fs/nilfs2/the_nilfs.h

   1 /*
   2  * the_nilfs.h - the_nilfs shared structure.
   3  *
   4  * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
   5  *
   6  * This program is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License as published by
   8  * the Free Software Foundation; either version 2 of the License, or
   9  * (at your option) any later version.
  10  *
  11  * This program is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  * GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License
  17  * along with this program; if not, write to the Free Software
  18  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  19  *
  20  * Written by Ryusuke Konishi <ryusuke@osrg.net>
  21  *
  22  */
  23
  24 #ifndef _THE_NILFS_H
  25 #define _THE_NILFS_H
  26
  27 #include <linux/types.h>
  28 #include <linux/buffer_head.h>
  29 #include <linux/rbtree.h>
  30 #include <linux/fs.h>
  31 #include <linux/blkdev.h>
  32 #include <linux/backing-dev.h>
  33 #include <linux/slab.h>
  34
  35 struct nilfs_sc_info;
  36 struct nilfs_sysfs_dev_subgroups;
  37
  38 /* the_nilfs struct */
  39 enum {
  40         THE_NILFS_INIT = 0,     /* Information from super_block is set */
  41         THE_NILFS_DISCONTINUED, /* 'next' pointer chain has broken */
  42         THE_NILFS_GC_RUNNING,   /* gc process is running */
  43         THE_NILFS_SB_DIRTY,     /* super block is dirty */
  44 };
  45
  46 /**
  47  * struct the_nilfs - struct to supervise multiple nilfs mount points
  48  * @ns_flags: flags
  49  * @ns_flushed_device: flag indicating if all volatile data was flushed
  50  * @ns_bdev: block device
  51  * @ns_sem: semaphore for shared states
  52  * @ns_snapshot_mount_mutex: mutex to protect snapshot mounts
  53  * @ns_sbh: buffer heads of on-disk super blocks
  54  * @ns_sbp: pointers to super block data
  55  * @ns_sbwtime: previous write time of super block
  56  * @ns_sbwcount: write count of super block
  57  * @ns_sbsize: size of valid data in super block
  58  * @ns_mount_state: file system state
  59  * @ns_sb_update_freq: interval of periodical update of superblocks (in seconds)
  60  * @ns_seg_seq: segment sequence counter
  61  * @ns_segnum: index number of the latest full segment.
  62  * @ns_nextnum: index number of the full segment index to be used next
  63  * @ns_pseg_offset: offset of next partial segment in the current full segment
  64  * @ns_cno: next checkpoint number
  65  * @ns_ctime: write time of the last segment
  66  * @ns_nongc_ctime: write time of the last segment not for cleaner operation
  67  * @ns_ndirtyblks: Number of dirty data blocks
  68  * @ns_last_segment_lock: lock protecting fields for the latest segment
  69  * @ns_last_pseg: start block number of the latest segment
  70  * @ns_last_seq: sequence value of the latest segment
  71  * @ns_last_cno: checkpoint number of the latest segment
  72  * @ns_prot_seq: least sequence number of segments which must not be reclaimed
  73  * @ns_prev_seq: base sequence number used to decide if advance log cursor
  74  * @ns_writer: log writer
  75  * @ns_segctor_sem: semaphore protecting log write
  76  * @ns_dat: DAT file inode
  77  * @ns_cpfile: checkpoint file inode
  78  * @ns_sufile: segusage file inode
  79  * @ns_cptree: rb-tree of all mounted checkpoints (nilfs_root)
  80  * @ns_cptree_lock: lock protecting @ns_cptree
  81  * @ns_dirty_files: list of dirty files
  82  * @ns_inode_lock: lock protecting @ns_dirty_files
  83  * @ns_gc_inodes: dummy inodes to keep live blocks
  84  * @ns_next_generation: next generation number for inodes
  85  * @ns_next_gen_lock: lock protecting @ns_next_generation
  86  * @ns_mount_opt: mount options
  87  * @ns_resuid: uid for reserved blocks
  88  * @ns_resgid: gid for reserved blocks
  89  * @ns_interval: checkpoint creation interval
  90  * @ns_watermark: watermark for the number of dirty buffers
  91  * @ns_blocksize_bits: bit length of block size
  92  * @ns_blocksize: block size
  93  * @ns_nsegments: number of segments in filesystem
  94  * @ns_blocks_per_segment: number of blocks per segment
  95  * @ns_r_segments_percentage: reserved segments percentage
  96  * @ns_nrsvsegs: number of reserved segments
  97  * @ns_first_data_block: block number of first data block
  98  * @ns_inode_size: size of on-disk inode
  99  * @ns_first_ino: first not-special inode number
 100  * @ns_crc_seed: seed value of CRC32 calculation
 101  * @ns_dev_kobj: /sys/fs/<nilfs>/<device>
 102  * @ns_dev_kobj_unregister: completion state
 103  * @ns_dev_subgroups: <device> subgroups pointer
 104  */
 105 struct the_nilfs {
 106         unsigned long           ns_flags;
 107         int                     ns_flushed_device;
 108
 109         struct block_device    *ns_bdev;
 110         struct rw_semaphore     ns_sem;
 111         struct mutex            ns_snapshot_mount_mutex;
 112
 113         /*
 114          * used for
 115          * - loading the latest checkpoint exclusively.
 116          * - allocating a new full segment.
 117          */
 118         struct buffer_head     *ns_sbh[2];
 119         struct nilfs_super_block *ns_sbp[2];
 120         time_t                  ns_sbwtime;
 121         unsigned                ns_sbwcount;
 122         unsigned                ns_sbsize;
 123         unsigned                ns_mount_state;
 124         unsigned                ns_sb_update_freq;
 125
 126         /*
 127          * Following fields are dedicated to a writable FS-instance.
 128          * Except for the period seeking checkpoint, code outside the segment
 129          * constructor must lock a segment semaphore while accessing these
 130          * fields.
 131          * The writable FS-instance is sole during a lifetime of the_nilfs.
 132          */
 133         u64                     ns_seg_seq;
 134         __u64                   ns_segnum;
 135         __u64                   ns_nextnum;
 136         unsigned long           ns_pseg_offset;
 137         __u64                   ns_cno;
 138         time_t                  ns_ctime;
 139         time_t                  ns_nongc_ctime;
 140         atomic_t                ns_ndirtyblks;
 141
 142         /*
 143          * The following fields hold information on the latest partial segment
 144          * written to disk with a super root.  These fields are protected by
 145          * ns_last_segment_lock.
 146          */
 147         spinlock_t              ns_last_segment_lock;
 148         sector_t                ns_last_pseg;
 149         u64                     ns_last_seq;
 150         __u64                   ns_last_cno;
 151         u64                     ns_prot_seq;
 152         u64                     ns_prev_seq;
 153
 154         struct nilfs_sc_info   *ns_writer;
 155         struct rw_semaphore     ns_segctor_sem;
 156
 157         /*
 158          * Following fields are lock free except for the period before
 159          * the_nilfs is initialized.
 160          */
 161         struct inode           *ns_dat;
 162         struct inode           *ns_cpfile;
 163         struct inode           *ns_sufile;
 164
 165         /* Checkpoint tree */
 166         struct rb_root          ns_cptree;
 167         spinlock_t              ns_cptree_lock;
 168
 169         /* Dirty inode list */
 170         struct list_head        ns_dirty_files;
 171         spinlock_t              ns_inode_lock;
 172
 173         /* GC inode list */
 174         struct list_head        ns_gc_inodes;
 175
 176         /* Inode allocator */
 177         u32                     ns_next_generation;
 178         spinlock_t              ns_next_gen_lock;
 179
 180         /* Mount options */
 181         unsigned long           ns_mount_opt;
 182
 183         uid_t                   ns_resuid;
 184         gid_t                   ns_resgid;
 185         unsigned long           ns_interval;
 186         unsigned long           ns_watermark;
 187
 188         /* Disk layout information (static) */
 189         unsigned int            ns_blocksize_bits;
 190         unsigned int            ns_blocksize;
 191         unsigned long           ns_nsegments;
 192         unsigned long           ns_blocks_per_segment;
 193         unsigned long           ns_r_segments_percentage;
 194         unsigned long           ns_nrsvsegs;
 195         unsigned long           ns_first_data_block;
 196         int                     ns_inode_size;
 197         int                     ns_first_ino;
 198         u32                     ns_crc_seed;
 199
 200         /* /sys/fs/<nilfs>/<device> */
 201         struct kobject ns_dev_kobj;
 202         struct completion ns_dev_kobj_unregister;
 203         struct nilfs_sysfs_dev_subgroups *ns_dev_subgroups;
 204 };
 205
 206 #define THE_NILFS_FNS(bit, name)                                        \
 207 static inline void set_nilfs_##name(struct the_nilfs *nilfs)            \
 208 {                                                                       \
 209         set_bit(THE_NILFS_##bit, &(nilfs)->ns_flags);                   \
 210 }                                                                       \
 211 static inline void clear_nilfs_##name(struct the_nilfs *nilfs)          \
 212 {                                                                       \
 213         clear_bit(THE_NILFS_##bit, &(nilfs)->ns_flags);                 \
 214 }                                                                       \
 215 static inline int nilfs_##name(struct the_nilfs *nilfs)                 \
 216 {                                                                       \
 217         return test_bit(THE_NILFS_##bit, &(nilfs)->ns_flags);           \
 218 }
 219
 220 THE_NILFS_FNS(INIT, init)
 221 THE_NILFS_FNS(DISCONTINUED, discontinued)
 222 THE_NILFS_FNS(GC_RUNNING, gc_running)
 223 THE_NILFS_FNS(SB_DIRTY, sb_dirty)
 224
 225 /*
 226  * Mount option operations
 227  */
 228 #define nilfs_clear_opt(nilfs, opt)  \
 229         do { (nilfs)->ns_mount_opt &= ~NILFS_MOUNT_##opt; } while (0)
 230 #define nilfs_set_opt(nilfs, opt)  \
 231         do { (nilfs)->ns_mount_opt |= NILFS_MOUNT_##opt; } while (0)
 232 #define nilfs_test_opt(nilfs, opt) ((nilfs)->ns_mount_opt & NILFS_MOUNT_##opt)
 233 #define nilfs_write_opt(nilfs, mask, opt)                               \
 234         do { (nilfs)->ns_mount_opt =                                    \
 235                 (((nilfs)->ns_mount_opt & ~NILFS_MOUNT_##mask) |        \
 236                  NILFS_MOUNT_##opt);                                    \
 237         } while (0)
 238
 239 /**
 240  * struct nilfs_root - nilfs root object
 241  * @cno: checkpoint number
 242  * @rb_node: red-black tree node
 243  * @count: refcount of this structure
 244  * @nilfs: nilfs object
 245  * @ifile: inode file
 246  * @inodes_count: number of inodes
 247  * @blocks_count: number of blocks
 248  * @snapshot_kobj: /sys/fs/<nilfs>/<device>/mounted_snapshots/<snapshot>
 249  * @snapshot_kobj_unregister: completion state for kernel object
 250  */
 251 struct nilfs_root {
 252         __u64 cno;
 253         struct rb_node rb_node;
 254
 255         atomic_t count;
 256         struct the_nilfs *nilfs;
 257         struct inode *ifile;
 258
 259         atomic64_t inodes_count;
 260         atomic64_t blocks_count;
 261
 262         /* /sys/fs/<nilfs>/<device>/mounted_snapshots/<snapshot> */
 263         struct kobject snapshot_kobj;
 264         struct completion snapshot_kobj_unregister;
 265 };
 266
 267 /* Special checkpoint number */
 268 #define NILFS_CPTREE_CURRENT_CNO        0
 269
 270 /* Minimum interval of periodical update of superblocks (in seconds) */
 271 #define NILFS_SB_FREQ           10
 272
 273 static inline int nilfs_sb_need_update(struct the_nilfs *nilfs)
 274 {
 275         u64 t = get_seconds();
 276         return t < nilfs->ns_sbwtime ||
 277                 t > nilfs->ns_sbwtime + nilfs->ns_sb_update_freq;
 278 }
 279
 280 static inline int nilfs_sb_will_flip(struct the_nilfs *nilfs)
 281 {
 282         int flip_bits = nilfs->ns_sbwcount & 0x0FL;
 283         return (flip_bits != 0x08 && flip_bits != 0x0F);
 284 }
 285
 286 void nilfs_set_last_segment(struct the_nilfs *, sector_t, u64, __u64);
 287 struct the_nilfs *alloc_nilfs(struct block_device *bdev);
 288 void destroy_nilfs(struct the_nilfs *nilfs);
 289 int init_nilfs(struct the_nilfs *nilfs, struct super_block *sb, char *data);
 290 int load_nilfs(struct the_nilfs *nilfs, struct super_block *sb);
 291 unsigned long nilfs_nrsvsegs(struct the_nilfs *nilfs, unsigned long nsegs);
 292 void nilfs_set_nsegments(struct the_nilfs *nilfs, unsigned long nsegs);
 293 int nilfs_discard_segments(struct the_nilfs *, __u64 *, size_t);
 294 int nilfs_count_free_blocks(struct the_nilfs *, sector_t *);
 295 struct nilfs_root *nilfs_lookup_root(struct the_nilfs *nilfs, __u64 cno);
 296 struct nilfs_root *nilfs_find_or_create_root(struct the_nilfs *nilfs,
 297                                              __u64 cno);
 298 void nilfs_put_root(struct nilfs_root *root);
 299 int nilfs_near_disk_full(struct the_nilfs *);
 300 void nilfs_fall_back_super_block(struct the_nilfs *);
 301 void nilfs_swap_super_block(struct the_nilfs *);
 302
 303
 304 static inline void nilfs_get_root(struct nilfs_root *root)
 305 {
 306         atomic_inc(&root->count);
 307 }
 308
 309 static inline int nilfs_valid_fs(struct the_nilfs *nilfs)
 310 {
 311         unsigned valid_fs;
 312
 313         down_read(&nilfs->ns_sem);
 314         valid_fs = (nilfs->ns_mount_state & NILFS_VALID_FS);
 315         up_read(&nilfs->ns_sem);
 316         return valid_fs;
 317 }
 318
 319 static inline void
 320 nilfs_get_segment_range(struct the_nilfs *nilfs, __u64 segnum,
 321                         sector_t *seg_start, sector_t *seg_end)
 322 {
 323         *seg_start = (sector_t)nilfs->ns_blocks_per_segment * segnum;
 324         *seg_end = *seg_start + nilfs->ns_blocks_per_segment - 1;
 325         if (segnum == 0)
 326                 *seg_start = nilfs->ns_first_data_block;
 327 }
 328
 329 static inline sector_t
 330 nilfs_get_segment_start_blocknr(struct the_nilfs *nilfs, __u64 segnum)
 331 {
 332         return (segnum == 0) ? nilfs->ns_first_data_block :
 333                 (sector_t)nilfs->ns_blocks_per_segment * segnum;
 334 }
 335
 336 static inline __u64
 337 nilfs_get_segnum_of_block(struct the_nilfs *nilfs, sector_t blocknr)
 338 {
 339         sector_t segnum = blocknr;
 340
 341         sector_div(segnum, nilfs->ns_blocks_per_segment);
 342         return segnum;
 343 }
 344
 345 static inline void
 346 nilfs_terminate_segment(struct the_nilfs *nilfs, sector_t seg_start,
 347                         sector_t seg_end)
 348 {
 349         /* terminate the current full segment (used in case of I/O-error) */
 350         nilfs->ns_pseg_offset = seg_end - seg_start + 1;
 351 }
 352
 353 static inline void nilfs_shift_to_next_segment(struct the_nilfs *nilfs)
 354 {
 355         /* move forward with a full segment */
 356         nilfs->ns_segnum = nilfs->ns_nextnum;
 357         nilfs->ns_pseg_offset = 0;
 358         nilfs->ns_seg_seq++;
 359 }
 360
 361 static inline __u64 nilfs_last_cno(struct the_nilfs *nilfs)
 362 {
 363         __u64 cno;
 364
 365         spin_lock(&nilfs->ns_last_segment_lock);
 366         cno = nilfs->ns_last_cno;
 367         spin_unlock(&nilfs->ns_last_segment_lock);
 368         return cno;
 369 }
 370
 371 static inline int nilfs_segment_is_active(struct the_nilfs *nilfs, __u64 n)
 372 {
 373         return n == nilfs->ns_segnum || n == nilfs->ns_nextnum;
 374 }
 375
 376 static inline int nilfs_flush_device(struct the_nilfs *nilfs)
 377 {
 378         int err;
 379
 380         if (!nilfs_test_opt(nilfs, BARRIER) || nilfs->ns_flushed_device)
 381                 return 0;
 382
 383         nilfs->ns_flushed_device = 1;
 384         /*
 385          * the store to ns_flushed_device must not be reordered after
 386          * blkdev_issue_flush().
 387          */
 388         smp_wmb();
 389
 390         err = blkdev_issue_flush(nilfs->ns_bdev, GFP_KERNEL, NULL);
 391         if (err != -EIO)
 392                 err = 0;
 393         return err;
 394 }
 395
 396 #endif /* _THE_NILFS_H */