LXR linux/fs/xfs/linux-2.6/xfs

   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_BUF_H__
  19#define __XFS_BUF_H__
  20
  21#include <linux/list.h>
  22#include <linux/types.h>
  23#include <linux/spinlock.h>
  24#include <asm/system.h>
  25#include <linux/mm.h>
  26#include <linux/fs.h>
  27#include <linux/buffer_head.h>
  28#include <linux/uio.h>
  29
  30/*
  31 *      Base types
  32 */
  33
  34#define XFS_BUF_DADDR_NULL      ((xfs_daddr_t) (-1LL))
  35
  36#define xfs_buf_ctob(pp)        ((pp) * PAGE_CACHE_SIZE)
  37#define xfs_buf_btoc(dd)        (((dd) + PAGE_CACHE_SIZE-1) >> PAGE_CACHE_SHIFT)
  38#define xfs_buf_btoct(dd)       ((dd) >> PAGE_CACHE_SHIFT)
  39#define xfs_buf_poff(aa)        ((aa) & ~PAGE_CACHE_MASK)
  40
  41typedef enum {
  42        XBRW_READ = 1,                  /* transfer into target memory */
  43        XBRW_WRITE = 2,                 /* transfer from target memory */
  44        XBRW_ZERO = 3,                  /* Zero target memory */
  45} xfs_buf_rw_t;
  46
  47#define XBF_READ        (1 << 0) /* buffer intended for reading from device */
  48#define XBF_WRITE       (1 << 1) /* buffer intended for writing to device */
  49#define XBF_MAPPED      (1 << 2) /* buffer mapped (b_addr valid) */
  50#define XBF_ASYNC       (1 << 4) /* initiator will not wait for completion */
  51#define XBF_DONE        (1 << 5) /* all pages in the buffer uptodate */
  52#define XBF_DELWRI      (1 << 6) /* buffer has dirty pages */
  53#define XBF_STALE       (1 << 7) /* buffer has been staled, do not find it */
  54#define XBF_ORDERED     (1 << 11)/* use ordered writes */
  55#define XBF_READ_AHEAD  (1 << 12)/* asynchronous read-ahead */
  56#define XBF_LOG_BUFFER  (1 << 13)/* this is a buffer used for the log */
  57
  58/* flags used only as arguments to access routines */
  59#define XBF_LOCK        (1 << 14)/* lock requested */
  60#define XBF_TRYLOCK     (1 << 15)/* lock requested, but do not wait */
  61#define XBF_DONT_BLOCK  (1 << 16)/* do not block in current thread */
  62
  63/* flags used only internally */
  64#define _XBF_PAGE_CACHE (1 << 17)/* backed by pagecache */
  65#define _XBF_PAGES      (1 << 18)/* backed by refcounted pages */
  66#define _XBF_RUN_QUEUES (1 << 19)/* run block device task queue */
  67#define _XBF_DELWRI_Q   (1 << 21)/* buffer on delwri queue */
  68
  69/*
  70 * Special flag for supporting metadata blocks smaller than a FSB.
  71 *
  72 * In this case we can have multiple xfs_buf_t on a single page and
  73 * need to lock out concurrent xfs_buf_t readers as they only
  74 * serialise access to the buffer.
  75 *
  76 * If the FSB size >= PAGE_CACHE_SIZE case, we have no serialisation
  77 * between reads of the page. Hence we can have one thread read the
  78 * page and modify it, but then race with another thread that thinks
  79 * the page is not up-to-date and hence reads it again.
  80 *
  81 * The result is that the first modifcation to the page is lost.
  82 * This sort of AGF/AGI reading race can happen when unlinking inodes
  83 * that require truncation and results in the AGI unlinked list
  84 * modifications being lost.
  85 */
  86#define _XBF_PAGE_LOCKED        (1 << 22)
  87
  88typedef unsigned int xfs_buf_flags_t;
  89
  90#define XFS_BUF_FLAGS \
  91        { XBF_READ,             "READ" }, \
  92        { XBF_WRITE,            "WRITE" }, \
  93        { XBF_MAPPED,           "MAPPED" }, \
  94        { XBF_ASYNC,            "ASYNC" }, \
  95        { XBF_DONE,             "DONE" }, \
  96        { XBF_DELWRI,           "DELWRI" }, \
  97        { XBF_STALE,            "STALE" }, \
  98        { XBF_ORDERED,          "ORDERED" }, \
  99        { XBF_READ_AHEAD,       "READ_AHEAD" }, \
 100        { XBF_LOCK,             "LOCK" },       /* should never be set */\
 101        { XBF_TRYLOCK,          "TRYLOCK" },    /* ditto */\
 102        { XBF_DONT_BLOCK,       "DONT_BLOCK" }, /* ditto */\
 103        { _XBF_PAGE_CACHE,      "PAGE_CACHE" }, \
 104        { _XBF_PAGES,           "PAGES" }, \
 105        { _XBF_RUN_QUEUES,      "RUN_QUEUES" }, \
 106        { _XBF_DELWRI_Q,        "DELWRI_Q" }, \
 107        { _XBF_PAGE_LOCKED,     "PAGE_LOCKED" }
 108
 109
 110typedef enum {
 111        XBT_FORCE_SLEEP = 0,
 112        XBT_FORCE_FLUSH = 1,
 113} xfs_buftarg_flags_t;
 114
 115typedef struct xfs_bufhash {
 116        struct list_head        bh_list;
 117        spinlock_t              bh_lock;
 118} xfs_bufhash_t;
 119
 120typedef struct xfs_buftarg {
 121        dev_t                   bt_dev;
 122        struct block_device     *bt_bdev;
 123        struct address_space    *bt_mapping;
 124        struct xfs_mount        *bt_mount;
 125        unsigned int            bt_bsize;
 126        unsigned int            bt_sshift;
 127        size_t                  bt_smask;
 128
 129        /* per device delwri queue */
 130        struct task_struct      *bt_task;
 131        struct list_head        bt_delwrite_queue;
 132        spinlock_t              bt_delwrite_lock;
 133        unsigned long           bt_flags;
 134
 135        /* LRU control structures */
 136        struct shrinker         bt_shrinker;
 137        struct list_head        bt_lru;
 138        spinlock_t              bt_lru_lock;
 139        unsigned int            bt_lru_nr;
 140} xfs_buftarg_t;
 141
 142/*
 143 *      xfs_buf_t:  Buffer structure for pagecache-based buffers
 144 *
 145 * This buffer structure is used by the pagecache buffer management routines
 146 * to refer to an assembly of pages forming a logical buffer.
 147 *
 148 * The buffer structure is used on a temporary basis only, and discarded when
 149 * released.  The real data storage is recorded in the pagecache. Buffers are
 150 * hashed to the block device on which the file system resides.
 151 */
 152
 153struct xfs_buf;
 154typedef void (*xfs_buf_iodone_t)(struct xfs_buf *);
 155
 156#define XB_PAGES        2
 157
 158typedef struct xfs_buf {
 159        /*
 160         * first cacheline holds all the fields needed for an uncontended cache
 161         * hit to be fully processed. The semaphore straddles the cacheline
 162         * boundary, but the counter and lock sits on the first cacheline,
 163         * which is the only bit that is touched if we hit the semaphore
 164         * fast-path on locking.
 165         */
 166        struct rb_node          b_rbnode;       /* rbtree node */
 167        xfs_off_t               b_file_offset;  /* offset in file */
 168        size_t                  b_buffer_length;/* size of buffer in bytes */
 169        atomic_t                b_hold;         /* reference count */
 170        atomic_t                b_lru_ref;      /* lru reclaim ref count */
 171        xfs_buf_flags_t         b_flags;        /* status flags */
 172        struct semaphore        b_sema;         /* semaphore for lockables */
 173
 174        struct list_head        b_lru;          /* lru list */
 175        wait_queue_head_t       b_waiters;      /* unpin waiters */
 176        struct list_head        b_list;
 177        struct xfs_perag        *b_pag;         /* contains rbtree root */
 178        xfs_buftarg_t           *b_target;      /* buffer target (device) */
 179        xfs_daddr_t             b_bn;           /* block number for I/O */
 180        size_t                  b_count_desired;/* desired transfer size */
 181        void                    *b_addr;        /* virtual address of buffer */
 182        struct work_struct      b_iodone_work;
 183        xfs_buf_iodone_t        b_iodone;       /* I/O completion function */
 184        struct completion       b_iowait;       /* queue for I/O waiters */
 185        void                    *b_fspriv;
 186        void                    *b_fspriv2;
 187        struct page             **b_pages;      /* array of page pointers */
 188        struct page             *b_page_array[XB_PAGES]; /* inline pages */
 189        unsigned long           b_queuetime;    /* time buffer was queued */
 190        atomic_t                b_pin_count;    /* pin count */
 191        atomic_t                b_io_remaining; /* #outstanding I/O requests */
 192        unsigned int            b_page_count;   /* size of page array */
 193        unsigned int            b_offset;       /* page offset in first page */
 194        unsigned short          b_error;        /* error code on I/O */
 195#ifdef XFS_BUF_LOCK_TRACKING
 196        int                     b_last_holder;
 197#endif
 198} xfs_buf_t;
 199
 200
 201/* Finding and Reading Buffers */
 202extern xfs_buf_t *_xfs_buf_find(xfs_buftarg_t *, xfs_off_t, size_t,
 203                                xfs_buf_flags_t, xfs_buf_t *);
 204#define xfs_incore(buftarg,blkno,len,lockit) \
 205        _xfs_buf_find(buftarg, blkno ,len, lockit, NULL)
 206
 207extern xfs_buf_t *xfs_buf_get(xfs_buftarg_t *, xfs_off_t, size_t,
 208                                xfs_buf_flags_t);
 209extern xfs_buf_t *xfs_buf_read(xfs_buftarg_t *, xfs_off_t, size_t,
 210                                xfs_buf_flags_t);
 211
 212extern xfs_buf_t *xfs_buf_get_empty(size_t, xfs_buftarg_t *);
 213extern xfs_buf_t *xfs_buf_get_uncached(struct xfs_buftarg *, size_t, int);
 214extern int xfs_buf_associate_memory(xfs_buf_t *, void *, size_t);
 215extern void xfs_buf_hold(xfs_buf_t *);
 216extern void xfs_buf_readahead(xfs_buftarg_t *, xfs_off_t, size_t);
 217struct xfs_buf *xfs_buf_read_uncached(struct xfs_mount *mp,
 218                                struct xfs_buftarg *target,
 219                                xfs_daddr_t daddr, size_t length, int flags);
 220
 221/* Releasing Buffers */
 222extern void xfs_buf_free(xfs_buf_t *);
 223extern void xfs_buf_rele(xfs_buf_t *);
 224
 225/* Locking and Unlocking Buffers */
 226extern int xfs_buf_cond_lock(xfs_buf_t *);
 227extern int xfs_buf_lock_value(xfs_buf_t *);
 228extern void xfs_buf_lock(xfs_buf_t *);
 229extern void xfs_buf_unlock(xfs_buf_t *);
 230
 231/* Buffer Read and Write Routines */
 232extern int xfs_bwrite(struct xfs_mount *mp, struct xfs_buf *bp);
 233extern void xfs_bdwrite(void *mp, xfs_buf_t *bp);
 234
 235extern void xfsbdstrat(struct xfs_mount *, struct xfs_buf *);
 236extern int xfs_bdstrat_cb(struct xfs_buf *);
 237
 238extern void xfs_buf_ioend(xfs_buf_t *,  int);
 239extern void xfs_buf_ioerror(xfs_buf_t *, int);
 240extern int xfs_buf_iorequest(xfs_buf_t *);
 241extern int xfs_buf_iowait(xfs_buf_t *);
 242extern void xfs_buf_iomove(xfs_buf_t *, size_t, size_t, void *,
 243                                xfs_buf_rw_t);
 244#define xfs_buf_zero(bp, off, len) \
 245            xfs_buf_iomove((bp), (off), (len), NULL, XBRW_ZERO)
 246
 247static inline int xfs_buf_geterror(xfs_buf_t *bp)
 248{
 249        return bp ? bp->b_error : ENOMEM;
 250}
 251
 252/* Buffer Utility Routines */
 253extern xfs_caddr_t xfs_buf_offset(xfs_buf_t *, size_t);
 254
 255/* Delayed Write Buffer Routines */
 256extern void xfs_buf_delwri_dequeue(xfs_buf_t *);
 257extern void xfs_buf_delwri_promote(xfs_buf_t *);
 258
 259/* Buffer Daemon Setup Routines */
 260extern int xfs_buf_init(void);
 261extern void xfs_buf_terminate(void);
 262
 263#define xfs_buf_target_name(target)     \
 264        ({ char __b[BDEVNAME_SIZE]; bdevname((target)->bt_bdev, __b); __b; })
 265
 266
 267#define XFS_BUF_BFLAGS(bp)      ((bp)->b_flags)
 268#define XFS_BUF_ZEROFLAGS(bp)   ((bp)->b_flags &= \
 269                ~(XBF_READ|XBF_WRITE|XBF_ASYNC|XBF_DELWRI|XBF_ORDERED))
 270
 271void xfs_buf_stale(struct xfs_buf *bp);
 272#define XFS_BUF_STALE(bp)       xfs_buf_stale(bp);
 273#define XFS_BUF_UNSTALE(bp)     ((bp)->b_flags &= ~XBF_STALE)
 274#define XFS_BUF_ISSTALE(bp)     ((bp)->b_flags & XBF_STALE)
 275#define XFS_BUF_SUPER_STALE(bp) do {                            \
 276                                        XFS_BUF_STALE(bp);      \
 277                                        xfs_buf_delwri_dequeue(bp);     \
 278                                        XFS_BUF_DONE(bp);       \
 279                                } while (0)
 280
 281#define XFS_BUF_DELAYWRITE(bp)          ((bp)->b_flags |= XBF_DELWRI)
 282#define XFS_BUF_UNDELAYWRITE(bp)        xfs_buf_delwri_dequeue(bp)
 283#define XFS_BUF_ISDELAYWRITE(bp)        ((bp)->b_flags & XBF_DELWRI)
 284
 285#define XFS_BUF_ERROR(bp,no)    xfs_buf_ioerror(bp,no)
 286#define XFS_BUF_GETERROR(bp)    xfs_buf_geterror(bp)
 287#define XFS_BUF_ISERROR(bp)     (xfs_buf_geterror(bp) ? 1 : 0)
 288
 289#define XFS_BUF_DONE(bp)        ((bp)->b_flags |= XBF_DONE)
 290#define XFS_BUF_UNDONE(bp)      ((bp)->b_flags &= ~XBF_DONE)
 291#define XFS_BUF_ISDONE(bp)      ((bp)->b_flags & XBF_DONE)
 292
 293#define XFS_BUF_BUSY(bp)        do { } while (0)
 294#define XFS_BUF_UNBUSY(bp)      do { } while (0)
 295#define XFS_BUF_ISBUSY(bp)      (1)
 296
 297#define XFS_BUF_ASYNC(bp)       ((bp)->b_flags |= XBF_ASYNC)
 298#define XFS_BUF_UNASYNC(bp)     ((bp)->b_flags &= ~XBF_ASYNC)
 299#define XFS_BUF_ISASYNC(bp)     ((bp)->b_flags & XBF_ASYNC)
 300
 301#define XFS_BUF_ORDERED(bp)     ((bp)->b_flags |= XBF_ORDERED)
 302#define XFS_BUF_UNORDERED(bp)   ((bp)->b_flags &= ~XBF_ORDERED)
 303#define XFS_BUF_ISORDERED(bp)   ((bp)->b_flags & XBF_ORDERED)
 304
 305#define XFS_BUF_HOLD(bp)        xfs_buf_hold(bp)
 306#define XFS_BUF_READ(bp)        ((bp)->b_flags |= XBF_READ)
 307#define XFS_BUF_UNREAD(bp)      ((bp)->b_flags &= ~XBF_READ)
 308#define XFS_BUF_ISREAD(bp)      ((bp)->b_flags & XBF_READ)
 309
 310#define XFS_BUF_WRITE(bp)       ((bp)->b_flags |= XBF_WRITE)
 311#define XFS_BUF_UNWRITE(bp)     ((bp)->b_flags &= ~XBF_WRITE)
 312#define XFS_BUF_ISWRITE(bp)     ((bp)->b_flags & XBF_WRITE)
 313
 314#define XFS_BUF_IODONE_FUNC(bp)                 ((bp)->b_iodone)
 315#define XFS_BUF_SET_IODONE_FUNC(bp, func)       ((bp)->b_iodone = (func))
 316#define XFS_BUF_CLR_IODONE_FUNC(bp)             ((bp)->b_iodone = NULL)
 317
 318#define XFS_BUF_FSPRIVATE(bp, type)             ((type)(bp)->b_fspriv)
 319#define XFS_BUF_SET_FSPRIVATE(bp, val)          ((bp)->b_fspriv = (void*)(val))
 320#define XFS_BUF_FSPRIVATE2(bp, type)            ((type)(bp)->b_fspriv2)
 321#define XFS_BUF_SET_FSPRIVATE2(bp, val)         ((bp)->b_fspriv2 = (void*)(val))
 322#define XFS_BUF_SET_START(bp)                   do { } while (0)
 323
 324#define XFS_BUF_PTR(bp)                 (xfs_caddr_t)((bp)->b_addr)
 325#define XFS_BUF_SET_PTR(bp, val, cnt)   xfs_buf_associate_memory(bp, val, cnt)
 326#define XFS_BUF_ADDR(bp)                ((bp)->b_bn)
 327#define XFS_BUF_SET_ADDR(bp, bno)       ((bp)->b_bn = (xfs_daddr_t)(bno))
 328#define XFS_BUF_OFFSET(bp)              ((bp)->b_file_offset)
 329#define XFS_BUF_SET_OFFSET(bp, off)     ((bp)->b_file_offset = (off))
 330#define XFS_BUF_COUNT(bp)               ((bp)->b_count_desired)
 331#define XFS_BUF_SET_COUNT(bp, cnt)      ((bp)->b_count_desired = (cnt))
 332#define XFS_BUF_SIZE(bp)                ((bp)->b_buffer_length)
 333#define XFS_BUF_SET_SIZE(bp, cnt)       ((bp)->b_buffer_length = (cnt))
 334
 335static inline void
 336xfs_buf_set_ref(
 337        struct xfs_buf  *bp,
 338        int             lru_ref)
 339{
 340        atomic_set(&bp->b_lru_ref, lru_ref);
 341}
 342#define XFS_BUF_SET_VTYPE_REF(bp, type, ref)    xfs_buf_set_ref(bp, ref)
 343#define XFS_BUF_SET_VTYPE(bp, type)             do { } while (0)
 344
 345#define XFS_BUF_ISPINNED(bp)    atomic_read(&((bp)->b_pin_count))
 346
 347#define XFS_BUF_VALUSEMA(bp)    xfs_buf_lock_value(bp)
 348#define XFS_BUF_CPSEMA(bp)      (xfs_buf_cond_lock(bp) == 0)
 349#define XFS_BUF_VSEMA(bp)       xfs_buf_unlock(bp)
 350#define XFS_BUF_PSEMA(bp,x)     xfs_buf_lock(bp)
 351#define XFS_BUF_FINISH_IOWAIT(bp)       complete(&bp->b_iowait);
 352
 353#define XFS_BUF_SET_TARGET(bp, target)  ((bp)->b_target = (target))
 354#define XFS_BUF_TARGET(bp)              ((bp)->b_target)
 355#define XFS_BUFTARG_NAME(target)        xfs_buf_target_name(target)
 356
 357static inline void xfs_buf_relse(xfs_buf_t *bp)
 358{
 359        xfs_buf_unlock(bp);
 360        xfs_buf_rele(bp);
 361}
 362
 363/*
 364 *      Handling of buftargs.
 365 */
 366extern xfs_buftarg_t *xfs_alloc_buftarg(struct xfs_mount *,
 367                        struct block_device *, int, const char *);
 368extern void xfs_free_buftarg(struct xfs_mount *, struct xfs_buftarg *);
 369extern void xfs_wait_buftarg(xfs_buftarg_t *);
 370extern int xfs_setsize_buftarg(xfs_buftarg_t *, unsigned int, unsigned int);
 371extern int xfs_flush_buftarg(xfs_buftarg_t *, int);
 372
 373#ifdef CONFIG_KDB_MODULES
 374extern struct list_head *xfs_get_buftarg_list(void);
 375#endif
 376
 377#define xfs_getsize_buftarg(buftarg)    block_size((buftarg)->bt_bdev)
 378#define xfs_readonly_buftarg(buftarg)   bdev_read_only((buftarg)->bt_bdev)
 379
 380#define xfs_binval(buftarg)             xfs_flush_buftarg(buftarg, 1)
 381#define XFS_bflush(buftarg)             xfs_flush_buftarg(buftarg, 1)
 382
 383#endif  /* __XFS_BUF_H__ */
 384