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
  47typedef enum {
  48        XBF_READ = (1 << 0),    /* buffer intended for reading from device */
  49        XBF_WRITE = (1 << 1),   /* buffer intended for writing to device   */
  50        XBF_MAPPED = (1 << 2),  /* buffer mapped (b_addr valid)            */
  51        XBF_ASYNC = (1 << 4),   /* initiator will not wait for completion  */
  52        XBF_DONE = (1 << 5),    /* all pages in the buffer uptodate        */
  53        XBF_DELWRI = (1 << 6),  /* buffer has dirty pages                  */
  54        XBF_STALE = (1 << 7),   /* buffer has been staled, do not find it  */
  55        XBF_FS_MANAGED = (1 << 8),  /* filesystem controls freeing memory  */
  56        XBF_ORDERED = (1 << 11),    /* use ordered writes                  */
  57        XBF_READ_AHEAD = (1 << 12), /* asynchronous read-ahead             */
  58
  59        /* flags used only as arguments to access routines */
  60        XBF_LOCK = (1 << 14),       /* lock requested                      */
  61        XBF_TRYLOCK = (1 << 15),    /* lock requested, but do not wait     */
  62        XBF_DONT_BLOCK = (1 << 16), /* do not block in current thread      */
  63
  64        /* flags used only internally */
  65        _XBF_PAGE_CACHE = (1 << 17),/* backed by pagecache                 */
  66        _XBF_PAGES = (1 << 18),     /* backed by refcounted pages          */
  67        _XBF_RUN_QUEUES = (1 << 19),/* run block device task queue         */
  68        _XBF_DELWRI_Q = (1 << 21),   /* buffer on delwri queue             */
  69
  70        /*
  71         * Special flag for supporting metadata blocks smaller than a FSB.
  72         *
  73         * In this case we can have multiple xfs_buf_t on a single page and
  74         * need to lock out concurrent xfs_buf_t readers as they only
  75         * serialise access to the buffer.
  76         *
  77         * If the FSB size >= PAGE_CACHE_SIZE case, we have no serialisation
  78         * between reads of the page. Hence we can have one thread read the
  79         * page and modify it, but then race with another thread that thinks
  80         * the page is not up-to-date and hence reads it again.
  81         *
  82         * The result is that the first modifcation to the page is lost.
  83         * This sort of AGF/AGI reading race can happen when unlinking inodes
  84         * that require truncation and results in the AGI unlinked list
  85         * modifications being lost.
  86         */
  87        _XBF_PAGE_LOCKED = (1 << 22),
  88
  89        /*
  90         * If we try a barrier write, but it fails we have to communicate
  91         * this to the upper layers.  Unfortunately b_error gets overwritten
  92         * when the buffer is re-issued so we have to add another flag to
  93         * keep this information.
  94         */
  95        _XFS_BARRIER_FAILED = (1 << 23),
  96} xfs_buf_flags_t;
  97
  98typedef enum {
  99        XBT_FORCE_SLEEP = 0,
 100        XBT_FORCE_FLUSH = 1,
 101} xfs_buftarg_flags_t;
 102
 103typedef struct xfs_bufhash {
 104        struct list_head        bh_list;
 105        spinlock_t              bh_lock;
 106} xfs_bufhash_t;
 107
 108typedef struct xfs_buftarg {
 109        dev_t                   bt_dev;
 110        struct block_device     *bt_bdev;
 111        struct address_space    *bt_mapping;
 112        unsigned int            bt_bsize;
 113        unsigned int            bt_sshift;
 114        size_t                  bt_smask;
 115
 116        /* per device buffer hash table */
 117        uint                    bt_hashmask;
 118        uint                    bt_hashshift;
 119        xfs_bufhash_t           *bt_hash;
 120
 121        /* per device delwri queue */
 122        struct task_struct      *bt_task;
 123        struct list_head        bt_list;
 124        struct list_head        bt_delwrite_queue;
 125        spinlock_t              bt_delwrite_lock;
 126        unsigned long           bt_flags;
 127} xfs_buftarg_t;
 128
 129/*
 130 *      xfs_buf_t:  Buffer structure for pagecache-based buffers
 131 *
 132 * This buffer structure is used by the pagecache buffer management routines
 133 * to refer to an assembly of pages forming a logical buffer.
 134 *
 135 * The buffer structure is used on a temporary basis only, and discarded when
 136 * released.  The real data storage is recorded in the pagecache. Buffers are
 137 * hashed to the block device on which the file system resides.
 138 */
 139
 140struct xfs_buf;
 141typedef void (*xfs_buf_iodone_t)(struct xfs_buf *);
 142typedef void (*xfs_buf_relse_t)(struct xfs_buf *);
 143typedef int (*xfs_buf_bdstrat_t)(struct xfs_buf *);
 144
 145#define XB_PAGES        2
 146
 147typedef struct xfs_buf {
 148        struct semaphore        b_sema;         /* semaphore for lockables */
 149        unsigned long           b_queuetime;    /* time buffer was queued */
 150        atomic_t                b_pin_count;    /* pin count */
 151        wait_queue_head_t       b_waiters;      /* unpin waiters */
 152        struct list_head        b_list;
 153        xfs_buf_flags_t         b_flags;        /* status flags */
 154        struct list_head        b_hash_list;    /* hash table list */
 155        xfs_bufhash_t           *b_hash;        /* hash table list start */
 156        xfs_buftarg_t           *b_target;      /* buffer target (device) */
 157        atomic_t                b_hold;         /* reference count */
 158        xfs_daddr_t             b_bn;           /* block number for I/O */
 159        xfs_off_t               b_file_offset;  /* offset in file */
 160        size_t                  b_buffer_length;/* size of buffer in bytes */
 161        size_t                  b_count_desired;/* desired transfer size */
 162        void                    *b_addr;        /* virtual address of buffer */
 163        struct work_struct      b_iodone_work;
 164        atomic_t                b_io_remaining; /* #outstanding I/O requests */
 165        xfs_buf_iodone_t        b_iodone;       /* I/O completion function */
 166        xfs_buf_relse_t         b_relse;        /* releasing function */
 167        xfs_buf_bdstrat_t       b_strat;        /* pre-write function */
 168        struct completion       b_iowait;       /* queue for I/O waiters */
 169        void                    *b_fspriv;
 170        void                    *b_fspriv2;
 171        struct xfs_mount        *b_mount;
 172        unsigned short          b_error;        /* error code on I/O */
 173        unsigned int            b_page_count;   /* size of page array */
 174        unsigned int            b_offset;       /* page offset in first page */
 175        struct page             **b_pages;      /* array of page pointers */
 176        struct page             *b_page_array[XB_PAGES]; /* inline pages */
 177#ifdef XFS_BUF_LOCK_TRACKING
 178        int                     b_last_holder;
 179#endif
 180} xfs_buf_t;
 181
 182
 183/* Finding and Reading Buffers */
 184extern xfs_buf_t *_xfs_buf_find(xfs_buftarg_t *, xfs_off_t, size_t,
 185                                xfs_buf_flags_t, xfs_buf_t *);
 186#define xfs_incore(buftarg,blkno,len,lockit) \
 187        _xfs_buf_find(buftarg, blkno ,len, lockit, NULL)
 188
 189extern xfs_buf_t *xfs_buf_get_flags(xfs_buftarg_t *, xfs_off_t, size_t,
 190                                xfs_buf_flags_t);
 191#define xfs_buf_get(target, blkno, len, flags) \
 192        xfs_buf_get_flags((target), (blkno), (len), XBF_LOCK | XBF_MAPPED)
 193
 194extern xfs_buf_t *xfs_buf_read_flags(xfs_buftarg_t *, xfs_off_t, size_t,
 195                                xfs_buf_flags_t);
 196#define xfs_buf_read(target, blkno, len, flags) \
 197        xfs_buf_read_flags((target), (blkno), (len), XBF_LOCK | XBF_MAPPED)
 198
 199extern xfs_buf_t *xfs_buf_get_empty(size_t, xfs_buftarg_t *);
 200extern xfs_buf_t *xfs_buf_get_noaddr(size_t, xfs_buftarg_t *);
 201extern int xfs_buf_associate_memory(xfs_buf_t *, void *, size_t);
 202extern void xfs_buf_hold(xfs_buf_t *);
 203extern void xfs_buf_readahead(xfs_buftarg_t *, xfs_off_t, size_t,
 204                                xfs_buf_flags_t);
 205
 206/* Releasing Buffers */
 207extern void xfs_buf_free(xfs_buf_t *);
 208extern void xfs_buf_rele(xfs_buf_t *);
 209
 210/* Locking and Unlocking Buffers */
 211extern int xfs_buf_cond_lock(xfs_buf_t *);
 212extern int xfs_buf_lock_value(xfs_buf_t *);
 213extern void xfs_buf_lock(xfs_buf_t *);
 214extern void xfs_buf_unlock(xfs_buf_t *);
 215
 216/* Buffer Read and Write Routines */
 217extern int xfs_bawrite(void *mp, xfs_buf_t *bp);
 218extern void xfs_bdwrite(void *mp, xfs_buf_t *bp);
 219extern void xfs_buf_ioend(xfs_buf_t *,  int);
 220extern void xfs_buf_ioerror(xfs_buf_t *, int);
 221extern int xfs_buf_iorequest(xfs_buf_t *);
 222extern int xfs_buf_iowait(xfs_buf_t *);
 223extern void xfs_buf_iomove(xfs_buf_t *, size_t, size_t, xfs_caddr_t,
 224                                xfs_buf_rw_t);
 225
 226static inline int xfs_buf_iostrategy(xfs_buf_t *bp)
 227{
 228        return bp->b_strat ? bp->b_strat(bp) : xfs_buf_iorequest(bp);
 229}
 230
 231static inline int xfs_buf_geterror(xfs_buf_t *bp)
 232{
 233        return bp ? bp->b_error : ENOMEM;
 234}
 235
 236/* Buffer Utility Routines */
 237extern xfs_caddr_t xfs_buf_offset(xfs_buf_t *, size_t);
 238
 239/* Pinning Buffer Storage in Memory */
 240extern void xfs_buf_pin(xfs_buf_t *);
 241extern void xfs_buf_unpin(xfs_buf_t *);
 242extern int xfs_buf_ispin(xfs_buf_t *);
 243
 244/* Delayed Write Buffer Routines */
 245extern void xfs_buf_delwri_dequeue(xfs_buf_t *);
 246
 247/* Buffer Daemon Setup Routines */
 248extern int xfs_buf_init(void);
 249extern void xfs_buf_terminate(void);
 250
 251#ifdef XFS_BUF_TRACE
 252extern ktrace_t *xfs_buf_trace_buf;
 253extern void xfs_buf_trace(xfs_buf_t *, char *, void *, void *);
 254#else
 255#define xfs_buf_trace(bp,id,ptr,ra)     do { } while (0)
 256#endif
 257
 258#define xfs_buf_target_name(target)     \
 259        ({ char __b[BDEVNAME_SIZE]; bdevname((target)->bt_bdev, __b); __b; })
 260
 261
 262#define XFS_B_ASYNC             XBF_ASYNC
 263#define XFS_B_DELWRI            XBF_DELWRI
 264#define XFS_B_READ              XBF_READ
 265#define XFS_B_WRITE             XBF_WRITE
 266#define XFS_B_STALE             XBF_STALE
 267
 268#define XFS_BUF_TRYLOCK         XBF_TRYLOCK
 269#define XFS_INCORE_TRYLOCK      XBF_TRYLOCK
 270#define XFS_BUF_LOCK            XBF_LOCK
 271#define XFS_BUF_MAPPED          XBF_MAPPED
 272
 273#define BUF_BUSY                XBF_DONT_BLOCK
 274
 275#define XFS_BUF_BFLAGS(bp)      ((bp)->b_flags)
 276#define XFS_BUF_ZEROFLAGS(bp)   ((bp)->b_flags &= \
 277                ~(XBF_READ|XBF_WRITE|XBF_ASYNC|XBF_DELWRI|XBF_ORDERED))
 278
 279#define XFS_BUF_STALE(bp)       ((bp)->b_flags |= XFS_B_STALE)
 280#define XFS_BUF_UNSTALE(bp)     ((bp)->b_flags &= ~XFS_B_STALE)
 281#define XFS_BUF_ISSTALE(bp)     ((bp)->b_flags & XFS_B_STALE)
 282#define XFS_BUF_SUPER_STALE(bp) do {                            \
 283                                        XFS_BUF_STALE(bp);      \
 284                                        xfs_buf_delwri_dequeue(bp);     \
 285                                        XFS_BUF_DONE(bp);       \
 286                                } while (0)
 287
 288#define XFS_BUF_MANAGE          XBF_FS_MANAGED
 289#define XFS_BUF_UNMANAGE(bp)    ((bp)->b_flags &= ~XBF_FS_MANAGED)
 290
 291#define XFS_BUF_DELAYWRITE(bp)          ((bp)->b_flags |= XBF_DELWRI)
 292#define XFS_BUF_UNDELAYWRITE(bp)        xfs_buf_delwri_dequeue(bp)
 293#define XFS_BUF_ISDELAYWRITE(bp)        ((bp)->b_flags & XBF_DELWRI)
 294
 295#define XFS_BUF_ERROR(bp,no)    xfs_buf_ioerror(bp,no)
 296#define XFS_BUF_GETERROR(bp)    xfs_buf_geterror(bp)
 297#define XFS_BUF_ISERROR(bp)     (xfs_buf_geterror(bp) ? 1 : 0)
 298
 299#define XFS_BUF_DONE(bp)        ((bp)->b_flags |= XBF_DONE)
 300#define XFS_BUF_UNDONE(bp)      ((bp)->b_flags &= ~XBF_DONE)
 301#define XFS_BUF_ISDONE(bp)      ((bp)->b_flags & XBF_DONE)
 302
 303#define XFS_BUF_BUSY(bp)        do { } while (0)
 304#define XFS_BUF_UNBUSY(bp)      do { } while (0)
 305#define XFS_BUF_ISBUSY(bp)      (1)
 306
 307#define XFS_BUF_ASYNC(bp)       ((bp)->b_flags |= XBF_ASYNC)
 308#define XFS_BUF_UNASYNC(bp)     ((bp)->b_flags &= ~XBF_ASYNC)
 309#define XFS_BUF_ISASYNC(bp)     ((bp)->b_flags & XBF_ASYNC)
 310
 311#define XFS_BUF_ORDERED(bp)     ((bp)->b_flags |= XBF_ORDERED)
 312#define XFS_BUF_UNORDERED(bp)   ((bp)->b_flags &= ~XBF_ORDERED)
 313#define XFS_BUF_ISORDERED(bp)   ((bp)->b_flags & XBF_ORDERED)
 314
 315#define XFS_BUF_HOLD(bp)        xfs_buf_hold(bp)
 316#define XFS_BUF_READ(bp)        ((bp)->b_flags |= XBF_READ)
 317#define XFS_BUF_UNREAD(bp)      ((bp)->b_flags &= ~XBF_READ)
 318#define XFS_BUF_ISREAD(bp)      ((bp)->b_flags & XBF_READ)
 319
 320#define XFS_BUF_WRITE(bp)       ((bp)->b_flags |= XBF_WRITE)
 321#define XFS_BUF_UNWRITE(bp)     ((bp)->b_flags &= ~XBF_WRITE)
 322#define XFS_BUF_ISWRITE(bp)     ((bp)->b_flags & XBF_WRITE)
 323
 324#define XFS_BUF_IODONE_FUNC(bp)                 ((bp)->b_iodone)
 325#define XFS_BUF_SET_IODONE_FUNC(bp, func)       ((bp)->b_iodone = (func))
 326#define XFS_BUF_CLR_IODONE_FUNC(bp)             ((bp)->b_iodone = NULL)
 327#define XFS_BUF_SET_BDSTRAT_FUNC(bp, func)      ((bp)->b_strat = (func))
 328#define XFS_BUF_CLR_BDSTRAT_FUNC(bp)            ((bp)->b_strat = NULL)
 329
 330#define XFS_BUF_FSPRIVATE(bp, type)             ((type)(bp)->b_fspriv)
 331#define XFS_BUF_SET_FSPRIVATE(bp, val)          ((bp)->b_fspriv = (void*)(val))
 332#define XFS_BUF_FSPRIVATE2(bp, type)            ((type)(bp)->b_fspriv2)
 333#define XFS_BUF_SET_FSPRIVATE2(bp, val)         ((bp)->b_fspriv2 = (void*)(val))
 334#define XFS_BUF_SET_START(bp)                   do { } while (0)
 335#define XFS_BUF_SET_BRELSE_FUNC(bp, func)       ((bp)->b_relse = (func))
 336
 337#define XFS_BUF_PTR(bp)                 (xfs_caddr_t)((bp)->b_addr)
 338#define XFS_BUF_SET_PTR(bp, val, cnt)   xfs_buf_associate_memory(bp, val, cnt)
 339#define XFS_BUF_ADDR(bp)                ((bp)->b_bn)
 340#define XFS_BUF_SET_ADDR(bp, bno)       ((bp)->b_bn = (xfs_daddr_t)(bno))
 341#define XFS_BUF_OFFSET(bp)              ((bp)->b_file_offset)
 342#define XFS_BUF_SET_OFFSET(bp, off)     ((bp)->b_file_offset = (off))
 343#define XFS_BUF_COUNT(bp)               ((bp)->b_count_desired)
 344#define XFS_BUF_SET_COUNT(bp, cnt)      ((bp)->b_count_desired = (cnt))
 345#define XFS_BUF_SIZE(bp)                ((bp)->b_buffer_length)
 346#define XFS_BUF_SET_SIZE(bp, cnt)       ((bp)->b_buffer_length = (cnt))
 347
 348#define XFS_BUF_SET_VTYPE_REF(bp, type, ref)    do { } while (0)
 349#define XFS_BUF_SET_VTYPE(bp, type)             do { } while (0)
 350#define XFS_BUF_SET_REF(bp, ref)                do { } while (0)
 351
 352#define XFS_BUF_ISPINNED(bp)    xfs_buf_ispin(bp)
 353
 354#define XFS_BUF_VALUSEMA(bp)    xfs_buf_lock_value(bp)
 355#define XFS_BUF_CPSEMA(bp)      (xfs_buf_cond_lock(bp) == 0)
 356#define XFS_BUF_VSEMA(bp)       xfs_buf_unlock(bp)
 357#define XFS_BUF_PSEMA(bp,x)     xfs_buf_lock(bp)
 358#define XFS_BUF_FINISH_IOWAIT(bp)       complete(&bp->b_iowait);
 359
 360#define XFS_BUF_SET_TARGET(bp, target)  ((bp)->b_target = (target))
 361#define XFS_BUF_TARGET(bp)              ((bp)->b_target)
 362#define XFS_BUFTARG_NAME(target)        xfs_buf_target_name(target)
 363
 364static inline void xfs_buf_relse(xfs_buf_t *bp)
 365{
 366        if (!bp->b_relse)
 367                xfs_buf_unlock(bp);
 368        xfs_buf_rele(bp);
 369}
 370
 371#define xfs_bpin(bp)            xfs_buf_pin(bp)
 372#define xfs_bunpin(bp)          xfs_buf_unpin(bp)
 373
 374#define xfs_buftrace(id, bp)    \
 375            xfs_buf_trace(bp, id, NULL, (void *)__builtin_return_address(0))
 376
 377#define xfs_biodone(bp)         xfs_buf_ioend(bp, 0)
 378
 379#define xfs_biomove(bp, off, len, data, rw) \
 380            xfs_buf_iomove((bp), (off), (len), (data), \
 381                ((rw) == XFS_B_WRITE) ? XBRW_WRITE : XBRW_READ)
 382
 383#define xfs_biozero(bp, off, len) \
 384            xfs_buf_iomove((bp), (off), (len), NULL, XBRW_ZERO)
 385
 386
 387static inline int XFS_bwrite(xfs_buf_t *bp)
 388{
 389        int     iowait = (bp->b_flags & XBF_ASYNC) == 0;
 390        int     error = 0;
 391
 392        if (!iowait)
 393                bp->b_flags |= _XBF_RUN_QUEUES;
 394
 395        xfs_buf_delwri_dequeue(bp);
 396        xfs_buf_iostrategy(bp);
 397        if (iowait) {
 398                error = xfs_buf_iowait(bp);
 399                xfs_buf_relse(bp);
 400        }
 401        return error;
 402}
 403
 404#define XFS_bdstrat(bp) xfs_buf_iorequest(bp)
 405
 406#define xfs_iowait(bp)  xfs_buf_iowait(bp)
 407
 408#define xfs_baread(target, rablkno, ralen)  \
 409        xfs_buf_readahead((target), (rablkno), (ralen), XBF_DONT_BLOCK)
 410
 411
 412/*
 413 *      Handling of buftargs.
 414 */
 415extern xfs_buftarg_t *xfs_alloc_buftarg(struct block_device *, int);
 416extern void xfs_free_buftarg(struct xfs_mount *, struct xfs_buftarg *);
 417extern void xfs_wait_buftarg(xfs_buftarg_t *);
 418extern int xfs_setsize_buftarg(xfs_buftarg_t *, unsigned int, unsigned int);
 419extern int xfs_flush_buftarg(xfs_buftarg_t *, int);
 420#ifdef CONFIG_KDB_MODULES
 421extern struct list_head *xfs_get_buftarg_list(void);
 422#endif
 423
 424#define xfs_getsize_buftarg(buftarg)    block_size((buftarg)->bt_bdev)
 425#define xfs_readonly_buftarg(buftarg)   bdev_read_only((buftarg)->bt_bdev)
 426
 427#define xfs_binval(buftarg)             xfs_flush_buftarg(buftarg, 1)
 428#define XFS_bflush(buftarg)             xfs_flush_buftarg(buftarg, 1)
 429
 430#endif  /* __XFS_BUF_H__ */
 431