fs/btrfs/volumes.h

   1 /*
   2  * Copyright (C) 2007 Oracle.  All rights reserved.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public
   6  * License v2 as published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  11  * General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public
  14  * License along with this program; if not, write to the
  15  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  16  * Boston, MA 021110-1307, USA.
  17  */
  18
  19 #ifndef __BTRFS_VOLUMES_
  20 #define __BTRFS_VOLUMES_
  21
  22 #include <linux/bio.h>
  23 #include <linux/sort.h>
  24 #include <linux/btrfs.h>
  25 #include "async-thread.h"
  26
  27 extern struct mutex uuid_mutex;
  28
  29 #define BTRFS_STRIPE_LEN        SZ_64K
  30
  31 struct buffer_head;
  32 struct btrfs_pending_bios {
  33         struct bio *head;
  34         struct bio *tail;
  35 };
  36
  37 /*
  38  * Use sequence counter to get consistent device stat data on
  39  * 32-bit processors.
  40  */
  41 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
  42 #include <linux/seqlock.h>
  43 #define __BTRFS_NEED_DEVICE_DATA_ORDERED
  44 #define btrfs_device_data_ordered_init(device)  \
  45         seqcount_init(&device->data_seqcount)
  46 #else
  47 #define btrfs_device_data_ordered_init(device) do { } while (0)
  48 #endif
  49
  50 struct btrfs_device {
  51         struct list_head dev_list;
  52         struct list_head dev_alloc_list;
  53         struct btrfs_fs_devices *fs_devices;
  54         struct btrfs_fs_info *fs_info;
  55
  56         struct rcu_string *name;
  57
  58         u64 generation;
  59
  60         spinlock_t io_lock ____cacheline_aligned;
  61         int running_pending;
  62         /* regular prio bios */
  63         struct btrfs_pending_bios pending_bios;
  64         /* sync bios */
  65         struct btrfs_pending_bios pending_sync_bios;
  66
  67         struct block_device *bdev;
  68
  69         /* the mode sent to blkdev_get */
  70         fmode_t mode;
  71
  72         int writeable;
  73         int in_fs_metadata;
  74         int missing;
  75         int can_discard;
  76         int is_tgtdev_for_dev_replace;
  77         int last_flush_error;
  78         int flush_bio_sent;
  79
  80 #ifdef __BTRFS_NEED_DEVICE_DATA_ORDERED
  81         seqcount_t data_seqcount;
  82 #endif
  83
  84         /* the internal btrfs device id */
  85         u64 devid;
  86
  87         /* size of the device in memory */
  88         u64 total_bytes;
  89
  90         /* size of the device on disk */
  91         u64 disk_total_bytes;
  92
  93         /* bytes used */
  94         u64 bytes_used;
  95
  96         /* optimal io alignment for this device */
  97         u32 io_align;
  98
  99         /* optimal io width for this device */
 100         u32 io_width;
 101         /* type and info about this device */
 102         u64 type;
 103
 104         /* minimal io size for this device */
 105         u32 sector_size;
 106
 107         /* physical drive uuid (or lvm uuid) */
 108         u8 uuid[BTRFS_UUID_SIZE];
 109
 110         /*
 111          * size of the device on the current transaction
 112          *
 113          * This variant is update when committing the transaction,
 114          * and protected by device_list_mutex
 115          */
 116         u64 commit_total_bytes;
 117
 118         /* bytes used on the current transaction */
 119         u64 commit_bytes_used;
 120         /*
 121          * used to manage the device which is resized
 122          *
 123          * It is protected by chunk_lock.
 124          */
 125         struct list_head resized_list;
 126
 127         /* for sending down flush barriers */
 128         struct bio *flush_bio;
 129         struct completion flush_wait;
 130
 131         /* per-device scrub information */
 132         struct scrub_ctx *scrub_device;
 133
 134         struct btrfs_work work;
 135         struct rcu_head rcu;
 136         struct work_struct rcu_work;
 137
 138         /* readahead state */
 139         spinlock_t reada_lock;
 140         atomic_t reada_in_flight;
 141         u64 reada_next;
 142         struct reada_zone *reada_curr_zone;
 143         struct radix_tree_root reada_zones;
 144         struct radix_tree_root reada_extents;
 145
 146         /* disk I/O failure stats. For detailed description refer to
 147          * enum btrfs_dev_stat_values in ioctl.h */
 148         int dev_stats_valid;
 149
 150         /* Counter to record the change of device stats */
 151         atomic_t dev_stats_ccnt;
 152         atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX];
 153 };
 154
 155 /*
 156  * If we read those variants at the context of their own lock, we needn't
 157  * use the following helpers, reading them directly is safe.
 158  */
 159 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 160 #define BTRFS_DEVICE_GETSET_FUNCS(name)                                 \
 161 static inline u64                                                       \
 162 btrfs_device_get_##name(const struct btrfs_device *dev)                 \
 163 {                                                                       \
 164         u64 size;                                                       \
 165         unsigned int seq;                                               \
 166                                                                         \
 167         do {                                                            \
 168                 seq = read_seqcount_begin(&dev->data_seqcount);         \
 169                 size = dev->name;                                       \
 170         } while (read_seqcount_retry(&dev->data_seqcount, seq));        \
 171         return size;                                                    \
 172 }                                                                       \
 173                                                                         \
 174 static inline void                                                      \
 175 btrfs_device_set_##name(struct btrfs_device *dev, u64 size)             \
 176 {                                                                       \
 177         preempt_disable();                                              \
 178         write_seqcount_begin(&dev->data_seqcount);                      \
 179         dev->name = size;                                               \
 180         write_seqcount_end(&dev->data_seqcount);                        \
 181         preempt_enable();                                               \
 182 }
 183 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
 184 #define BTRFS_DEVICE_GETSET_FUNCS(name)                                 \
 185 static inline u64                                                       \
 186 btrfs_device_get_##name(const struct btrfs_device *dev)                 \
 187 {                                                                       \
 188         u64 size;                                                       \
 189                                                                         \
 190         preempt_disable();                                              \
 191         size = dev->name;                                               \
 192         preempt_enable();                                               \
 193         return size;                                                    \
 194 }                                                                       \
 195                                                                         \
 196 static inline void                                                      \
 197 btrfs_device_set_##name(struct btrfs_device *dev, u64 size)             \
 198 {                                                                       \
 199         preempt_disable();                                              \
 200         dev->name = size;                                               \
 201         preempt_enable();                                               \
 202 }
 203 #else
 204 #define BTRFS_DEVICE_GETSET_FUNCS(name)                                 \
 205 static inline u64                                                       \
 206 btrfs_device_get_##name(const struct btrfs_device *dev)                 \
 207 {                                                                       \
 208         return dev->name;                                               \
 209 }                                                                       \
 210                                                                         \
 211 static inline void                                                      \
 212 btrfs_device_set_##name(struct btrfs_device *dev, u64 size)             \
 213 {                                                                       \
 214         dev->name = size;                                               \
 215 }
 216 #endif
 217
 218 BTRFS_DEVICE_GETSET_FUNCS(total_bytes);
 219 BTRFS_DEVICE_GETSET_FUNCS(disk_total_bytes);
 220 BTRFS_DEVICE_GETSET_FUNCS(bytes_used);
 221
 222 struct btrfs_fs_devices {
 223         u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
 224
 225         u64 num_devices;
 226         u64 open_devices;
 227         u64 rw_devices;
 228         u64 missing_devices;
 229         u64 total_rw_bytes;
 230         u64 total_devices;
 231         struct block_device *latest_bdev;
 232
 233         /* all of the devices in the FS, protected by a mutex
 234          * so we can safely walk it to write out the supers without
 235          * worrying about add/remove by the multi-device code.
 236          * Scrubbing super can kick off supers writing by holding
 237          * this mutex lock.
 238          */
 239         struct mutex device_list_mutex;
 240         struct list_head devices;
 241
 242         struct list_head resized_devices;
 243         /* devices not currently being allocated */
 244         struct list_head alloc_list;
 245         struct list_head list;
 246
 247         struct btrfs_fs_devices *seed;
 248         int seeding;
 249
 250         int opened;
 251
 252         /* set when we find or add a device that doesn't have the
 253          * nonrot flag set
 254          */
 255         int rotating;
 256
 257         struct btrfs_fs_info *fs_info;
 258         /* sysfs kobjects */
 259         struct kobject fsid_kobj;
 260         struct kobject *device_dir_kobj;
 261         struct completion kobj_unregister;
 262 };
 263
 264 #define BTRFS_BIO_INLINE_CSUM_SIZE      64
 265
 266 /*
 267  * we need the mirror number and stripe index to be passed around
 268  * the call chain while we are processing end_io (especially errors).
 269  * Really, what we need is a btrfs_bio structure that has this info
 270  * and is properly sized with its stripe array, but we're not there
 271  * quite yet.  We have our own btrfs bioset, and all of the bios
 272  * we allocate are actually btrfs_io_bios.  We'll cram as much of
 273  * struct btrfs_bio as we can into this over time.
 274  */
 275 typedef void (btrfs_io_bio_end_io_t) (struct btrfs_io_bio *bio, int err);
 276 struct btrfs_io_bio {
 277         unsigned int mirror_num;
 278         unsigned int stripe_index;
 279         u64 logical;
 280         u8 *csum;
 281         u8 csum_inline[BTRFS_BIO_INLINE_CSUM_SIZE];
 282         u8 *csum_allocated;
 283         btrfs_io_bio_end_io_t *end_io;
 284         struct bvec_iter iter;
 285         /*
 286          * This member must come last, bio_alloc_bioset will allocate enough
 287          * bytes for entire btrfs_io_bio but relies on bio being last.
 288          */
 289         struct bio bio;
 290 };
 291
 292 static inline struct btrfs_io_bio *btrfs_io_bio(struct bio *bio)
 293 {
 294         return container_of(bio, struct btrfs_io_bio, bio);
 295 }
 296
 297 struct btrfs_bio_stripe {
 298         struct btrfs_device *dev;
 299         u64 physical;
 300         u64 length; /* only used for discard mappings */
 301 };
 302
 303 struct btrfs_bio;
 304 typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err);
 305
 306 struct btrfs_bio {
 307         refcount_t refs;
 308         atomic_t stripes_pending;
 309         struct btrfs_fs_info *fs_info;
 310         u64 map_type; /* get from map_lookup->type */
 311         bio_end_io_t *end_io;
 312         struct bio *orig_bio;
 313         unsigned long flags;
 314         void *private;
 315         atomic_t error;
 316         int max_errors;
 317         int num_stripes;
 318         int mirror_num;
 319         int num_tgtdevs;
 320         int *tgtdev_map;
 321         /*
 322          * logical block numbers for the start of each stripe
 323          * The last one or two are p/q.  These are sorted,
 324          * so raid_map[0] is the start of our full stripe
 325          */
 326         u64 *raid_map;
 327         struct btrfs_bio_stripe stripes[];
 328 };
 329
 330 struct btrfs_device_info {
 331         struct btrfs_device *dev;
 332         u64 dev_offset;
 333         u64 max_avail;
 334         u64 total_avail;
 335 };
 336
 337 struct btrfs_raid_attr {
 338         int sub_stripes;        /* sub_stripes info for map */
 339         int dev_stripes;        /* stripes per dev */
 340         int devs_max;           /* max devs to use */
 341         int devs_min;           /* min devs needed */
 342         int tolerated_failures; /* max tolerated fail devs */
 343         int devs_increment;     /* ndevs has to be a multiple of this */
 344         int ncopies;            /* how many copies to data has */
 345 };
 346
 347 extern const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES];
 348 extern const int btrfs_raid_mindev_error[BTRFS_NR_RAID_TYPES];
 349 extern const u64 btrfs_raid_group[BTRFS_NR_RAID_TYPES];
 350
 351 struct map_lookup {
 352         u64 type;
 353         int io_align;
 354         int io_width;
 355         u64 stripe_len;
 356         int sector_size;
 357         int num_stripes;
 358         int sub_stripes;
 359         struct btrfs_bio_stripe stripes[];
 360 };
 361
 362 #define map_lookup_size(n) (sizeof(struct map_lookup) + \
 363                             (sizeof(struct btrfs_bio_stripe) * (n)))
 364
 365 struct btrfs_balance_args;
 366 struct btrfs_balance_progress;
 367 struct btrfs_balance_control {
 368         struct btrfs_fs_info *fs_info;
 369
 370         struct btrfs_balance_args data;
 371         struct btrfs_balance_args meta;
 372         struct btrfs_balance_args sys;
 373
 374         u64 flags;
 375
 376         struct btrfs_balance_progress stat;
 377 };
 378
 379 enum btrfs_map_op {
 380         BTRFS_MAP_READ,
 381         BTRFS_MAP_WRITE,
 382         BTRFS_MAP_DISCARD,
 383         BTRFS_MAP_GET_READ_MIRRORS,
 384 };
 385
 386 static inline enum btrfs_map_op btrfs_op(struct bio *bio)
 387 {
 388         switch (bio_op(bio)) {
 389         case REQ_OP_DISCARD:
 390                 return BTRFS_MAP_DISCARD;
 391         case REQ_OP_WRITE:
 392                 return BTRFS_MAP_WRITE;
 393         default:
 394                 WARN_ON_ONCE(1);
 395         case REQ_OP_READ:
 396                 return BTRFS_MAP_READ;
 397         }
 398 }
 399
 400 int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start,
 401                                    u64 end, u64 *length);
 402 void btrfs_get_bbio(struct btrfs_bio *bbio);
 403 void btrfs_put_bbio(struct btrfs_bio *bbio);
 404 int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 405                     u64 logical, u64 *length,
 406                     struct btrfs_bio **bbio_ret, int mirror_num);
 407 int btrfs_map_sblock(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 408                      u64 logical, u64 *length,
 409                      struct btrfs_bio **bbio_ret);
 410 int btrfs_rmap_block(struct btrfs_fs_info *fs_info,
 411                      u64 chunk_start, u64 physical, u64 devid,
 412                      u64 **logical, int *naddrs, int *stripe_len);
 413 int btrfs_read_sys_array(struct btrfs_fs_info *fs_info);
 414 int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info);
 415 int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 416                       struct btrfs_fs_info *fs_info, u64 type);
 417 void btrfs_mapping_init(struct btrfs_mapping_tree *tree);
 418 void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree);
 419 int btrfs_map_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
 420                   int mirror_num, int async_submit);
 421 int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
 422                        fmode_t flags, void *holder);
 423 int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
 424                           struct btrfs_fs_devices **fs_devices_ret);
 425 int btrfs_close_devices(struct btrfs_fs_devices *fs_devices);
 426 void btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices, int step);
 427 void btrfs_assign_next_active_device(struct btrfs_fs_info *fs_info,
 428                 struct btrfs_device *device, struct btrfs_device *this_dev);
 429 int btrfs_find_device_missing_or_by_path(struct btrfs_fs_info *fs_info,
 430                                          const char *device_path,
 431                                          struct btrfs_device **device);
 432 int btrfs_find_device_by_devspec(struct btrfs_fs_info *fs_info, u64 devid,
 433                                          const char *devpath,
 434                                          struct btrfs_device **device);
 435 struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info,
 436                                         const u64 *devid,
 437                                         const u8 *uuid);
 438 int btrfs_rm_device(struct btrfs_fs_info *fs_info,
 439                     const char *device_path, u64 devid);
 440 void btrfs_cleanup_fs_uuids(void);
 441 int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len);
 442 int btrfs_grow_device(struct btrfs_trans_handle *trans,
 443                       struct btrfs_device *device, u64 new_size);
 444 struct btrfs_device *btrfs_find_device(struct btrfs_fs_info *fs_info, u64 devid,
 445                                        u8 *uuid, u8 *fsid);
 446 int btrfs_shrink_device(struct btrfs_device *device, u64 new_size);
 447 int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *path);
 448 int btrfs_init_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
 449                                   const char *device_path,
 450                                   struct btrfs_device *srcdev,
 451                                   struct btrfs_device **device_out);
 452 int btrfs_balance(struct btrfs_balance_control *bctl,
 453                   struct btrfs_ioctl_balance_args *bargs);
 454 int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info);
 455 int btrfs_recover_balance(struct btrfs_fs_info *fs_info);
 456 int btrfs_pause_balance(struct btrfs_fs_info *fs_info);
 457 int btrfs_cancel_balance(struct btrfs_fs_info *fs_info);
 458 int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info);
 459 int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info);
 460 int btrfs_chunk_readonly(struct btrfs_fs_info *fs_info, u64 chunk_offset);
 461 int find_free_dev_extent_start(struct btrfs_transaction *transaction,
 462                          struct btrfs_device *device, u64 num_bytes,
 463                          u64 search_start, u64 *start, u64 *max_avail);
 464 int find_free_dev_extent(struct btrfs_trans_handle *trans,
 465                          struct btrfs_device *device, u64 num_bytes,
 466                          u64 *start, u64 *max_avail);
 467 void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index);
 468 int btrfs_get_dev_stats(struct btrfs_fs_info *fs_info,
 469                         struct btrfs_ioctl_get_dev_stats *stats);
 470 void btrfs_init_devices_late(struct btrfs_fs_info *fs_info);
 471 int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info);
 472 int btrfs_run_dev_stats(struct btrfs_trans_handle *trans,
 473                         struct btrfs_fs_info *fs_info);
 474 void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_fs_info *fs_info,
 475                                         struct btrfs_device *srcdev);
 476 void btrfs_rm_dev_replace_free_srcdev(struct btrfs_fs_info *fs_info,
 477                                       struct btrfs_device *srcdev);
 478 void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
 479                                       struct btrfs_device *tgtdev);
 480 void btrfs_init_dev_replace_tgtdev_for_resume(struct btrfs_fs_info *fs_info,
 481                                               struct btrfs_device *tgtdev);
 482 void btrfs_scratch_superblocks(struct block_device *bdev, const char *device_path);
 483 int btrfs_is_parity_mirror(struct btrfs_fs_info *fs_info,
 484                            u64 logical, u64 len, int mirror_num);
 485 unsigned long btrfs_full_stripe_len(struct btrfs_fs_info *fs_info,
 486                                     struct btrfs_mapping_tree *map_tree,
 487                                     u64 logical);
 488 int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
 489                                 struct btrfs_fs_info *fs_info,
 490                                 u64 chunk_offset, u64 chunk_size);
 491 int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
 492                        struct btrfs_fs_info *fs_info, u64 chunk_offset);
 493
 494 static inline int btrfs_dev_stats_dirty(struct btrfs_device *dev)
 495 {
 496         return atomic_read(&dev->dev_stats_ccnt);
 497 }
 498
 499 static inline void btrfs_dev_stat_inc(struct btrfs_device *dev,
 500                                       int index)
 501 {
 502         atomic_inc(dev->dev_stat_values + index);
 503         smp_mb__before_atomic();
 504         atomic_inc(&dev->dev_stats_ccnt);
 505 }
 506
 507 static inline int btrfs_dev_stat_read(struct btrfs_device *dev,
 508                                       int index)
 509 {
 510         return atomic_read(dev->dev_stat_values + index);
 511 }
 512
 513 static inline int btrfs_dev_stat_read_and_reset(struct btrfs_device *dev,
 514                                                 int index)
 515 {
 516         int ret;
 517
 518         ret = atomic_xchg(dev->dev_stat_values + index, 0);
 519         smp_mb__before_atomic();
 520         atomic_inc(&dev->dev_stats_ccnt);
 521         return ret;
 522 }
 523
 524 static inline void btrfs_dev_stat_set(struct btrfs_device *dev,
 525                                       int index, unsigned long val)
 526 {
 527         atomic_set(dev->dev_stat_values + index, val);
 528         smp_mb__before_atomic();
 529         atomic_inc(&dev->dev_stats_ccnt);
 530 }
 531
 532 static inline void btrfs_dev_stat_reset(struct btrfs_device *dev,
 533                                         int index)
 534 {
 535         btrfs_dev_stat_set(dev, index, 0);
 536 }
 537
 538 void btrfs_update_commit_device_size(struct btrfs_fs_info *fs_info);
 539 void btrfs_update_commit_device_bytes_used(struct btrfs_fs_info *fs_info,
 540                                         struct btrfs_transaction *transaction);
 541
 542 struct list_head *btrfs_get_fs_uuids(void);
 543 void btrfs_set_fs_info_ptr(struct btrfs_fs_info *fs_info);
 544 void btrfs_reset_fs_info_ptr(struct btrfs_fs_info *fs_info);
 545
 546 #endif