1/* SPDX-License-Identifier: GPL-2.0 */ 2/* 3 * Copyright (C) 2007 Oracle. All rights reserved. 4 */ 5 6#ifndef BTRFS_VOLUMES_H 7#define BTRFS_VOLUMES_H 8 9#include <linux/bio.h> 10#include <linux/sort.h> 11#include <linux/btrfs.h> 12#include "async-thread.h" 13 14#define BTRFS_MAX_DATA_CHUNK_SIZE (10ULL * SZ_1G) 15 16extern struct mutex uuid_mutex; 17 18#define BTRFS_STRIPE_LEN SZ_64K 19 20struct btrfs_io_geometry { 21 /* remaining bytes before crossing a stripe */ 22 u64 len; 23 /* offset of logical address in chunk */ 24 u64 offset; 25 /* length of single IO stripe */ 26 u64 stripe_len; 27 /* number of stripe where address falls */ 28 u64 stripe_nr; 29 /* offset of address in stripe */ 30 u64 stripe_offset; 31 /* offset of raid56 stripe into the chunk */ 32 u64 raid56_stripe_offset; 33}; 34 35/* 36 * Use sequence counter to get consistent device stat data on 37 * 32-bit processors. 38 */ 39#if BITS_PER_LONG==32 && defined(CONFIG_SMP) 40#include <linux/seqlock.h> 41#define __BTRFS_NEED_DEVICE_DATA_ORDERED 42#define btrfs_device_data_ordered_init(device) \ 43 seqcount_init(&device->data_seqcount) 44#else 45#define btrfs_device_data_ordered_init(device) do { } while (0) 46#endif 47 48#define BTRFS_DEV_STATE_WRITEABLE (0) 49#define BTRFS_DEV_STATE_IN_FS_METADATA (1) 50#define BTRFS_DEV_STATE_MISSING (2) 51#define BTRFS_DEV_STATE_REPLACE_TGT (3) 52#define BTRFS_DEV_STATE_FLUSH_SENT (4) 53#define BTRFS_DEV_STATE_NO_READA (5) 54 55struct btrfs_device { 56 struct list_head dev_list; /* device_list_mutex */ 57 struct list_head dev_alloc_list; /* chunk mutex */ 58 struct list_head post_commit_list; /* chunk mutex */ 59 struct btrfs_fs_devices *fs_devices; 60 struct btrfs_fs_info *fs_info; 61 62 struct rcu_string __rcu *name; 63 64 u64 generation; 65 66 struct block_device *bdev; 67 68 /* the mode sent to blkdev_get */ 69 fmode_t mode; 70 71 unsigned long dev_state; 72 blk_status_t last_flush_error; 73 74#ifdef __BTRFS_NEED_DEVICE_DATA_ORDERED 75 seqcount_t data_seqcount; 76#endif 77 78 /* the internal btrfs device id */ 79 u64 devid; 80 81 /* size of the device in memory */ 82 u64 total_bytes; 83 84 /* size of the device on disk */ 85 u64 disk_total_bytes; 86 87 /* bytes used */ 88 u64 bytes_used; 89 90 /* optimal io alignment for this device */ 91 u32 io_align; 92 93 /* optimal io width for this device */ 94 u32 io_width; 95 /* type and info about this device */ 96 u64 type; 97 98 /* minimal io size for this device */ 99 u32 sector_size; 100 101 /* physical drive uuid (or lvm uuid) */ 102 u8 uuid[BTRFS_UUID_SIZE]; 103 104 /* 105 * size of the device on the current transaction 106 * 107 * This variant is update when committing the transaction, 108 * and protected by chunk mutex 109 */ 110 u64 commit_total_bytes; 111 112 /* bytes used on the current transaction */ 113 u64 commit_bytes_used; 114 115 /* for sending down flush barriers */ 116 struct bio *flush_bio; 117 struct completion flush_wait; 118 119 /* per-device scrub information */ 120 struct scrub_ctx *scrub_ctx; 121 122 /* readahead state */ 123 atomic_t reada_in_flight; 124 u64 reada_next; 125 struct reada_zone *reada_curr_zone; 126 struct radix_tree_root reada_zones; 127 struct radix_tree_root reada_extents; 128 129 /* disk I/O failure stats. For detailed description refer to 130 * enum btrfs_dev_stat_values in ioctl.h */ 131 int dev_stats_valid; 132 133 /* Counter to record the change of device stats */ 134 atomic_t dev_stats_ccnt; 135 atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX]; 136 137 struct extent_io_tree alloc_state; 138 139 struct completion kobj_unregister; 140 /* For sysfs/FSID/devinfo/devid/ */ 141 struct kobject devid_kobj; 142}; 143 144/* 145 * If we read those variants at the context of their own lock, we needn't 146 * use the following helpers, reading them directly is safe. 147 */ 148#if BITS_PER_LONG==32 && defined(CONFIG_SMP) 149#define BTRFS_DEVICE_GETSET_FUNCS(name) \ 150static inline u64 \ 151btrfs_device_get_##name(const struct btrfs_device *dev) \ 152{ \ 153 u64 size; \ 154 unsigned int seq; \ 155 \ 156 do { \ 157 seq = read_seqcount_begin(&dev->data_seqcount); \ 158 size = dev->name; \ 159 } while (read_seqcount_retry(&dev->data_seqcount, seq)); \ 160 return size; \ 161} \ 162 \ 163static inline void \ 164btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \ 165{ \ 166 preempt_disable(); \ 167 write_seqcount_begin(&dev->data_seqcount); \ 168 dev->name = size; \ 169 write_seqcount_end(&dev->data_seqcount); \ 170 preempt_enable(); \ 171} 172#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION) 173#define BTRFS_DEVICE_GETSET_FUNCS(name) \ 174static inline u64 \ 175btrfs_device_get_##name(const struct btrfs_device *dev) \ 176{ \ 177 u64 size; \ 178 \ 179 preempt_disable(); \ 180 size = dev->name; \ 181 preempt_enable(); \ 182 return size; \ 183} \ 184 \ 185static inline void \ 186btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \ 187{ \ 188 preempt_disable(); \ 189 dev->name = size; \ 190 preempt_enable(); \ 191} 192#else 193#define BTRFS_DEVICE_GETSET_FUNCS(name) \ 194static inline u64 \ 195btrfs_device_get_##name(const struct btrfs_device *dev) \ 196{ \ 197 return dev->name; \ 198} \ 199 \ 200static inline void \ 201btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \ 202{ \ 203 dev->name = size; \ 204} 205#endif 206 207BTRFS_DEVICE_GETSET_FUNCS(total_bytes); 208BTRFS_DEVICE_GETSET_FUNCS(disk_total_bytes); 209BTRFS_DEVICE_GETSET_FUNCS(bytes_used); 210 211enum btrfs_chunk_allocation_policy { 212 BTRFS_CHUNK_ALLOC_REGULAR, 213}; 214 215struct btrfs_fs_devices { 216 u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */ 217 u8 metadata_uuid[BTRFS_FSID_SIZE]; 218 bool fsid_change; 219 struct list_head fs_list; 220 221 u64 num_devices; 222 u64 open_devices; 223 u64 rw_devices; 224 u64 missing_devices; 225 u64 total_rw_bytes; 226 u64 total_devices; 227 228 /* Highest generation number of seen devices */ 229 u64 latest_generation; 230 231 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; 240 241 /* List of all devices, protected by device_list_mutex */ 242 struct list_head devices; 243 244 /* 245 * Devices which can satisfy space allocation. Protected by 246 * chunk_mutex 247 */ 248 struct list_head alloc_list; 249 250 struct list_head seed_list; 251 bool seeding; 252 253 int opened; 254 255 /* set when we find or add a device that doesn't have the 256 * nonrot flag set 257 */ 258 bool rotating; 259 260 struct btrfs_fs_info *fs_info; 261 /* sysfs kobjects */ 262 struct kobject fsid_kobj; 263 struct kobject *devices_kobj; 264 struct kobject *devinfo_kobj; 265 struct completion kobj_unregister; 266 267 enum btrfs_chunk_allocation_policy chunk_alloc_policy; 268}; 269 270#define BTRFS_BIO_INLINE_CSUM_SIZE 64 271 272#define BTRFS_MAX_DEVS(info) ((BTRFS_MAX_ITEM_SIZE(info) \ 273 - sizeof(struct btrfs_chunk)) \ 274 / sizeof(struct btrfs_stripe) + 1) 275 276#define BTRFS_MAX_DEVS_SYS_CHUNK ((BTRFS_SYSTEM_CHUNK_ARRAY_SIZE \ 277 - 2 * sizeof(struct btrfs_disk_key) \ 278 - 2 * sizeof(struct btrfs_chunk)) \ 279 / sizeof(struct btrfs_stripe) + 1) 280 281/* 282 * we need the mirror number and stripe index to be passed around 283 * the call chain while we are processing end_io (especially errors). 284 * Really, what we need is a btrfs_bio structure that has this info 285 * and is properly sized with its stripe array, but we're not there 286 * quite yet. We have our own btrfs bioset, and all of the bios 287 * we allocate are actually btrfs_io_bios. We'll cram as much of 288 * struct btrfs_bio as we can into this over time. 289 */ 290struct btrfs_io_bio { 291 unsigned int mirror_num; 292 struct btrfs_device *device; 293 u64 logical; 294 u8 *csum; 295 u8 csum_inline[BTRFS_BIO_INLINE_CSUM_SIZE]; 296 struct bvec_iter iter; 297 /* 298 * This member must come last, bio_alloc_bioset will allocate enough 299 * bytes for entire btrfs_io_bio but relies on bio being last. 300 */ 301 struct bio bio; 302}; 303 304static inline struct btrfs_io_bio *btrfs_io_bio(struct bio *bio) 305{ 306 return container_of(bio, struct btrfs_io_bio, bio); 307} 308 309static inline void btrfs_io_bio_free_csum(struct btrfs_io_bio *io_bio) 310{ 311 if (io_bio->csum != io_bio->csum_inline) { 312 kfree(io_bio->csum); 313 io_bio->csum = NULL; 314 } 315} 316 317struct btrfs_bio_stripe { 318 struct btrfs_device *dev; 319 u64 physical; 320 u64 length; /* only used for discard mappings */ 321}; 322 323struct btrfs_bio { 324 refcount_t refs; 325 atomic_t stripes_pending; 326 struct btrfs_fs_info *fs_info; 327 u64 map_type; /* get from map_lookup->type */ 328 bio_end_io_t *end_io; 329 struct bio *orig_bio; 330 void *private; 331 atomic_t error; 332 int max_errors; 333 int num_stripes; 334 int mirror_num; 335 int num_tgtdevs; 336 int *tgtdev_map; 337 /* 338 * logical block numbers for the start of each stripe 339 * The last one or two are p/q. These are sorted, 340 * so raid_map[0] is the start of our full stripe 341 */ 342 u64 *raid_map; 343 struct btrfs_bio_stripe stripes[]; 344}; 345 346struct btrfs_device_info { 347 struct btrfs_device *dev; 348 u64 dev_offset; 349 u64 max_avail; 350 u64 total_avail; 351}; 352 353struct btrfs_raid_attr { 354 u8 sub_stripes; /* sub_stripes info for map */ 355 u8 dev_stripes; /* stripes per dev */ 356 u8 devs_max; /* max devs to use */ 357 u8 devs_min; /* min devs needed */ 358 u8 tolerated_failures; /* max tolerated fail devs */ 359 u8 devs_increment; /* ndevs has to be a multiple of this */ 360 u8 ncopies; /* how many copies to data has */ 361 u8 nparity; /* number of stripes worth of bytes to store 362 * parity information */ 363 u8 mindev_error; /* error code if min devs requisite is unmet */ 364 const char raid_name[8]; /* name of the raid */ 365 u64 bg_flag; /* block group flag of the raid */ 366}; 367 368extern const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES]; 369 370struct map_lookup { 371 u64 type; 372 int io_align; 373 int io_width; 374 u64 stripe_len; 375 int num_stripes; 376 int sub_stripes; 377 int verified_stripes; /* For mount time dev extent verification */ 378 struct btrfs_bio_stripe stripes[]; 379}; 380 381#define map_lookup_size(n) (sizeof(struct map_lookup) + \ 382 (sizeof(struct btrfs_bio_stripe) * (n))) 383 384struct btrfs_balance_args; 385struct btrfs_balance_progress; 386struct btrfs_balance_control { 387 struct btrfs_balance_args data; 388 struct btrfs_balance_args meta; 389 struct btrfs_balance_args sys; 390 391 u64 flags; 392 393 struct btrfs_balance_progress stat; 394}; 395 396enum btrfs_map_op { 397 BTRFS_MAP_READ, 398 BTRFS_MAP_WRITE, 399 BTRFS_MAP_DISCARD, 400 BTRFS_MAP_GET_READ_MIRRORS, 401}; 402 403static inline enum btrfs_map_op btrfs_op(struct bio *bio) 404{ 405 switch (bio_op(bio)) { 406 case REQ_OP_DISCARD: 407 return BTRFS_MAP_DISCARD; 408 case REQ_OP_WRITE: 409 return BTRFS_MAP_WRITE; 410 default: 411 WARN_ON_ONCE(1); 412 fallthrough; 413 case REQ_OP_READ: 414 return BTRFS_MAP_READ; 415 } 416} 417 418void btrfs_get_bbio(struct btrfs_bio *bbio); 419void btrfs_put_bbio(struct btrfs_bio *bbio); 420int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op, 421 u64 logical, u64 *length, 422 struct btrfs_bio **bbio_ret, int mirror_num); 423int btrfs_map_sblock(struct btrfs_fs_info *fs_info, enum btrfs_map_op op, 424 u64 logical, u64 *length, 425 struct btrfs_bio **bbio_ret); 426int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, enum btrfs_map_op op, 427 u64 logical, u64 len, struct btrfs_io_geometry *io_geom); 428int btrfs_read_sys_array(struct btrfs_fs_info *fs_info); 429int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info); 430int btrfs_alloc_chunk(struct btrfs_trans_handle *trans, u64 type); 431void btrfs_mapping_tree_free(struct extent_map_tree *tree); 432blk_status_t btrfs_map_bio(struct btrfs_fs_info *fs_info, struct bio *bio, 433 int mirror_num); 434int btrfs_open_devices(struct btrfs_fs_devices *fs_devices, 435 fmode_t flags, void *holder); 436struct btrfs_device *btrfs_scan_one_device(const char *path, 437 fmode_t flags, void *holder); 438int btrfs_forget_devices(const char *path); 439void btrfs_close_devices(struct btrfs_fs_devices *fs_devices); 440void btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices, int step); 441void btrfs_assign_next_active_device(struct btrfs_device *device, 442 struct btrfs_device *this_dev); 443struct btrfs_device *btrfs_find_device_by_devspec(struct btrfs_fs_info *fs_info, 444 u64 devid, 445 const char *devpath); 446struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info, 447 const u64 *devid, 448 const u8 *uuid); 449void btrfs_free_device(struct btrfs_device *device); 450int btrfs_rm_device(struct btrfs_fs_info *fs_info, 451 const char *device_path, u64 devid); 452void __exit btrfs_cleanup_fs_uuids(void); 453int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len); 454int btrfs_grow_device(struct btrfs_trans_handle *trans, 455 struct btrfs_device *device, u64 new_size); 456struct btrfs_device *btrfs_find_device(struct btrfs_fs_devices *fs_devices, 457 u64 devid, u8 *uuid, u8 *fsid, bool seed); 458int btrfs_shrink_device(struct btrfs_device *device, u64 new_size); 459int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *path); 460int btrfs_balance(struct btrfs_fs_info *fs_info, 461 struct btrfs_balance_control *bctl, 462 struct btrfs_ioctl_balance_args *bargs); 463void btrfs_describe_block_groups(u64 flags, char *buf, u32 size_buf); 464int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info); 465int btrfs_recover_balance(struct btrfs_fs_info *fs_info); 466int btrfs_pause_balance(struct btrfs_fs_info *fs_info); 467int btrfs_cancel_balance(struct btrfs_fs_info *fs_info); 468int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info); 469int btrfs_uuid_scan_kthread(void *data); 470int btrfs_chunk_readonly(struct btrfs_fs_info *fs_info, u64 chunk_offset); 471int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes, 472 u64 *start, u64 *max_avail); 473void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index); 474int btrfs_get_dev_stats(struct btrfs_fs_info *fs_info, 475 struct btrfs_ioctl_get_dev_stats *stats); 476void btrfs_init_devices_late(struct btrfs_fs_info *fs_info); 477int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info); 478int btrfs_run_dev_stats(struct btrfs_trans_handle *trans); 479void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_device *srcdev); 480void btrfs_rm_dev_replace_free_srcdev(struct btrfs_device *srcdev); 481void btrfs_destroy_dev_replace_tgtdev(struct btrfs_device *tgtdev); 482int btrfs_is_parity_mirror(struct btrfs_fs_info *fs_info, 483 u64 logical, u64 len); 484unsigned long btrfs_full_stripe_len(struct btrfs_fs_info *fs_info, 485 u64 logical); 486int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans, 487 u64 chunk_offset, u64 chunk_size); 488int btrfs_remove_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset); 489struct extent_map *btrfs_get_chunk_map(struct btrfs_fs_info *fs_info, 490 u64 logical, u64 length); 491void btrfs_release_disk_super(struct btrfs_super_block *super); 492 493static inline void btrfs_dev_stat_inc(struct btrfs_device *dev, 494 int index) 495{ 496 atomic_inc(dev->dev_stat_values + index); 497 /* 498 * This memory barrier orders stores updating statistics before stores 499 * updating dev_stats_ccnt. 500 * 501 * It pairs with smp_rmb() in btrfs_run_dev_stats(). 502 */ 503 smp_mb__before_atomic(); 504 atomic_inc(&dev->dev_stats_ccnt); 505} 506 507static 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 513static 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 /* 520 * atomic_xchg implies a full memory barriers as per atomic_t.txt: 521 * - RMW operations that have a return value are fully ordered; 522 * 523 * This implicit memory barriers is paired with the smp_rmb in 524 * btrfs_run_dev_stats 525 */ 526 atomic_inc(&dev->dev_stats_ccnt); 527 return ret; 528} 529 530static inline void btrfs_dev_stat_set(struct btrfs_device *dev, 531 int index, unsigned long val) 532{ 533 atomic_set(dev->dev_stat_values + index, val); 534 /* 535 * This memory barrier orders stores updating statistics before stores 536 * updating dev_stats_ccnt. 537 * 538 * It pairs with smp_rmb() in btrfs_run_dev_stats(). 539 */ 540 smp_mb__before_atomic(); 541 atomic_inc(&dev->dev_stats_ccnt); 542} 543 544/* 545 * Convert block group flags (BTRFS_BLOCK_GROUP_*) to btrfs_raid_types, which 546 * can be used as index to access btrfs_raid_array[]. 547 */ 548static inline enum btrfs_raid_types btrfs_bg_flags_to_raid_index(u64 flags) 549{ 550 if (flags & BTRFS_BLOCK_GROUP_RAID10) 551 return BTRFS_RAID_RAID10; 552 else if (flags & BTRFS_BLOCK_GROUP_RAID1) 553 return BTRFS_RAID_RAID1; 554 else if (flags & BTRFS_BLOCK_GROUP_RAID1C3) 555 return BTRFS_RAID_RAID1C3; 556 else if (flags & BTRFS_BLOCK_GROUP_RAID1C4) 557 return BTRFS_RAID_RAID1C4; 558 else if (flags & BTRFS_BLOCK_GROUP_DUP) 559 return BTRFS_RAID_DUP; 560 else if (flags & BTRFS_BLOCK_GROUP_RAID0) 561 return BTRFS_RAID_RAID0; 562 else if (flags & BTRFS_BLOCK_GROUP_RAID5) 563 return BTRFS_RAID_RAID5; 564 else if (flags & BTRFS_BLOCK_GROUP_RAID6) 565 return BTRFS_RAID_RAID6; 566 567 return BTRFS_RAID_SINGLE; /* BTRFS_BLOCK_GROUP_SINGLE */ 568} 569 570void btrfs_commit_device_sizes(struct btrfs_transaction *trans); 571 572struct list_head * __attribute_const__ btrfs_get_fs_uuids(void); 573bool btrfs_check_rw_degradable(struct btrfs_fs_info *fs_info, 574 struct btrfs_device *failing_dev); 575void btrfs_scratch_superblocks(struct btrfs_fs_info *fs_info, 576 struct block_device *bdev, 577 const char *device_path); 578 579int btrfs_bg_type_to_factor(u64 flags); 580const char *btrfs_bg_type_to_raid_name(u64 flags); 581int btrfs_verify_dev_extents(struct btrfs_fs_info *fs_info); 582 583bool btrfs_pinned_by_swapfile(struct btrfs_fs_info *fs_info, void *ptr); 584u8 *btrfs_sb_fsid_ptr(struct btrfs_super_block *sb); 585 586#endif 587