1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
|
#include "server.h"
#include "bio.h"
#include "atomicvar.h"
#include "cluster.h"
static size_t lazyfree_objects = 0;
pthread_mutex_t lazyfree_objects_mutex = PTHREAD_MUTEX_INITIALIZER;
/* Return the number of currently pending objects to free. */
size_t lazyfreeGetPendingObjectsCount(void) {
size_t aux;
atomicGet(lazyfree_objects,aux);
return aux;
}
/* Return the amount of work needed in order to free an object.
* The return value is not always the actual number of allocations the
* object is compoesd of, but a number proportional to it.
*
* For strings the function always returns 1.
*
* For aggregated objects represented by hash tables or other data structures
* the function just returns the number of elements the object is composed of.
*
* Objects composed of single allocations are always reported as having a
* single item even if they are actually logical composed of multiple
* elements.
*
* For lists the function returns the number of elements in the quicklist
* representing the list. */
size_t lazyfreeGetFreeEffort(robj *obj) {
if (obj->type == OBJ_LIST) {
quicklist *ql = obj->ptr;
return ql->len;
} else if (obj->type == OBJ_SET && obj->encoding == OBJ_ENCODING_HT) {
dict *ht = obj->ptr;
return dictSize(ht);
} else if (obj->type == OBJ_ZSET && obj->encoding == OBJ_ENCODING_SKIPLIST){
zset *zs = obj->ptr;
return zs->zsl->length;
} else if (obj->type == OBJ_HASH && obj->encoding == OBJ_ENCODING_HT) {
dict *ht = obj->ptr;
return dictSize(ht);
} else {
return 1; /* Everything else is a single allocation. */
}
}
/* Delete a key, value, and associated expiration entry if any, from the DB.
* If there are enough allocations to free the value object may be put into
* a lazy free list instead of being freed synchronously. The lazy free list
* will be reclaimed in a different bio.c thread. */
#define LAZYFREE_THRESHOLD 64
int dbAsyncDelete(redisDb *db, robj *keyname) {
/* If the value is composed of a few allocations, to free in a lazy way
* is actually just slower... So under a certain limit we just free
* the object synchronously. */
dictEntry *de = dictUnlink(db->dict,keyname->ptr);
if (de) {
rkey *key = dictGetKey(de);
robj *val = dictGetVal(de);
size_t free_effort = lazyfreeGetFreeEffort(val);
/* Remove the entry from the expire tree since this is fast anyway. */
if (key->flags & KEY_FLAG_EXPIRE) removeExpireFromTree(db,key);
/* If releasing the object is too much work, do it in the background
* by adding the object to the lazy free list.
* Note that if the object is shared, to reclaim it now it is not
* possible. This rarely happens, however sometimes the implementation
* of parts of the Redis core may call incrRefCount() to protect
* objects, and then call dbDelete(). In this case we'll fall
* through and reach the dictFreeUnlinkedEntry() call, that will be
* equivalent to just calling decrRefCount(). */
if (free_effort > LAZYFREE_THRESHOLD && val->refcount == 1) {
atomicIncr(lazyfree_objects,1);
bioCreateBackgroundJob(BIO_LAZY_FREE,val,NULL,NULL);
dictSetVal(db->dict,de,NULL);
}
}
/* Release the key-val pair, or just the key if we set the val
* field to NULL in order to lazy free it later. */
if (de) {
dictFreeUnlinkedEntry(db->dict,de);
if (server.cluster_enabled) slotToKeyDel(keyname);
return 1;
} else {
return 0;
}
}
/* Free an object, if the object is huge enough, free it in async way. */
void freeObjAsync(robj *o) {
size_t free_effort = lazyfreeGetFreeEffort(o);
if (free_effort > LAZYFREE_THRESHOLD && o->refcount == 1) {
atomicIncr(lazyfree_objects,1);
bioCreateBackgroundJob(BIO_LAZY_FREE,o,NULL,NULL);
} else {
decrRefCount(o);
}
}
/* Empty a Redis DB asynchronously. What the function does actually is to
* create a new empty set of hash tables and scheduling the old ones for
* lazy freeing. */
void emptyDbAsync(redisDb *db) {
void *hashtable = db->dict, *tree = db->expires;
db->dict = dictCreate(&dbDictType,NULL);
db->expires = raxNew();
atomicIncr(lazyfree_objects,dictSize((dict*)hashtable));
bioCreateBackgroundJob(BIO_LAZY_FREE,NULL,hashtable,tree);
}
/* Empty the slots-keys map of Redis CLuster by creating a new empty one
* and scheduiling the old for lazy freeing. */
void slotToKeyFlushAsync(void) {
rax *old = server.cluster->slots_to_keys;
server.cluster->slots_to_keys = raxNew();
memset(server.cluster->slots_keys_count,0,
sizeof(server.cluster->slots_keys_count));
atomicIncr(lazyfree_objects,old->numele);
bioCreateBackgroundJob(BIO_LAZY_FREE,NULL,NULL,old);
}
/* Release objects from the lazyfree thread. It's just decrRefCount()
* updating the count of objects to release. */
void lazyfreeFreeObjectFromBioThread(robj *o) {
decrRefCount(o);
atomicDecr(lazyfree_objects,1);
}
/* Release a database from the lazyfree thread. The 'db' pointer is the
* database which was substitutied with a fresh one in the main thread
* when the database was logically deleted. 'sl' is a skiplist used by
* Redis Cluster in order to take the hash slots -> keys mapping. This
* may be NULL if Redis Cluster is disabled. */
void lazyfreeFreeDatabaseFromBioThread(dict *hashtable, rax *tree) {
size_t numkeys = dictSize(hashtable);
dictRelease(hashtable);
raxFree(tree);
atomicDecr(lazyfree_objects,numkeys);
}
/* Release the skiplist mapping Redis Cluster keys to slots in the
* lazyfree thread. */
void lazyfreeFreeSlotsMapFromBioThread(rax *rt) {
size_t len = rt->numele;
raxFree(rt);
atomicDecr(lazyfree_objects,len);
}
|
