             *m_log;         /* log specific stuff */
  94         int                     m_logbufs;      /* number of log buffers */
  95         int                     m_logbsize;     /* size of each log buffer */
  96         uint                    m_rsumlevels;   /* rt summary levels */
  97         uint                    m_rsumsize;     /* size of rt summary, bytes */
  98         struct xfs_inode        *m_rbmip;       /* pointer to bitmap inode */
  99         struct xfs_inode        *m_rsumip;      /* pointer to summary inode */
 100         struct xfs_inode        *m_rootip;      /* pointer to root directory */
 101         struct xfs_quotainfo    *m_quotainfo;   /* disk quota information */
 102         xfs_buftarg_t           *m_ddev_targp;  /* saves taking the address */
 103         xfs_buftarg_t           *m_logdev_targp;/* ptr to log device */
 104         xfs_buftarg_t           *m_rtdev_targp; /* ptr to rt device */
 105         __uint8_t               m_blkbit_log;   /* blocklog + NBBY */
 106         __uint8_t               m_blkbb_log;    /* blocklog - BBSHIFT */
 107         __uint8_t               m_agno_log;     /* log #ag's */
 108         __uint8_t               m_agino_log;    /* #bits for agino in inum */
 109         uint                    m_inode_cluster_size;/* min inode buf size */
 110         uint                    m_blockmask;    /* sb_blocksize-1 */
 111         uint                    m_blockwsize;   /* sb_blocksize in words */
 112         uint                    m_blockwmask;   /* blockwsize-1 */
 113         uint                    m_alloc_mxr[2]; /* max alloc btree records */
 114         uint                    m_alloc_mnr[2]; /* min alloc btree records */
 115         uint                    m_bmap_dmxr[2]; /* max bmap btree records */
 116         uint                    m_bmap_dmnr[2]; /* min bmap btree records */
 117         uint                    m_inobt_mxr[2]; /* max inobt btree records */
 118         uint                    m_inobt_mnr[2]; /* min inobt btree records */
 119         uint                    m_ag_maxlevels; /* XFS_AG_MAXLEVELS */
 120         uint                    m_bm_maxlevels[2]; /* XFS_BM_MAXLEVELS */
 121         uint                    m_in_maxlevels; /* max inobt btree levels. */
 122         struct radix_tree_root  m_perag_tree;   /* per-ag accounting info */
 123         spinlock_t              m_perag_lock;   /* lock for m_perag_tree */
 124         struct mutex            m_growlock;     /* growfs mutex */
 125         int                     m_fixedfsid[2]; /* unchanged for life of FS */
 126         uint                    m_dmevmask;     /* DMI events for this FS */
 127         __uint64_t              m_flags;        /* global mount flags */
 128         int                     m_ialloc_inos;  /* inodes in inode allocation */
 129         int                     m_ialloc_blks;  /* blocks in inode allocation */
 130         int                     m_ialloc_min_blks;/* min blocks in sparse inode
 131                                                    * allocation */
 132         int                     m_inoalign_mask;/* mask sb_inoalignmt if used */
 133         uint                    m_qflags;       /* quota status flags */
 134         struct xfs_trans_resv   m_resv;         /* precomputed res values */
 135         __uint64_t              m_maxicount;    /* maximum inode count */
 136         __uint64_t              m_resblks;      /* total reserved blocks */
 137         __uint64_t              m_resblks_avail;/* available reserved blocks */
 138         __uint64_t              m_resblks_save; /* reserved blks @ remount,ro */
 139         int                     m_dalign;       /* stripe unit */
 140         int                     m_swidth;       /* stripe width */
 141         int                     m_sinoalign;    /* stripe unit inode alignment */
 142         __uint8_t               m_sectbb_log;   /* sectlog - BBSHIFT */
 143         const struct xfs_nameops *m_dirnameops; /* vector of dir name ops */
 144         const struct xfs_dir_ops *m_dir_inode_ops; /* vector of dir inode ops */
 145         const struct xfs_dir_ops *m_nondir_inode_ops; /* !dir inode ops */
 146         uint                    m_chsize;       /* size of next field */
 147         atomic_t                m_active_trans; /* number trans frozen */
 148         struct xfs_mru_cache    *m_filestream;  /* per-mount filestream data */
 149         struct delayed_work     m_reclaim_work; /* background inode reclaim */
 150         struct delayed_work     m_eofblocks_work; /* background eof blocks
 151                                                      trimming */
 152         bool                    m_update_sb;    /* sb needs update in mount */
 153         int64_t                 m_low_space[XFS_LOWSP_MAX];
 154                                                 /* low free space thresholds */
 155         struct xfs_kobj         m_kobj;
 156         struct xfs_kobj         m_error_kobj;
 157         struct xfs_kobj         m_error_meta_kobj;
 158         struct xfs_error_cfg    m_error_cfg[XFS_ERR_CLASS_MAX][XFS_ERR_ERRNO_MAX];
 159         struct xstats           m_stats;        /* per-fs stats */
 160
 161         struct workqueue_struct *m_buf_workqueue;
 162         struct workqueue_struct *m_data_workqueue;
 163         struct workqueue_struct *m_unwritten_workqueue;
 164         struct workqueue_struct *m_cil_workqueue;
 165         struct workqueue_struct *m_reclaim_workqueue;
 166         struct workqueue_struct *m_log_workqueue;
 167         struct workqueue_struct *m_eofblocks_workqueue;
 168
 169         /*
 170          * Generation of the filesysyem layout.  This is incremented by each
 171          * growfs, and used by the pNFS server to ensure the client updates
 172          * its view of the block device once it gets a layout that might
 173          * reference the newly added blocks.  Does not need to be persistent
 174          * as long as we only allow file system size increments, but if we
 175          * ever support shrinks it would have to be persisted in addition
 176          * to various other kinds of pain inflicted on the pNFS server.
 177          */
 178         __uint32_t              m_generation;
 179
 180 #ifdef DEBUG
 181         /*
 182          * DEBUG mode instrumentation to test and/or trigger delayed allocation
 183          * block killing in the event of failed writes. When enabled, all
 184          * buffered writes are forced to fail. All delalloc blocks in the range
 185          * of the write (including pre-existing delalloc blocks!) are tossed as
 186          * part of the write failure error handling sequence.
 187          */
 188         bool                    m_fail_writes;
 189 #endif
 190 } xfs_mount_t;
 191
 192 /*
 193  * Flags for m_flags.
 194  */
 195 #define XFS_MOUNT_WSYNC         (1ULL << 0)     /* for nfs - all metadata ops
 196                                                    must be synchronous except
 197                                                    for space allocations */
 198 #define XFS_MOUNT_WAS_CLEAN     (1ULL << 3)
 199 #define XFS_MOUNT_FS_SHUTDOWN   (1ULL << 4)     /* atomic stop of all filesystem
 200                                                    operations, typically for
 201                                                    disk errors in metadata */
 202 #define XFS_MOUNT_DISCARD       (1ULL << 5)     /* discard unused blocks */
 203 #define XFS_MOUNT_NOALIGN       (1ULL << 7)     /* turn off stripe alignment
 204                                                    allocations */
 205 #define XFS_MOUNT_ATTR2         (1ULL << 8)     /* allow use of attr2 format */
 206 #define XFS_MOUNT_GRPID         (1ULL << 9)     /* group-ID assigned from directory */
 207 #define XFS_MOUNT_NORECOVERY    (1ULL << 10)    /* no recovery - dirty fs */
 208 #define XFS_MOUNT_DFLT_IOSIZE   (1ULL << 12)    /* set default i/o size */
 209 #define XFS_MOUNT_SMALL_INUMS   (1ULL << 14)    /* user wants 32bit inodes */
 210 #define XFS_MOUNT_32BITINODES   (1ULL << 15)    /* inode32 allocator active */
 211 #define XFS_MOUNT_NOUUID        (1ULL << 16)    /* ignore uuid during mount */
 212 #define XFS_MOUNT_BARRIER       (1ULL << 17)
 213 #define XFS_MOUNT_IKEEP         (1ULL << 18)    /* keep empty inode clusters*/
 214 #define XFS_MOUNT_SWALLOC       (1ULL << 19)    /* turn on stripe width
 215                                                  * allocation */
 216 #define XFS_MOUNT_RDONLY        (1ULL << 20)    /* read-only fs */
 217 #define XFS_MOUNT_DIRSYNC       (1ULL << 21)    /* synchronous directory ops */
 218 #define XFS_MOUNT_COMPAT_IOSIZE (1ULL << 22)    /* don't report large preferred
 219                                                  * I/O size in stat() */
 220 #define XFS_MOUNT_FILESTREAMS   (1ULL << 24)    /* enable the filestreams
 221                                                    allocator */
 222 #define XFS_MOUNT_NOATTR2       (1ULL << 25)    /* disable use of attr2 format */
 223
 224 #define XFS_MOUNT_DAX           (1ULL << 62)    /* TEST ONLY! */
 225
 226
 227 /*
 228  * Default minimum read and write sizes.
 229  */
 230 #define XFS_READIO_LOG_LARGE    16
 231 #define XFS_WRITEIO_LOG_LARGE   16
 232
 233 /*
 234  * Max and min values for mount-option defined I/O
 235  * preallocation sizes.
 236  */
 237 #define XFS_MAX_IO_LOG          30      /* 1G */
 238 #define XFS_MIN_IO_LOG          PAGE_SHIFT
 239
 240 /*
 241  * Synchronous read and write sizes.  This should be
 242  * better for NFSv2 wsync filesystems.
 243  */
 244 #define XFS_WSYNC_READIO_LOG    15      /* 32k */
 245 #define XFS_WSYNC_WRITEIO_LOG   14      /* 16k */
 246
 247 /*
 248  * Allow large block sizes to be reported to userspace programs if the
 249  * "largeio" mount option is used.
 250  *
 251  * If compatibility mode is specified, simply return the basic unit of caching
 252  * so that we don't get inefficient read/modify/write I/O from user apps.
 253  * Otherwise....
 254  *
 255  * If the underlying volume is a stripe, then return the stripe width in bytes
 256  * as the recommended I/O size. It is not a stripe and we've set a default
 257  * buffered I/O size, return that, otherwise return the compat default.
 258  */
 259 static inline unsigned long
 260 xfs_preferred_iosize(xfs_mount_t *mp)
 261 {
 262         if (mp->m_flags & XFS_MOUNT_COMPAT_IOSIZE)
 263                 return PAGE_CACHE_SIZE;
 264         return (mp->m_swidth ?
 265                 (mp->m_swidth << mp->m_sb.sb_blocklog) :
 266                 ((mp->m_flags & XFS_MOUNT_DFLT_IOSIZE) ?
 267                         (1 << (int)MAX(mp->m_readio_log, mp->m_writeio_log)) :
 268                         PAGE_CACHE_SIZE));
 269 }
 270
 271 #define XFS_LAST_UNMOUNT_WAS_CLEAN(mp)  \
 272                                 ((mp)->m_flags & XFS_MOUNT_WAS_CLEAN)
 273 #define XFS_FORCED_SHUTDOWN(mp) ((mp)->m_flags & XFS_MOUNT_FS_SHUTDOWN)
 274 void xfs_do_force_shutdown(struct xfs_mount *mp, int flags, char *fname,
 275                 int lnnum);
 276 #define xfs_force_shutdown(m,f) \
 277         xfs_do_force_shutdown(m, f, __FILE__, __LINE__)
 278
 279 #define SHUTDOWN_META_IO_ERROR  0x0001  /* write attempt to metadata failed */
 280 #define SHUTDOWN_LOG_IO_ERROR   0x0002  /* write attempt to the log failed */
 281 #define SHUTDOWN_FORCE_UMOUNT   0x0004  /* shutdown from a forced unmount */
 282 #define SHUTDOWN_CORRUPT_INCORE 0x0008  /* corrupt in-memory data structures */
 283 #define SHUTDOWN_REMOTE_REQ     0x0010  /* shutdown came from remote cell */
 284 #define SHUTDOWN_DEVICE_REQ     0x0020  /* failed all paths to the device */
 285
 286 /*
 287  * Flags for xfs_mountfs
 288  */
 289 #define XFS_MFSI_QUIET          0x40    /* Be silent if mount errors found */
 290
 291 static inline xfs_agnumber_t
 292 xfs_daddr_to_agno(struct xfs_mount *mp, xfs_daddr_t d)
 293 {
 294         xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
 295         do_div(ld, mp->m_sb.sb_agblocks);
 296         return (xfs_agnumber_t) ld;
 297 }
 298
 299 static inline xfs_agblock_t
 300 xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
 301 {
 302         xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
 303         return (xfs_agblock_t) do_div(ld, mp->m_sb.sb_agblocks);
 304 }
 305
 306 #ifdef DEBUG
 307 static inline bool
 308 xfs_mp_fail_writes(struct xfs_mount *mp)
 309 {
 310         return mp->m_fail_writes;
 311 }
 312 #else
 313 static inline bool
 314 xfs_mp_fail_writes(struct xfs_mount *mp)
 315 {
 316         return 0;
 317 }
 318 #endif
 319
 320 /*
 321  * Per-ag incore structure, copies of information in agf and agi, to improve the
 322  * performance of allocation group selection.
 323  */
 324 typedef struct xfs_perag {
 325         struct xfs_mount *pag_mount;    /* owner filesystem */
 326         xfs_agnumber_t  pag_agno;       /* AG this structure belongs to */
 327         atomic_t        pag_ref;        /* perag reference count */
 328         char            pagf_init;      /* this agf's entry is initialized */
 329         char            pagi_init;      /* this agi's entry is initialized */
 330         char            pagf_metadata;  /* the agf is preferred to be metadata */
 331         char            pagi_inodeok;   /* The agi is ok for inodes */
 332         __uint8_t       pagf_levels[XFS_BTNUM_AGF];
 333                                         /* # of levels in bno & cnt btree */
 334         __uint32_t      pagf_flcount;   /* count of blocks in freelist */
 335         xfs_extlen_t    pagf_freeblks;  /* total free blocks */
 336         xfs_extlen_t    pagf_longest;   /* longest free space */
 337         __uint32_t      pagf_btreeblks; /* # of blocks held in AGF btrees */
 338         xfs_agino_t     pagi_freecount; /* number of free inodes */
 339         xfs_agino_t     pagi_count;     /* number of allocated inodes */
 340
 341         /*
 342          * Inode allocation search lookup optimisation.
 343          * If the pagino matches, the search for new inodes
 344          * doesn't need to search the near ones again straight away
 345          */
 346         xfs_agino_t     pagl_pagino;
 347         xfs_agino_t     pagl_leftrec;
 348         xfs_agino_t     pagl_rightrec;
 349         spinlock_t      pagb_lock;      /* lock for pagb_tree */
 350         struct rb_root  pagb_tree;      /* ordered tree of busy extents */
 351
 352         atomic_t        pagf_fstrms;    /* # of filestreams active in this AG */
 353
 354         spinlock_t      pag_ici_lock;   /* incore inode cache lock */
 355         struct radix_tree_root pag_ici_root;    /* incore inode cache root */
 356         int             pag_ici_reclaimable;    /* reclaimable inodes */
 357         struct mutex    pag_ici_reclaim_lock;   /* serialisation point */
 358         unsigned long   pag_ici_reclaim_cursor; /* reclaim restart point */
 359
 360         /* buffer cache index */
 361         spinlock_t      pag_buf_lock;   /* lock for pag_buf_tree */
 362         struct rb_root  pag_buf_tree;   /* ordered tree of active buffers */
 363
 364         /* for rcu-safe freeing */
 365         struct rcu_head rcu_head;
 366         int             pagb_count;     /* pagb slots in use */
 367 } xfs_perag_t;
 368
 369 extern void     xfs_uuid_table_free(void);
 370 extern int      xfs_log_sbcount(xfs_mount_t *);
 371 extern __uint64_t xfs_default_resblks(xfs_mount_t *mp);
 372 extern int      xfs_mountfs(xfs_mount_t *mp);
 373 extern int      xfs_initialize_perag(xfs_mount_t *mp, xfs_agnumber_t agcount,
 374                                      xfs_agnumber_t *maxagi);
 375 extern void     xfs_unmountfs(xfs_mount_t *);
 376
 377 extern int      xfs_mod_icount(struct xfs_mount *mp, int64_t delta);
 378 extern int      xfs_mod_ifree(struct xfs_mount *mp, int64_t delta);
 379 extern int      xfs_mod_fdblocks(struct xfs_mount *mp, int64_t delta,
 380                                  bool reserved);
 381 extern int      xfs_mod_frextents(struct xfs_mount *mp, int64_t delta);
 382
 383 extern struct xfs_buf *xfs_getsb(xfs_mount_t *, int);
 384 extern int      xfs_readsb(xfs_mount_t *, int);
 385 extern void     xfs_freesb(xfs_mount_t *);
 386 extern bool     xfs_fs_writable(struct xfs_mount *mp, int level);
 387 extern int      xfs_sb_validate_fsb_count(struct xfs_sb *, __uint64_t);
 388
 389 extern int      xfs_dev_is_read_only(struct xfs_mount *, char *);
 390
 391 extern void     xfs_set_low_space_thresholds(struct xfs_mount *);
 392
 393 int     xfs_zero_extent(struct xfs_inode *ip, xfs_fsblock_t start_fsb,
 394                         xfs_off_t count_fsb);
 395
 396 struct xfs_error_cfg * xfs_error_get_cfg(struct xfs_mount *mp,
 397                 int error_class, int error);
 398
 399 #endif  /* __XFS_MOUNT_H__ */