1 /* SPDX-License-Identifier: GPL-2.0 */
3 * Copyright (C) 2008 Oracle. All rights reserved.
6 #ifndef BTRFS_LOCKING_H
7 #define BTRFS_LOCKING_H
9 #include <linux/atomic.h>
10 #include <linux/wait.h>
11 #include <linux/lockdep.h>
12 #include <linux/percpu_counter.h>
13 #include "extent_io.h"
19 #define BTRFS_WRITE_LOCK 1
20 #define BTRFS_READ_LOCK 2
23 * We are limited in number of subclasses by MAX_LOCKDEP_SUBCLASSES, which at
24 * the time of this patch is 8, which is how many we use. Keep this in mind if
25 * you decide you want to add another subclass.
27 enum btrfs_lock_nesting
{
31 * When we COW a block we are holding the lock on the original block,
32 * and since our lockdep maps are rootid+level, this confuses lockdep
33 * when we lock the newly allocated COW'd block. Handle this by having
34 * a subclass for COW'ed blocks so that lockdep doesn't complain.
39 * Oftentimes we need to lock adjacent nodes on the same level while
40 * still holding the lock on the original node we searched to, such as
41 * for searching forward or for split/balance.
43 * Because of this we need to indicate to lockdep that this is
44 * acceptable by having a different subclass for each of these
51 * When splitting we will be holding a lock on the left/right node when
52 * we need to cow that node, thus we need a new set of subclasses for
53 * these two operations.
55 BTRFS_NESTING_LEFT_COW
,
56 BTRFS_NESTING_RIGHT_COW
,
59 * When splitting we may push nodes to the left or right, but still use
60 * the subsequent nodes in our path, keeping our locks on those adjacent
61 * blocks. Thus when we go to allocate a new split block we've already
62 * used up all of our available subclasses, so this subclass exists to
63 * handle this case where we need to allocate a new split block.
68 * When promoting a new block to a root we need to have a special
69 * subclass so we don't confuse lockdep, as it will appear that we are
70 * locking a higher level node before a lower level one. Copying also
71 * has this problem as it appears we're locking the same block again
72 * when we make a snapshot of an existing root.
74 BTRFS_NESTING_NEW_ROOT
,
77 * We are limited to MAX_LOCKDEP_SUBLCLASSES number of subclasses, so
78 * add this in here and add a static_assert to keep us from going over
79 * the limit. As of this writing we're limited to 8, and we're
80 * definitely using 8, hence this check to keep us from messing up in
86 enum btrfs_lockdep_trans_states
{
87 BTRFS_LOCKDEP_TRANS_COMMIT_PREP
,
88 BTRFS_LOCKDEP_TRANS_UNBLOCKED
,
89 BTRFS_LOCKDEP_TRANS_SUPER_COMMITTED
,
90 BTRFS_LOCKDEP_TRANS_COMPLETED
,
94 * Lockdep annotation for wait events.
96 * @owner: The struct where the lockdep map is defined
97 * @lock: The lockdep map corresponding to a wait event
99 * This macro is used to annotate a wait event. In this case a thread acquires
100 * the lockdep map as writer (exclusive lock) because it has to block until all
101 * the threads that hold the lock as readers signal the condition for the wait
102 * event and release their locks.
104 #define btrfs_might_wait_for_event(owner, lock) \
106 rwsem_acquire(&owner->lock##_map, 0, 0, _THIS_IP_); \
107 rwsem_release(&owner->lock##_map, _THIS_IP_); \
111 * Protection for the resource/condition of a wait event.
113 * @owner: The struct where the lockdep map is defined
114 * @lock: The lockdep map corresponding to a wait event
116 * Many threads can modify the condition for the wait event at the same time
117 * and signal the threads that block on the wait event. The threads that modify
118 * the condition and do the signaling acquire the lock as readers (shared
121 #define btrfs_lockdep_acquire(owner, lock) \
122 rwsem_acquire_read(&owner->lock##_map, 0, 0, _THIS_IP_)
125 * Used after signaling the condition for a wait event to release the lockdep
126 * map held by a reader thread.
128 #define btrfs_lockdep_release(owner, lock) \
129 rwsem_release(&owner->lock##_map, _THIS_IP_)
132 * Used to account for the fact that when doing io_uring encoded I/O, we can
133 * return to userspace with the inode lock still held.
135 #define btrfs_lockdep_inode_acquire(owner, lock) \
136 rwsem_acquire_read(&owner->vfs_inode.lock.dep_map, 0, 0, _THIS_IP_)
138 #define btrfs_lockdep_inode_release(owner, lock) \
139 rwsem_release(&owner->vfs_inode.lock.dep_map, _THIS_IP_)
142 * Macros for the transaction states wait events, similar to the generic wait
145 #define btrfs_might_wait_for_state(owner, i) \
147 rwsem_acquire(&owner->btrfs_state_change_map[i], 0, 0, _THIS_IP_); \
148 rwsem_release(&owner->btrfs_state_change_map[i], _THIS_IP_); \
151 #define btrfs_trans_state_lockdep_acquire(owner, i) \
152 rwsem_acquire_read(&owner->btrfs_state_change_map[i], 0, 0, _THIS_IP_)
154 #define btrfs_trans_state_lockdep_release(owner, i) \
155 rwsem_release(&owner->btrfs_state_change_map[i], _THIS_IP_)
157 /* Initialization of the lockdep map */
158 #define btrfs_lockdep_init_map(owner, lock) \
160 static struct lock_class_key lock##_key; \
161 lockdep_init_map(&owner->lock##_map, #lock, &lock##_key, 0); \
164 /* Initialization of the transaction states lockdep maps. */
165 #define btrfs_state_lockdep_init_map(owner, lock, state) \
167 static struct lock_class_key lock##_key; \
168 lockdep_init_map(&owner->btrfs_state_change_map[state], #lock, \
172 static_assert(BTRFS_NESTING_MAX
<= MAX_LOCKDEP_SUBCLASSES
,
173 "too many lock subclasses defined");
175 void btrfs_tree_lock_nested(struct extent_buffer
*eb
, enum btrfs_lock_nesting nest
);
177 static inline void btrfs_tree_lock(struct extent_buffer
*eb
)
179 btrfs_tree_lock_nested(eb
, BTRFS_NESTING_NORMAL
);
182 void btrfs_tree_unlock(struct extent_buffer
*eb
);
184 void btrfs_tree_read_lock_nested(struct extent_buffer
*eb
, enum btrfs_lock_nesting nest
);
186 static inline void btrfs_tree_read_lock(struct extent_buffer
*eb
)
188 btrfs_tree_read_lock_nested(eb
, BTRFS_NESTING_NORMAL
);
191 void btrfs_tree_read_unlock(struct extent_buffer
*eb
);
192 int btrfs_try_tree_read_lock(struct extent_buffer
*eb
);
193 struct extent_buffer
*btrfs_lock_root_node(struct btrfs_root
*root
);
194 struct extent_buffer
*btrfs_read_lock_root_node(struct btrfs_root
*root
);
195 struct extent_buffer
*btrfs_try_read_lock_root_node(struct btrfs_root
*root
);
197 #ifdef CONFIG_BTRFS_DEBUG
198 static inline void btrfs_assert_tree_write_locked(struct extent_buffer
*eb
)
200 lockdep_assert_held_write(&eb
->lock
);
203 static inline void btrfs_assert_tree_write_locked(struct extent_buffer
*eb
) { }
206 void btrfs_unlock_up_safe(struct btrfs_path
*path
, int level
);
208 static inline void btrfs_tree_unlock_rw(struct extent_buffer
*eb
, int rw
)
210 if (rw
== BTRFS_WRITE_LOCK
)
211 btrfs_tree_unlock(eb
);
212 else if (rw
== BTRFS_READ_LOCK
)
213 btrfs_tree_read_unlock(eb
);
218 struct btrfs_drew_lock
{
221 wait_queue_head_t pending_writers
;
222 wait_queue_head_t pending_readers
;
225 void btrfs_drew_lock_init(struct btrfs_drew_lock
*lock
);
226 void btrfs_drew_write_lock(struct btrfs_drew_lock
*lock
);
227 bool btrfs_drew_try_write_lock(struct btrfs_drew_lock
*lock
);
228 void btrfs_drew_write_unlock(struct btrfs_drew_lock
*lock
);
229 void btrfs_drew_read_lock(struct btrfs_drew_lock
*lock
);
230 void btrfs_drew_read_unlock(struct btrfs_drew_lock
*lock
);
232 #ifdef CONFIG_DEBUG_LOCK_ALLOC
233 void btrfs_set_buffer_lockdep_class(u64 objectid
, struct extent_buffer
*eb
, int level
);
234 void btrfs_maybe_reset_lockdep_class(struct btrfs_root
*root
, struct extent_buffer
*eb
);
236 static inline void btrfs_set_buffer_lockdep_class(u64 objectid
,
237 struct extent_buffer
*eb
, int level
)
240 static inline void btrfs_maybe_reset_lockdep_class(struct btrfs_root
*root
,
241 struct extent_buffer
*eb
)