fs/btrfs/delayed-ref.h

   1 /* SPDX-License-Identifier: GPL-2.0 */
   2 /*
   3  * Copyright (C) 2008 Oracle.  All rights reserved.
   4  */
   5
   6 #ifndef BTRFS_DELAYED_REF_H
   7 #define BTRFS_DELAYED_REF_H
   8
   9 #include <linux/refcount.h>
  10
  11 /* these are the possible values of struct btrfs_delayed_ref_node->action */
  12 #define BTRFS_ADD_DELAYED_REF    1 /* add one backref to the tree */
  13 #define BTRFS_DROP_DELAYED_REF   2 /* delete one backref from the tree */
  14 #define BTRFS_ADD_DELAYED_EXTENT 3 /* record a full extent allocation */
  15 #define BTRFS_UPDATE_DELAYED_HEAD 4 /* not changing ref count on head ref */
  16
  17 struct btrfs_delayed_ref_node {
  18         struct rb_node ref_node;
  19         /*
  20          * If action is BTRFS_ADD_DELAYED_REF, also link this node to
  21          * ref_head->ref_add_list, then we do not need to iterate the
  22          * whole ref_head->ref_list to find BTRFS_ADD_DELAYED_REF nodes.
  23          */
  24         struct list_head add_list;
  25
  26         /* the starting bytenr of the extent */
  27         u64 bytenr;
  28
  29         /* the size of the extent */
  30         u64 num_bytes;
  31
  32         /* seq number to keep track of insertion order */
  33         u64 seq;
  34
  35         /* ref count on this data structure */
  36         refcount_t refs;
  37
  38         /*
  39          * how many refs is this entry adding or deleting.  For
  40          * head refs, this may be a negative number because it is keeping
  41          * track of the total mods done to the reference count.
  42          * For individual refs, this will always be a positive number
  43          *
  44          * It may be more than one, since it is possible for a single
  45          * parent to have more than one ref on an extent
  46          */
  47         int ref_mod;
  48
  49         unsigned int action:8;
  50         unsigned int type:8;
  51 };
  52
  53 struct btrfs_delayed_extent_op {
  54         struct btrfs_disk_key key;
  55         u8 level;
  56         bool update_key;
  57         bool update_flags;
  58         u64 flags_to_set;
  59 };
  60
  61 /*
  62  * the head refs are used to hold a lock on a given extent, which allows us
  63  * to make sure that only one process is running the delayed refs
  64  * at a time for a single extent.  They also store the sum of all the
  65  * reference count modifications we've queued up.
  66  */
  67 struct btrfs_delayed_ref_head {
  68         u64 bytenr;
  69         u64 num_bytes;
  70         /*
  71          * For insertion into struct btrfs_delayed_ref_root::href_root.
  72          * Keep it in the same cache line as 'bytenr' for more efficient
  73          * searches in the rbtree.
  74          */
  75         struct rb_node href_node;
  76         /*
  77          * the mutex is held while running the refs, and it is also
  78          * held when checking the sum of reference modifications.
  79          */
  80         struct mutex mutex;
  81
  82         refcount_t refs;
  83
  84         /* Protects 'ref_tree' and 'ref_add_list'. */
  85         spinlock_t lock;
  86         struct rb_root_cached ref_tree;
  87         /* accumulate add BTRFS_ADD_DELAYED_REF nodes to this ref_add_list. */
  88         struct list_head ref_add_list;
  89
  90         struct btrfs_delayed_extent_op *extent_op;
  91
  92         /*
  93          * This is used to track the final ref_mod from all the refs associated
  94          * with this head ref, this is not adjusted as delayed refs are run,
  95          * this is meant to track if we need to do the csum accounting or not.
  96          */
  97         int total_ref_mod;
  98
  99         /*
 100          * This is the current outstanding mod references for this bytenr.  This
 101          * is used with lookup_extent_info to get an accurate reference count
 102          * for a bytenr, so it is adjusted as delayed refs are run so that any
 103          * on disk reference count + ref_mod is accurate.
 104          */
 105         int ref_mod;
 106
 107         /*
 108          * when a new extent is allocated, it is just reserved in memory
 109          * The actual extent isn't inserted into the extent allocation tree
 110          * until the delayed ref is processed.  must_insert_reserved is
 111          * used to flag a delayed ref so the accounting can be updated
 112          * when a full insert is done.
 113          *
 114          * It is possible the extent will be freed before it is ever
 115          * inserted into the extent allocation tree.  In this case
 116          * we need to update the in ram accounting to properly reflect
 117          * the free has happened.
 118          */
 119         bool must_insert_reserved;
 120         bool is_data;
 121         bool is_system;
 122         bool processing;
 123 };
 124
 125 struct btrfs_delayed_tree_ref {
 126         struct btrfs_delayed_ref_node node;
 127         u64 root;
 128         u64 parent;
 129         int level;
 130 };
 131
 132 struct btrfs_delayed_data_ref {
 133         struct btrfs_delayed_ref_node node;
 134         u64 root;
 135         u64 parent;
 136         u64 objectid;
 137         u64 offset;
 138 };
 139
 140 enum btrfs_delayed_ref_flags {
 141         /* Indicate that we are flushing delayed refs for the commit */
 142         BTRFS_DELAYED_REFS_FLUSHING,
 143 };
 144
 145 struct btrfs_delayed_ref_root {
 146         /* head ref rbtree */
 147         struct rb_root_cached href_root;
 148
 149         /* dirty extent records */
 150         struct rb_root dirty_extent_root;
 151
 152         /* this spin lock protects the rbtree and the entries inside */
 153         spinlock_t lock;
 154
 155         /* how many delayed ref updates we've queued, used by the
 156          * throttling code
 157          */
 158         atomic_t num_entries;
 159
 160         /* total number of head nodes in tree */
 161         unsigned long num_heads;
 162
 163         /* total number of head nodes ready for processing */
 164         unsigned long num_heads_ready;
 165
 166         u64 pending_csums;
 167
 168         unsigned long flags;
 169
 170         u64 run_delayed_start;
 171
 172         /*
 173          * To make qgroup to skip given root.
 174          * This is for snapshot, as btrfs_qgroup_inherit() will manually
 175          * modify counters for snapshot and its source, so we should skip
 176          * the snapshot in new_root/old_roots or it will get calculated twice
 177          */
 178         u64 qgroup_to_skip;
 179 };
 180
 181 enum btrfs_ref_type {
 182         BTRFS_REF_NOT_SET,
 183         BTRFS_REF_DATA,
 184         BTRFS_REF_METADATA,
 185         BTRFS_REF_LAST,
 186 };
 187
 188 struct btrfs_data_ref {
 189         /* For EXTENT_DATA_REF */
 190
 191         /* Original root this data extent belongs to */
 192         u64 owning_root;
 193
 194         /* Inode which refers to this data extent */
 195         u64 ino;
 196
 197         /*
 198          * file_offset - extent_offset
 199          *
 200          * file_offset is the key.offset of the EXTENT_DATA key.
 201          * extent_offset is btrfs_file_extent_offset() of the EXTENT_DATA data.
 202          */
 203         u64 offset;
 204 };
 205
 206 struct btrfs_tree_ref {
 207         /*
 208          * Level of this tree block
 209          *
 210          * Shared for skinny (TREE_BLOCK_REF) and normal tree ref.
 211          */
 212         int level;
 213
 214         /*
 215          * Root which owns this tree block.
 216          *
 217          * For TREE_BLOCK_REF (skinny metadata, either inline or keyed)
 218          */
 219         u64 owning_root;
 220
 221         /* For non-skinny metadata, no special member needed */
 222 };
 223
 224 struct btrfs_ref {
 225         enum btrfs_ref_type type;
 226         int action;
 227
 228         /*
 229          * Whether this extent should go through qgroup record.
 230          *
 231          * Normally false, but for certain cases like delayed subtree scan,
 232          * setting this flag can hugely reduce qgroup overhead.
 233          */
 234         bool skip_qgroup;
 235
 236 #ifdef CONFIG_BTRFS_FS_REF_VERIFY
 237         /* Through which root is this modification. */
 238         u64 real_root;
 239 #endif
 240         u64 bytenr;
 241         u64 len;
 242
 243         /* Bytenr of the parent tree block */
 244         u64 parent;
 245         union {
 246                 struct btrfs_data_ref data_ref;
 247                 struct btrfs_tree_ref tree_ref;
 248         };
 249 };
 250
 251 extern struct kmem_cache *btrfs_delayed_ref_head_cachep;
 252 extern struct kmem_cache *btrfs_delayed_tree_ref_cachep;
 253 extern struct kmem_cache *btrfs_delayed_data_ref_cachep;
 254 extern struct kmem_cache *btrfs_delayed_extent_op_cachep;
 255
 256 int __init btrfs_delayed_ref_init(void);
 257 void __cold btrfs_delayed_ref_exit(void);
 258
 259 static inline u64 btrfs_calc_delayed_ref_bytes(const struct btrfs_fs_info *fs_info,
 260                                                int num_delayed_refs)
 261 {
 262         u64 num_bytes;
 263
 264         num_bytes = btrfs_calc_insert_metadata_size(fs_info, num_delayed_refs);
 265
 266         /*
 267          * We have to check the mount option here because we could be enabling
 268          * the free space tree for the first time and don't have the compat_ro
 269          * option set yet.
 270          *
 271          * We need extra reservations if we have the free space tree because
 272          * we'll have to modify that tree as well.
 273          */
 274         if (btrfs_test_opt(fs_info, FREE_SPACE_TREE))
 275                 num_bytes *= 2;
 276
 277         return num_bytes;
 278 }
 279
 280 static inline void btrfs_init_generic_ref(struct btrfs_ref *generic_ref,
 281                                 int action, u64 bytenr, u64 len, u64 parent)
 282 {
 283         generic_ref->action = action;
 284         generic_ref->bytenr = bytenr;
 285         generic_ref->len = len;
 286         generic_ref->parent = parent;
 287 }
 288
 289 static inline void btrfs_init_tree_ref(struct btrfs_ref *generic_ref,
 290                                 int level, u64 root, u64 mod_root, bool skip_qgroup)
 291 {
 292 #ifdef CONFIG_BTRFS_FS_REF_VERIFY
 293         /* If @real_root not set, use @root as fallback */
 294         generic_ref->real_root = mod_root ?: root;
 295 #endif
 296         generic_ref->tree_ref.level = level;
 297         generic_ref->tree_ref.owning_root = root;
 298         generic_ref->type = BTRFS_REF_METADATA;
 299         if (skip_qgroup || !(is_fstree(root) &&
 300                              (!mod_root || is_fstree(mod_root))))
 301                 generic_ref->skip_qgroup = true;
 302         else
 303                 generic_ref->skip_qgroup = false;
 304
 305 }
 306
 307 static inline void btrfs_init_data_ref(struct btrfs_ref *generic_ref,
 308                                 u64 ref_root, u64 ino, u64 offset, u64 mod_root,
 309                                 bool skip_qgroup)
 310 {
 311 #ifdef CONFIG_BTRFS_FS_REF_VERIFY
 312         /* If @real_root not set, use @root as fallback */
 313         generic_ref->real_root = mod_root ?: ref_root;
 314 #endif
 315         generic_ref->data_ref.owning_root = ref_root;
 316         generic_ref->data_ref.ino = ino;
 317         generic_ref->data_ref.offset = offset;
 318         generic_ref->type = BTRFS_REF_DATA;
 319         if (skip_qgroup || !(is_fstree(ref_root) &&
 320                              (!mod_root || is_fstree(mod_root))))
 321                 generic_ref->skip_qgroup = true;
 322         else
 323                 generic_ref->skip_qgroup = false;
 324 }
 325
 326 static inline struct btrfs_delayed_extent_op *
 327 btrfs_alloc_delayed_extent_op(void)
 328 {
 329         return kmem_cache_alloc(btrfs_delayed_extent_op_cachep, GFP_NOFS);
 330 }
 331
 332 static inline void
 333 btrfs_free_delayed_extent_op(struct btrfs_delayed_extent_op *op)
 334 {
 335         if (op)
 336                 kmem_cache_free(btrfs_delayed_extent_op_cachep, op);
 337 }
 338
 339 static inline void btrfs_put_delayed_ref(struct btrfs_delayed_ref_node *ref)
 340 {
 341         WARN_ON(refcount_read(&ref->refs) == 0);
 342         if (refcount_dec_and_test(&ref->refs)) {
 343                 WARN_ON(!RB_EMPTY_NODE(&ref->ref_node));
 344                 switch (ref->type) {
 345                 case BTRFS_TREE_BLOCK_REF_KEY:
 346                 case BTRFS_SHARED_BLOCK_REF_KEY:
 347                         kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref);
 348                         break;
 349                 case BTRFS_EXTENT_DATA_REF_KEY:
 350                 case BTRFS_SHARED_DATA_REF_KEY:
 351                         kmem_cache_free(btrfs_delayed_data_ref_cachep, ref);
 352                         break;
 353                 default:
 354                         BUG();
 355                 }
 356         }
 357 }
 358
 359 static inline u64 btrfs_ref_head_to_space_flags(
 360                                 struct btrfs_delayed_ref_head *head_ref)
 361 {
 362         if (head_ref->is_data)
 363                 return BTRFS_BLOCK_GROUP_DATA;
 364         else if (head_ref->is_system)
 365                 return BTRFS_BLOCK_GROUP_SYSTEM;
 366         return BTRFS_BLOCK_GROUP_METADATA;
 367 }
 368
 369 static inline void btrfs_put_delayed_ref_head(struct btrfs_delayed_ref_head *head)
 370 {
 371         if (refcount_dec_and_test(&head->refs))
 372                 kmem_cache_free(btrfs_delayed_ref_head_cachep, head);
 373 }
 374
 375 int btrfs_add_delayed_tree_ref(struct btrfs_trans_handle *trans,
 376                                struct btrfs_ref *generic_ref,
 377                                struct btrfs_delayed_extent_op *extent_op);
 378 int btrfs_add_delayed_data_ref(struct btrfs_trans_handle *trans,
 379                                struct btrfs_ref *generic_ref,
 380                                u64 reserved);
 381 int btrfs_add_delayed_extent_op(struct btrfs_trans_handle *trans,
 382                                 u64 bytenr, u64 num_bytes,
 383                                 struct btrfs_delayed_extent_op *extent_op);
 384 void btrfs_merge_delayed_refs(struct btrfs_fs_info *fs_info,
 385                               struct btrfs_delayed_ref_root *delayed_refs,
 386                               struct btrfs_delayed_ref_head *head);
 387
 388 struct btrfs_delayed_ref_head *
 389 btrfs_find_delayed_ref_head(struct btrfs_delayed_ref_root *delayed_refs,
 390                             u64 bytenr);
 391 int btrfs_delayed_ref_lock(struct btrfs_delayed_ref_root *delayed_refs,
 392                            struct btrfs_delayed_ref_head *head);
 393 static inline void btrfs_delayed_ref_unlock(struct btrfs_delayed_ref_head *head)
 394 {
 395         mutex_unlock(&head->mutex);
 396 }
 397 void btrfs_delete_ref_head(struct btrfs_delayed_ref_root *delayed_refs,
 398                            struct btrfs_delayed_ref_head *head);
 399
 400 struct btrfs_delayed_ref_head *btrfs_select_ref_head(
 401                 struct btrfs_delayed_ref_root *delayed_refs);
 402
 403 int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info, u64 seq);
 404
 405 void btrfs_delayed_refs_rsv_release(struct btrfs_fs_info *fs_info, int nr);
 406 void btrfs_update_delayed_refs_rsv(struct btrfs_trans_handle *trans);
 407 int btrfs_delayed_refs_rsv_refill(struct btrfs_fs_info *fs_info,
 408                                   enum btrfs_reserve_flush_enum flush);
 409 void btrfs_migrate_to_delayed_refs_rsv(struct btrfs_fs_info *fs_info,
 410                                        u64 num_bytes);
 411 bool btrfs_check_space_for_delayed_refs(struct btrfs_fs_info *fs_info);
 412
 413 /*
 414  * helper functions to cast a node into its container
 415  */
 416 static inline struct btrfs_delayed_tree_ref *
 417 btrfs_delayed_node_to_tree_ref(struct btrfs_delayed_ref_node *node)
 418 {
 419         return container_of(node, struct btrfs_delayed_tree_ref, node);
 420 }
 421
 422 static inline struct btrfs_delayed_data_ref *
 423 btrfs_delayed_node_to_data_ref(struct btrfs_delayed_ref_node *node)
 424 {
 425         return container_of(node, struct btrfs_delayed_data_ref, node);
 426 }
 427
 428 #endif