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_FS_MANAGED (1 << 8) /* filesystem controls freeing memory */ 55#define XBF_ORDERED (1 << 11)/* use ordered writes */ 56#define XBF_READ_AHEAD (1 << 12)/* asynchronous read-ahead */ 57#define XBF_LOG_BUFFER (1 << 13)/* this is a buffer used for the log */ 58 59/* flags used only as arguments to access routines */ 60#define XBF_LOCK (1 << 14)/* lock requested */ 61#define XBF_TRYLOCK (1 << 15)/* lock requested, but do not wait */ 62#define XBF_DONT_BLOCK (1 << 16)/* do not block in current thread */ 63 64/* flags used only internally */ 65#define _XBF_PAGE_CACHE (1 << 17)/* backed by pagecache */ 66#define _XBF_PAGES (1 << 18)/* backed by refcounted pages */ 67#define _XBF_RUN_QUEUES (1 << 19)/* run block device task queue */ 68#define _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#define _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#define _XFS_BARRIER_FAILED (1 << 23) 96 97typedef unsigned int xfs_buf_flags_t; 98 99#define XFS_BUF_FLAGS \ 100 { XBF_READ, "READ" }, \ 101 { XBF_WRITE, "WRITE" }, \ 102 { XBF_MAPPED, "MAPPED" }, \ 103 { XBF_ASYNC, "ASYNC" }, \ 104 { XBF_DONE, "DONE" }, \ 105 { XBF_DELWRI, "DELWRI" }, \ 106 { XBF_STALE, "STALE" }, \ 107 { XBF_FS_MANAGED, "FS_MANAGED" }, \ 108 { XBF_ORDERED, "ORDERED" }, \ 109 { XBF_READ_AHEAD, "READ_AHEAD" }, \ 110 { XBF_LOCK, "LOCK" }, /* should never be set */\ 111 { XBF_TRYLOCK, "TRYLOCK" }, /* ditto */\ 112 { XBF_DONT_BLOCK, "DONT_BLOCK" }, /* ditto */\ 113 { _XBF_PAGE_CACHE, "PAGE_CACHE" }, \ 114 { _XBF_PAGES, "PAGES" }, \ 115 { _XBF_RUN_QUEUES, "RUN_QUEUES" }, \ 116 { _XBF_DELWRI_Q, "DELWRI_Q" }, \ 117 { _XBF_PAGE_LOCKED, "PAGE_LOCKED" }, \ 118 { _XFS_BARRIER_FAILED, "BARRIER_FAILED" } 119 120 121typedef enum { 122 XBT_FORCE_SLEEP = 0, 123 XBT_FORCE_FLUSH = 1, 124} xfs_buftarg_flags_t; 125 126typedef struct xfs_bufhash { 127 struct list_head bh_list; 128 spinlock_t bh_lock; 129} xfs_bufhash_t; 130 131typedef struct xfs_buftarg { 132 dev_t bt_dev; 133 struct block_device *bt_bdev; 134 struct address_space *bt_mapping; 135 unsigned int bt_bsize; 136 unsigned int bt_sshift; 137 size_t bt_smask; 138 139 /* per device buffer hash table */ 140 uint bt_hashshift; 141 xfs_bufhash_t *bt_hash; 142 143 /* per device delwri queue */ 144 struct task_struct *bt_task; 145 struct list_head bt_list; 146 struct list_head bt_delwrite_queue; 147 spinlock_t bt_delwrite_lock; 148 unsigned long bt_flags; 149} xfs_buftarg_t; 150 151/* 152 * xfs_buf_t: Buffer structure for pagecache-based buffers 153 * 154 * This buffer structure is used by the pagecache buffer management routines 155 * to refer to an assembly of pages forming a logical buffer. 156 * 157 * The buffer structure is used on a temporary basis only, and discarded when 158 * released. The real data storage is recorded in the pagecache. Buffers are 159 * hashed to the block device on which the file system resides. 160 */ 161 162struct xfs_buf; 163typedef void (*xfs_buf_iodone_t)(struct xfs_buf *); 164typedef void (*xfs_buf_relse_t)(struct xfs_buf *); 165typedef int (*xfs_buf_bdstrat_t)(struct xfs_buf *); 166 167#define XB_PAGES 2 168 169typedef struct xfs_buf { 170 struct semaphore b_sema; /* semaphore for lockables */ 171 unsigned long b_queuetime; /* time buffer was queued */ 172 atomic_t b_pin_count; /* pin count */ 173 wait_queue_head_t b_waiters; /* unpin waiters */ 174 struct list_head b_list; 175 xfs_buf_flags_t b_flags; /* status flags */ 176 struct list_head b_hash_list; /* hash table list */ 177 xfs_bufhash_t *b_hash; /* hash table list start */ 178 xfs_buftarg_t *b_target; /* buffer target (device) */ 179 atomic_t b_hold; /* reference count */ 180 xfs_daddr_t b_bn; /* block number for I/O */ 181 xfs_off_t b_file_offset; /* offset in file */ 182 size_t b_buffer_length;/* size of buffer in bytes */ 183 size_t b_count_desired;/* desired transfer size */ 184 void *b_addr; /* virtual address of buffer */ 185 struct work_struct b_iodone_work; 186 atomic_t b_io_remaining; /* #outstanding I/O requests */ 187 xfs_buf_iodone_t b_iodone; /* I/O completion function */ 188 xfs_buf_relse_t b_relse; /* releasing function */ 189 struct completion b_iowait; /* queue for I/O waiters */ 190 void *b_fspriv; 191 void *b_fspriv2; 192 struct xfs_mount *b_mount; 193 unsigned short b_error; /* error code on I/O */ 194 unsigned int b_page_count; /* size of page array */ 195 unsigned int b_offset; /* page offset in first page */ 196 struct page **b_pages; /* array of page pointers */ 197 struct page *b_page_array[XB_PAGES]; /* inline pages */ 198#ifdef XFS_BUF_LOCK_TRACKING 199 int b_last_holder; 200#endif 201} xfs_buf_t; 202 203 204/* Finding and Reading Buffers */ 205extern xfs_buf_t *_xfs_buf_find(xfs_buftarg_t *, xfs_off_t, size_t, 206 xfs_buf_flags_t, xfs_buf_t *); 207#define xfs_incore(buftarg,blkno,len,lockit) \ 208 _xfs_buf_find(buftarg, blkno ,len, lockit, NULL) 209 210extern xfs_buf_t *xfs_buf_get(xfs_buftarg_t *, xfs_off_t, size_t, 211 xfs_buf_flags_t); 212extern xfs_buf_t *xfs_buf_read(xfs_buftarg_t *, xfs_off_t, size_t, 213 xfs_buf_flags_t); 214 215extern xfs_buf_t *xfs_buf_get_empty(size_t, xfs_buftarg_t *); 216extern xfs_buf_t *xfs_buf_get_noaddr(size_t, xfs_buftarg_t *); 217extern int xfs_buf_associate_memory(xfs_buf_t *, void *, size_t); 218extern void xfs_buf_hold(xfs_buf_t *); 219extern void xfs_buf_readahead(xfs_buftarg_t *, xfs_off_t, size_t, 220 xfs_buf_flags_t); 221 222/* Releasing Buffers */ 223extern void xfs_buf_free(xfs_buf_t *); 224extern void xfs_buf_rele(xfs_buf_t *); 225 226/* Locking and Unlocking Buffers */ 227extern int xfs_buf_cond_lock(xfs_buf_t *); 228extern int xfs_buf_lock_value(xfs_buf_t *); 229extern void xfs_buf_lock(xfs_buf_t *); 230extern void xfs_buf_unlock(xfs_buf_t *); 231 232/* Buffer Read and Write Routines */ 233extern int xfs_bwrite(struct xfs_mount *mp, struct xfs_buf *bp); 234extern void xfs_bdwrite(void *mp, xfs_buf_t *bp); 235 236extern void xfsbdstrat(struct xfs_mount *, struct xfs_buf *); 237extern int xfs_bdstrat_cb(struct xfs_buf *); 238 239extern void xfs_buf_ioend(xfs_buf_t *, int); 240extern void xfs_buf_ioerror(xfs_buf_t *, int); 241extern int xfs_buf_iorequest(xfs_buf_t *); 242extern int xfs_buf_iowait(xfs_buf_t *); 243extern void xfs_buf_iomove(xfs_buf_t *, size_t, size_t, void *, 244 xfs_buf_rw_t); 245 246static inline int xfs_buf_geterror(xfs_buf_t *bp) 247{ 248 return bp ? bp->b_error : ENOMEM; 249} 250 251/* Buffer Utility Routines */ 252extern xfs_caddr_t xfs_buf_offset(xfs_buf_t *, size_t); 253 254/* Delayed Write Buffer Routines */ 255extern void xfs_buf_delwri_dequeue(xfs_buf_t *); 256extern void xfs_buf_delwri_promote(xfs_buf_t *); 257 258/* Buffer Daemon Setup Routines */ 259extern int xfs_buf_init(void); 260extern void xfs_buf_terminate(void); 261 262#define xfs_buf_target_name(target) \ 263 ({ char __b[BDEVNAME_SIZE]; bdevname((target)->bt_bdev, __b); __b; }) 264 265 266#define XFS_BUF_BFLAGS(bp) ((bp)->b_flags) 267#define XFS_BUF_ZEROFLAGS(bp) ((bp)->b_flags &= \ 268 ~(XBF_READ|XBF_WRITE|XBF_ASYNC|XBF_DELWRI|XBF_ORDERED)) 269 270#define XFS_BUF_STALE(bp) ((bp)->b_flags |= XBF_STALE) 271#define XFS_BUF_UNSTALE(bp) ((bp)->b_flags &= ~XBF_STALE) 272#define XFS_BUF_ISSTALE(bp) ((bp)->b_flags & XBF_STALE) 273#define XFS_BUF_SUPER_STALE(bp) do { \ 274 XFS_BUF_STALE(bp); \ 275 xfs_buf_delwri_dequeue(bp); \ 276 XFS_BUF_DONE(bp); \ 277 } while (0) 278 279#define XFS_BUF_UNMANAGE(bp) ((bp)->b_flags &= ~XBF_FS_MANAGED) 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#define XFS_BUF_SET_BRELSE_FUNC(bp, func) ((bp)->b_relse = (func)) 324 325#define XFS_BUF_PTR(bp) (xfs_caddr_t)((bp)->b_addr) 326#define XFS_BUF_SET_PTR(bp, val, cnt) xfs_buf_associate_memory(bp, val, cnt) 327#define XFS_BUF_ADDR(bp) ((bp)->b_bn) 328#define XFS_BUF_SET_ADDR(bp, bno) ((bp)->b_bn = (xfs_daddr_t)(bno)) 329#define XFS_BUF_OFFSET(bp) ((bp)->b_file_offset) 330#define XFS_BUF_SET_OFFSET(bp, off) ((bp)->b_file_offset = (off)) 331#define XFS_BUF_COUNT(bp) ((bp)->b_count_desired) 332#define XFS_BUF_SET_COUNT(bp, cnt) ((bp)->b_count_desired = (cnt)) 333#define XFS_BUF_SIZE(bp) ((bp)->b_buffer_length) 334#define XFS_BUF_SET_SIZE(bp, cnt) ((bp)->b_buffer_length = (cnt)) 335 336#define XFS_BUF_SET_VTYPE_REF(bp, type, ref) do { } while (0) 337#define XFS_BUF_SET_VTYPE(bp, type) do { } while (0) 338#define XFS_BUF_SET_REF(bp, ref) do { } while (0) 339 340#define XFS_BUF_ISPINNED(bp) atomic_read(&((bp)->b_pin_count)) 341 342#define XFS_BUF_VALUSEMA(bp) xfs_buf_lock_value(bp) 343#define XFS_BUF_CPSEMA(bp) (xfs_buf_cond_lock(bp) == 0) 344#define XFS_BUF_VSEMA(bp) xfs_buf_unlock(bp) 345#define XFS_BUF_PSEMA(bp,x) xfs_buf_lock(bp) 346#define XFS_BUF_FINISH_IOWAIT(bp) complete(&bp->b_iowait); 347 348#define XFS_BUF_SET_TARGET(bp, target) ((bp)->b_target = (target)) 349#define XFS_BUF_TARGET(bp) ((bp)->b_target) 350#define XFS_BUFTARG_NAME(target) xfs_buf_target_name(target) 351 352static inline void xfs_buf_relse(xfs_buf_t *bp) 353{ 354 if (!bp->b_relse) 355 xfs_buf_unlock(bp); 356 xfs_buf_rele(bp); 357} 358 359#define xfs_biodone(bp) xfs_buf_ioend(bp, 0) 360 361#define xfs_biomove(bp, off, len, data, rw) \ 362 xfs_buf_iomove((bp), (off), (len), (data), \ 363 ((rw) == XBF_WRITE) ? XBRW_WRITE : XBRW_READ) 364 365#define xfs_biozero(bp, off, len) \ 366 xfs_buf_iomove((bp), (off), (len), NULL, XBRW_ZERO) 367 368#define xfs_iowait(bp) xfs_buf_iowait(bp) 369 370#define xfs_baread(target, rablkno, ralen) \ 371 xfs_buf_readahead((target), (rablkno), (ralen), XBF_DONT_BLOCK) 372 373 374/* 375 * Handling of buftargs. 376 */ 377extern xfs_buftarg_t *xfs_alloc_buftarg(struct block_device *, int, const char *); 378extern void xfs_free_buftarg(struct xfs_mount *, struct xfs_buftarg *); 379extern void xfs_wait_buftarg(xfs_buftarg_t *); 380extern int xfs_setsize_buftarg(xfs_buftarg_t *, unsigned int, unsigned int); 381extern int xfs_flush_buftarg(xfs_buftarg_t *, int); 382 383#ifdef CONFIG_KDB_MODULES 384extern struct list_head *xfs_get_buftarg_list(void); 385#endif 386 387#define xfs_getsize_buftarg(buftarg) block_size((buftarg)->bt_bdev) 388#define xfs_readonly_buftarg(buftarg) bdev_read_only((buftarg)->bt_bdev) 389 390#define xfs_binval(buftarg) xfs_flush_buftarg(buftarg, 1) 391#define XFS_bflush(buftarg) xfs_flush_buftarg(buftarg, 1) 392 393#endif /* __XFS_BUF_H__ */ 394