redis源码分析——list

什么是list

list是简单的字符串列表，按照插入顺序排序。你可以添加一个元素到列表的头部（左边）或者尾部（右边）。list内部结构为quicklist。

quicklist

quicklist是由经过lzf算法压缩的ziplist组成的双向链表。

数据结构

/* Optimization levels for size-based filling */
static const size_t optimization_level[] = {4096, 8192, 16384, 32768, 65536};

typedef struct quicklist {
    quicklistNode *head;
    quicklistNode *tail;
    unsigned long count;        /* total count of all entries in all ziplists */
    unsigned long len;          /* number of quicklistNodes */
    int fill : 16;              /* fill factor for individual nodes */
    unsigned int compress : 16; /* depth of end nodes not to compress;0=off */
} quicklist

quicklist是由quicklistNode组成的双向链表，其中compress表示链表两端不压缩节点的个数，若compress=n，则这个链表头部的n个结点和尾部的n个结点的ziplist是非压缩的，而其它结点采用lzf算法压缩。fill表示每个节点的装载因子，即决定了每个结点的最大长度=optimization_level[-fill - 1]。

typedef struct quicklistNode {
    struct quicklistNode *prev;
    struct quicklistNode *next;
    unsigned char *zl;
    unsigned int sz;             /* ziplist size in bytes */
    unsigned int count : 16;     /* count of items in ziplist */
    unsigned int encoding : 2;   /* RAW==1 or LZF==2 */
    unsigned int container : 2;  /* NONE==1 or ZIPLIST==2 */
    unsigned int recompress : 1; /* was this node previous compressed? */
    unsigned int attempted_compress : 1; /* node can't compress; too small */
    unsigned int extra : 10; /* more bits to steal for future usage */
} quicklistNode;

若这个节点没有压缩，则zl指向ziplist，否则zl指向压缩过的数据，即

typedef struct quicklistLZF {
    unsigned int sz; /* LZF size in bytes*/
    char compressed[];
} quicklistLZF

另一个结构体quicklistEntry表示quicklist中某一具体的元素的信息，

typedef struct quicklistEntry {
    const quicklist *quicklist;
    quicklistNode *node;
    unsigned char *zi;
    unsigned char *value;
    long long longval;
    unsigned int sz;
    int offset;
} quicklistEntry;

其中zi指向ziplist中的某个entry，sz为该entry的数据的长度，offset表示在ziplist元素中的偏移个数。

quicklist的创建

quicklist *quicklistCreate(void) {
    struct quicklist *quicklist;

    quicklist = zmalloc(sizeof(*quicklist));
    quicklist->head = quicklist->tail = NULL;
    quicklist->len = 0;
    quicklist->count = 0;
    quicklist->compress = 0;
    quicklist->fill = -2;
    return quicklist;
}

fill默认为-2，即ziplist中元素个数最多为8192。

插入元素

首先获取待插入位置的quicklistNode，及其内部的ziplist，若该ziplist未满，则直接插入ziplist中，否则新建一个quicklistNode插入数据，并将该node挂到链表中。

/* Wrapper to allow argument-based switching between HEAD/TAIL pop */
void quicklistPush(quicklist *quicklist, void *value, const size_t sz,
                   int where) {
    if (where == QUICKLIST_HEAD) {
        quicklistPushHead(quicklist, value, sz);
    } else if (where == QUICKLIST_TAIL) {
        quicklistPushTail(quicklist, value, sz);
    }
}

/* Add new entry to tail node of quicklist.
 *
 * Returns 0 if used existing tail.
 * Returns 1 if new tail created. */
int quicklistPushTail(quicklist *quicklist, void *value, size_t sz) {
    quicklistNode *orig_tail = quicklist->tail;
    if (likely(
            _quicklistNodeAllowInsert(quicklist->tail, quicklist->fill, sz))) {
        quicklist->tail->zl =
            ziplistPush(quicklist->tail->zl, value, sz, ZIPLIST_TAIL);
        quicklistNodeUpdateSz(quicklist->tail);
    } else {
        quicklistNode *node = quicklistCreateNode();
        node->zl = ziplistPush(ziplistNew(), value, sz, ZIPLIST_TAIL);

        quicklistNodeUpdateSz(node);
        _quicklistInsertNodeAfter(quicklist, quicklist->tail, node);
    }
    quicklist->count++;
    quicklist->tail->count++;
    return (orig_tail != quicklist->tail);
}

REDIS_STATIC int _quicklistNodeAllowInsert(const quicklistNode *node,
                                           const int fill, const size_t sz) {
    if (unlikely(!node))
        return 0;

    int ziplist_overhead;
    /* size of previous offset */
    if (sz < 254)
        ziplist_overhead = 1;
    else
        ziplist_overhead = 5;

    /* size of forward offset */
    if (sz < 64)
        ziplist_overhead += 1;
    else if (likely(sz < 16384))
        ziplist_overhead += 2;
    else
        ziplist_overhead += 5;

    /* new_sz overestimates if 'sz' encodes to an integer type */
    unsigned int new_sz = node->sz + sz + ziplist_overhead;
    if (likely(_quicklistNodeSizeMeetsOptimizationRequirement(new_sz, fill)))
        return 1;
    else if (!sizeMeetsSafetyLimit(new_sz))
        return 0;
    else if ((int)node->count < fill)
        return 1;
    else
        return 0;
}

删除元素

/* Default pop function
 *
 * Returns malloc'd value from quicklist */
int quicklistPop(quicklist *quicklist, int where, unsigned char **data,
                 unsigned int *sz, long long *slong) {
    unsigned char *vstr;
    unsigned int vlen;
    long long vlong;
    if (quicklist->count == 0)
        return 0;
    int ret = quicklistPopCustom(quicklist, where, &vstr, &vlen, &vlong,
                                 _quicklistSaver);
    if (data)
        *data = vstr;
    if (slong)
        *slong = vlong;
    if (sz)
        *sz = vlen;
    return ret;
}

/* pop from quicklist and return result in 'data' ptr.  Value of 'data'
 * is the return value of 'saver' function pointer if the data is NOT a number.
 *
 * If the quicklist element is a long long, then the return value is returned in
 * 'sval'.
 *
 * Return value of 0 means no elements available.
 * Return value of 1 means check 'data' and 'sval' for values.
 * If 'data' is set, use 'data' and 'sz'.  Otherwise, use 'sval'. */
int quicklistPopCustom(quicklist *quicklist, int where, unsigned char **data,
                       unsigned int *sz, long long *sval,
                       void *(*saver)(unsigned char *data, unsigned int sz)) {
    unsigned char *p;
    unsigned char *vstr;
    unsigned int vlen;
    long long vlong;
    int pos = (where == QUICKLIST_HEAD) ? 0 : -1;

    if (quicklist->count == 0)
        return 0;

    if (data)
        *data = NULL;
    if (sz)
        *sz = 0;
    if (sval)
        *sval = -123456789;

    quicklistNode *node;
    if (where == QUICKLIST_HEAD && quicklist->head) {
        node = quicklist->head;
    } else if (where == QUICKLIST_TAIL && quicklist->tail) {
        node = quicklist->tail;
    } else {
        return 0;
    }

    p = ziplistIndex(node->zl, pos);
    if (ziplistGet(p, &vstr, &vlen, &vlong)) {
        if (vstr) {
            if (data)
                *data = saver(vstr, vlen);
            if (sz)
                *sz = vlen;
        } else {
            if (data)
                *data = NULL;
            if (sval)
                *sval = vlong;
        }
        quicklistDelIndex(quicklist, node, &p);
        return 1;
    }
    return 0;
}

/* Delete one entry from list given the node for the entry and a pointer
 * to the entry in the node.
 *
 * Note: quicklistDelIndex() *requires* uncompressed nodes because you
 *       already had to get *p from an uncompressed node somewhere.
 *
 * Returns 1 if the entire node was deleted, 0 if node still exists.
 * Also updates in/out param 'p' with the next offset in the ziplist. */
REDIS_STATIC int quicklistDelIndex(quicklist *quicklist, quicklistNode *node,
                                   unsigned char **p) {
    int gone = 0;

    node->zl = ziplistDelete(node->zl, p);
    node->count--;
    if (node->count == 0) {
        gone = 1;
        __quicklistDelNode(quicklist, node);
    } else {
        quicklistNodeUpdateSz(node);
    }
    quicklist->count--;
    /* If we deleted the node, the original node is no longer valid */
    return gone ? 1 : 0;
}

REDIS_STATIC void __quicklistDelNode(quicklist *quicklist,
                                     quicklistNode *node) {
    if (node->next)
        node->next->prev = node->prev;
    if (node->prev)
        node->prev->next = node->next;

    if (node == quicklist->tail) {
        quicklist->tail = node->prev;
    }

    if (node == quicklist->head) {
        quicklist->head = node->next;
    }

    /* If we deleted a node within our compress depth, we
     * now have compressed nodes needing to be decompressed. */
    __quicklistCompress(quicklist, NULL);

    quicklist->count -= node->count;

    zfree(node->zl);
    zfree(node);
    quicklist->len--;
}

list的阻塞操作

数据结构

阻塞操作涉及到的数据结构如下

/* This structure holds the blocking operation state for a client.
 * The fields used depend on client->btype. */
typedef struct blockingState {
    /* Generic fields. */
    mstime_t timeout;       /* Blocking operation timeout. If UNIX current time
                             * is > timeout then the operation timed out. */

    /* BLOCKED_LIST */
    dict *keys;             /* The keys we are waiting to terminate a blocking
                             * operation such as BLPOP. Otherwise NULL. */
    robj *target;           /* The key that should receive the element,
                             * for BRPOPLPUSH. */

    /* BLOCKED_WAIT */
    int numreplicas;        /* Number of replicas we are waiting for ACK. */
    long long reploffset;   /* Replication offset to reach. */

    /* BLOCKED_MODULE */
    void *module_blocked_handle; /* RedisModuleBlockedClient structure.
                                    which is opaque for the Redis core, only
                                    handled in module.c. */
} blockingState;

blockingState是client的 一个属性，其中keys存放了该client被哪些key所阻塞。

/* Redis database representation. There are multiple databases identified
 * by integers from 0 (the default database) up to the max configured
 * database. The database number is the 'id' field in the structure. */
typedef struct redisDb {
    dict *dict;                 /* The keyspace for this DB */
    dict *expires;              /* Timeout of keys with a timeout set */
    dict *blocking_keys;        /* Keys with clients waiting for data (BLPOP)*/
    dict *ready_keys;           /* Blocked keys that received a PUSH */
    dict *watched_keys;         /* WATCHED keys for MULTI/EXEC CAS */
    int id;                     /* Database ID */
    long long avg_ttl;          /* Average TTL, just for stats */
} redisDb;

redisDb中有一个blocking_keys属性，其键为阻塞的key，值为一个列表，在列表中存放了被该key所阻塞的所有client。当blocking_keys中的某个key执行了一次push操作后，该key就会放入ready_keys中。

block

/* Blocking RPOP/LPOP */
void blockingPopGenericCommand(client *c, int where) {
    robj *o;
    mstime_t timeout;
    int j;

    if (getTimeoutFromObjectOrReply(c,c->argv[c->argc-1],&timeout,UNIT_SECONDS)
        != C_OK) return;

    for (j = 1; j < c->argc-1; j++) {
    //从db中查找要pop的key，若所有key都不存在，则阻塞，否则从第一个存在的key中pop元素
        o = lookupKeyWrite(c->db,c->argv[j]);
        if (o != NULL) {
            if (o->type != OBJ_LIST) {
                addReply(c,shared.wrongtypeerr);
                return;
            } else {
                if (listTypeLength(o) != 0) {
                    /* Non empty list, this is like a non normal [LR]POP. */
                    char *event = (where == LIST_HEAD) ? "lpop" : "rpop";
                    robj *value = listTypePop(o,where);
                    serverAssert(value != NULL);

                    addReplyMultiBulkLen(c,2);
                    addReplyBulk(c,c->argv[j]);
                    addReplyBulk(c,value);
                    decrRefCount(value);
                    notifyKeyspaceEvent(NOTIFY_LIST,event,
                                        c->argv[j],c->db->id);
                    if (listTypeLength(o) == 0) {
                        dbDelete(c->db,c->argv[j]);
                        notifyKeyspaceEvent(NOTIFY_GENERIC,"del",
                                            c->argv[j],c->db->id);
                    }
                    signalModifiedKey(c->db,c->argv[j]);
                    server.dirty++;

                    /* Replicate it as an [LR]POP instead of B[LR]POP. */
                    rewriteClientCommandVector(c,2,
                        (where == LIST_HEAD) ? shared.lpop : shared.rpop,
                        c->argv[j]);
                    return;
                }
            }
        }
    }

    /* If we are inside a MULTI/EXEC and the list is empty the only thing
     * we can do is treating it as a timeout (even with timeout 0). */
    if (c->flags & CLIENT_MULTI) {
        addReply(c,shared.nullmultibulk);
        return;
    }

    /* If the list is empty or the key does not exists we must block */
    blockForKeys(c, c->argv + 1, c->argc - 2, timeout, NULL);
}

/* Set a client in blocking mode for the specified key, with the specified
 * timeout */
void blockForKeys(client *c, robj **keys, int numkeys, mstime_t timeout, robj *target) {
    dictEntry *de;
    list *l;
    int j;

    c->bpop.timeout = timeout;
    c->bpop.target = target;

    if (target != NULL) incrRefCount(target);

    for (j = 0; j < numkeys; j++) {
        /* If the key already exists in the dict ignore it. */
        //将要pop的所有key加到blockingState的keys中
        if (dictAdd(c->bpop.keys,keys[j],NULL) != DICT_OK) continue;
        incrRefCount(keys[j]);

        /* And in the other "side", to map keys -> clients */
        //将client加到blocking_keys中key对应的列表中
        de = dictFind(c->db->blocking_keys,keys[j]);
        if (de == NULL) {
            int retval;

            /* For every key we take a list of clients blocked for it */
            l = listCreate();
            retval = dictAdd(c->db->blocking_keys,keys[j],l);
            incrRefCount(keys[j]);
            serverAssertWithInfo(c,keys[j],retval == DICT_OK);
        } else {
            l = dictGetVal(de);
        }
        listAddNodeTail(l,c);
    }
    blockClient(c,BLOCKED_LIST);
}

/* Block a client for the specific operation type. Once the CLIENT_BLOCKED
 * flag is set client query buffer is not longer processed, but accumulated,
 * and will be processed when the client is unblocked. */
void blockClient(client *c, int btype) {
    c->flags |= CLIENT_BLOCKED;
    c->btype = btype;
    server.bpop_blocked_clients++;
}

unblock

若对list执行push操作，则在该list上被阻塞的client会被unblock

void pushGenericCommand(client *c, int where) {
    int j, pushed = 0;
    //查找key，因为list中pop出最后一个元素时会将list本身删除，
    //所以若某个client阻塞在某个key上，则这个key肯定是不存在的
    robj *lobj = lookupKeyWrite(c->db,c->argv[1]);

    if (lobj && lobj->type != OBJ_LIST) {
        addReply(c,shared.wrongtypeerr);
        return;
    }

    for (j = 2; j < c->argc; j++) {
    //若list不存在则重新创建
        if (!lobj) {
            lobj = createQuicklistObject();
            quicklistSetOptions(lobj->ptr, server.list_max_ziplist_size,
                                server.list_compress_depth);
            //在dbAdd中会进行unblock操作
            dbAdd(c->db,c->argv[1],lobj);
        }
        listTypePush(lobj,c->argv[j],where);
        pushed++;
    }
    addReplyLongLong(c, (lobj ? listTypeLength(lobj) : 0));
    if (pushed) {
        char *event = (where == LIST_HEAD) ? "lpush" : "rpush";

        signalModifiedKey(c->db,c->argv[1]);
        notifyKeyspaceEvent(NOTIFY_LIST,event,c->argv[1],c->db->id);
    }
    server.dirty += pushed;
}

void dbAdd(redisDb *db, robj *key, robj *val) {
    sds copy = sdsdup(key->ptr);
    int retval = dictAdd(db->dict, copy, val);

    serverAssertWithInfo(NULL,key,retval == DICT_OK);
    //将key放到ready_keys中
    if (val->type == OBJ_LIST) signalListAsReady(db, key);
    if (server.cluster_enabled) slotToKeyAdd(key);
 }
 
void signalListAsReady(redisDb *db, robj *key) {
    readyList *rl;

    /* No clients blocking for this key? No need to queue it. */
    if (dictFind(db->blocking_keys,key) == NULL) return;

    /* Key was already signaled? No need to queue it again. */
    if (dictFind(db->ready_keys,key) != NULL) return;

    /* Ok, we need to queue this key into server.ready_keys. */
    rl = zmalloc(sizeof(*rl));
    rl->key = key;
    rl->db = db;
    incrRefCount(key);
    listAddNodeTail(server.ready_keys,rl);

    /* We also add the key in the db->ready_keys dictionary in order
     * to avoid adding it multiple times into a list with a simple O(1)
     * check. */
    incrRefCount(key);
    serverAssert(dictAdd(db->ready_keys,key,NULL) == DICT_OK);
}

真正的unblock操作位于processCommand函数中，在processCommand调用call执行完push操作后，接着会判断ready_keys是否为空，若不为空，则进行unblock

int processCommand(client *c) {
	...
	...
	...
    /* Exec the command */
    if (c->flags & CLIENT_MULTI &&
        c->cmd->proc != execCommand && c->cmd->proc != discardCommand &&
        c->cmd->proc != multiCommand && c->cmd->proc != watchCommand)
    {
        queueMultiCommand(c);
        addReply(c,shared.queued);
    } else {
        call(c,CMD_CALL_FULL);
        c->woff = server.master_repl_offset;
        if (listLength(server.ready_keys))
            handleClientsBlockedOnLists();
    }
    return C_OK;
}

void handleClientsBlockedOnLists(void) {
    while(listLength(server.ready_keys) != 0) {
        list *l;

        /* Point server.ready_keys to a fresh list and save the current one
         * locally. This way as we run the old list we are free to call
         * signalListAsReady() that may push new elements in server.ready_keys
         * when handling clients blocked into BRPOPLPUSH. */
        l = server.ready_keys;
        server.ready_keys = listCreate();

        while(listLength(l) != 0) {
            listNode *ln = listFirst(l);
            readyList *rl = ln->value;

            /* First of all remove this key from db->ready_keys so that
             * we can safely call signalListAsReady() against this key. */
            dictDelete(rl->db->ready_keys,rl->key);

            /* If the key exists and it's a list, serve blocked clients
             * with data. */
            robj *o = lookupKeyWrite(rl->db,rl->key);
            if (o != NULL && o->type == OBJ_LIST) {
                dictEntry *de;

                /* We serve clients in the same order they blocked for
                 * this key, from the first blocked to the last. */
                de = dictFind(rl->db->blocking_keys,rl->key);
                if (de) {
                    list *clients = dictGetVal(de);
                    int numclients = listLength(clients);

                    while(numclients--) {
                        listNode *clientnode = listFirst(clients);
                        client *receiver = clientnode->value;
                        robj *dstkey = receiver->bpop.target;
                        int where = (receiver->lastcmd &&
                                     receiver->lastcmd->proc == blpopCommand) ?
                                    LIST_HEAD : LIST_TAIL;
                        robj *value = listTypePop(o,where);

                        if (value) {
                            /* Protect receiver->bpop.target, that will be
                             * freed by the next unblockClient()
                             * call. */
                            if (dstkey) incrRefCount(dstkey);
                            unblockClient(receiver);

							//通知slave执行pop操作，若dstkey不为空，则将value push到dstkey中
                            if (serveClientBlockedOnList(receiver,
                                rl->key,dstkey,rl->db,value,
                                where) == C_ERR)
                            {
                                /* If we failed serving the client we need
                                 * to also undo the POP operation. */
                                    listTypePush(o,value,where);
                            }

                            if (dstkey) decrRefCount(dstkey);
                            decrRefCount(value);
                        } else {
                            break;
                        }
                    }
                }

                if (listTypeLength(o) == 0) {
                    dbDelete(rl->db,rl->key);
                }
                /* We don't call signalModifiedKey() as it was already called
                 * when an element was pushed on the list. */
            }

            /* Free this item. */
            decrRefCount(rl->key);
            zfree(rl);
            listDelNode(l,ln);
        }
        listRelease(l); /* We have the new list on place at this point. */
    }
}

/* Unblock a client calling the right function depending on the kind
 * of operation the client is blocking for. */
void unblockClient(client *c) {
    if (c->btype == BLOCKED_LIST) {
        unblockClientWaitingData(c);
    } else if (c->btype == BLOCKED_WAIT) {
        unblockClientWaitingReplicas(c);
    } else if (c->btype == BLOCKED_MODULE) {
        unblockClientFromModule(c);
    } else {
        serverPanic("Unknown btype in unblockClient().");
    }
    /* Clear the flags, and put the client in the unblocked list so that
     * we'll process new commands in its query buffer ASAP. */
    c->flags &= ~CLIENT_BLOCKED;
    c->btype = BLOCKED_NONE;
    server.bpop_blocked_clients--;
    /* The client may already be into the unblocked list because of a previous
     * blocking operation, don't add back it into the list multiple times. */
    if (!(c->flags & CLIENT_UNBLOCKED)) {
        c->flags |= CLIENT_UNBLOCKED;
        listAddNodeTail(server.unblocked_clients,c);
    }
}