#include "redis.h"
#include <signal.h>
+#include <ctype.h>
void SlotToKeyAdd(robj *key);
void SlotToKeyDel(robj *key);
* C-level DB API
*----------------------------------------------------------------------------*/
-/* Important notes on lookup and disk store.
- *
- * When disk store is enabled on lookup we can have different cases.
- *
- * a) The key is in memory:
- * - If the key is not in IO_SAVEINPROG state we can access it.
- * As if it's just IO_SAVE this means we have the key in the IO queue
- * but can't be accessed by the IO thread (it requires to be
- * translated into an IO Job by the cache cron function.)
- * - If the key is in IO_SAVEINPROG we can't touch the key and have
- * to blocking wait completion of operations.
- * b) The key is not in memory:
- * - If it's marked as non existing on disk as well (negative cache)
- * we don't need to perform the disk access.
- * - if the key MAY EXIST, but is not in memory, and it is marked as IO_SAVE
- * then the key can only be a deleted one. As IO_SAVE keys are never
- * evicted (dirty state), so the only possibility is that key was deleted.
- * - if the key MAY EXIST we need to blocking load it.
- * We check that the key is not in IO_SAVEINPROG state before accessing
- * the disk object. If it is in this state, we wait.
- */
-
-void lookupWaitBusyKey(redisDb *db, robj *key) {
- /* FIXME: wait just for this key, not everything */
- waitEmptyIOJobsQueue();
- processAllPendingIOJobs();
- redisAssert((cacheScheduleIOGetFlags(db,key) & REDIS_IO_SAVEINPROG) == 0);
-}
-
robj *lookupKey(redisDb *db, robj *key) {
dictEntry *de = dictFind(db->dict,key->ptr);
if (de) {
- robj *val = dictGetEntryVal(de);
+ robj *val = dictGetVal(de);
/* Update the access time for the aging algorithm.
* Don't do it if we have a saving child, as this will trigger
* a copy on write madness. */
- if (server.bgsavechildpid == -1 && server.bgrewritechildpid == -1)
+ if (server.rdb_child_pid == -1 && server.aof_child_pid == -1)
val->lru = server.lruclock;
-
- if (server.ds_enabled &&
- cacheScheduleIOGetFlags(db,key) & REDIS_IO_SAVEINPROG)
- {
- /* Need to wait for the key to get unbusy */
- redisLog(REDIS_DEBUG,"Lookup found a key in SAVEINPROG state. Waiting. (Key was in the cache)");
- lookupWaitBusyKey(db,key);
- }
- server.stat_keyspace_hits++;
return val;
} else {
- time_t expire;
- robj *val;
-
- /* Key not found in the in memory hash table, but if disk store is
- * enabled we may have this key on disk. If so load it in memory
- * in a blocking way. */
- if (server.ds_enabled && cacheKeyMayExist(db,key)) {
- long flags = cacheScheduleIOGetFlags(db,key);
-
- /* They key is not in cache, but it has a SAVE op in queue?
- * The only possibility is that the key was deleted, since
- * dirty keys are not evicted. */
- if (flags & REDIS_IO_SAVE) {
- server.stat_keyspace_misses++;
- return NULL;
- }
-
- /* At this point we need to blocking load the key in memory.
- * The first thing we do is waiting here if the key is busy. */
- if (flags & REDIS_IO_SAVEINPROG) {
- redisLog(REDIS_DEBUG,"Lookup found a key in SAVEINPROG state. Waiting (while force loading).");
- lookupWaitBusyKey(db,key);
- }
-
- redisLog(REDIS_DEBUG,"Force loading key %s via lookup", key->ptr);
- val = dsGet(db,key,&expire);
- if (val) {
- dbAdd(db,key,val);
- if (expire != -1) setExpire(db,key,expire);
- server.stat_keyspace_hits++;
- return val;
- } else {
- cacheSetKeyDoesNotExist(db,key);
- }
- }
- server.stat_keyspace_misses++;
return NULL;
}
}
robj *lookupKeyRead(redisDb *db, robj *key) {
+ robj *val;
+
expireIfNeeded(db,key);
- return lookupKey(db,key);
+ val = lookupKey(db,key);
+ if (val == NULL)
+ server.stat_keyspace_misses++;
+ else
+ server.stat_keyspace_hits++;
+ return val;
}
robj *lookupKeyWrite(redisDb *db, robj *key) {
sds copy = sdsdup(key->ptr);
int retval = dictAdd(db->dict, copy, val);
- redisAssert(retval == REDIS_OK);
- if (server.ds_enabled) cacheSetKeyMayExist(db,key);
- if (server.cluster_enabled) SlotToKeyAdd(key);
+ redisAssertWithInfo(NULL,key,retval == REDIS_OK);
}
/* Overwrite an existing key with a new value. Incrementing the reference
void dbOverwrite(redisDb *db, robj *key, robj *val) {
struct dictEntry *de = dictFind(db->dict,key->ptr);
- redisAssert(de != NULL);
+ redisAssertWithInfo(NULL,key,de != NULL);
dictReplace(db->dict, key->ptr, val);
- if (server.ds_enabled) cacheSetKeyMayExist(db,key);
}
/* High level Set operation. This function can be used in order to set
}
incrRefCount(val);
removeExpire(db,key);
- touchWatchedKey(db,key);
+ signalModifiedKey(db,key);
}
int dbExists(redisDb *db, robj *key) {
de = dictGetRandomKey(db->dict);
if (de == NULL) return NULL;
- key = dictGetEntryKey(de);
+ key = dictGetKey(de);
keyobj = createStringObject(key,sdslen(key));
if (dictFind(db->expires,key)) {
if (expireIfNeeded(db,keyobj)) {
/* Delete a key, value, and associated expiration entry if any, from the DB */
int dbDelete(redisDb *db, robj *key) {
- /* If diskstore is enabled make sure to awake waiting clients for this key
- * as it is not really useful to wait for a key already deleted to be
- * loaded from disk. */
- if (server.ds_enabled) {
- handleClientsBlockedOnSwappedKey(db,key);
- cacheSetKeyDoesNotExist(db,key);
- }
-
/* Deleting an entry from the expires dict will not free the sds of
* the key, because it is shared with the main dictionary. */
if (dictSize(db->expires) > 0) dictDelete(db->expires,key->ptr);
if (dictDelete(db->dict,key->ptr) == DICT_OK) {
- if (server.cluster_enabled) SlotToKeyDel(key);
return 1;
} else {
return 0;
}
}
-/* Empty the whole database.
- * If diskstore is enabled this function will just flush the in-memory cache. */
long long emptyDb() {
int j;
long long removed = 0;
removed += dictSize(server.db[j].dict);
dictEmpty(server.db[j].dict);
dictEmpty(server.db[j].expires);
- if (server.ds_enabled) dictEmpty(server.db[j].io_negcache);
}
return removed;
}
void signalModifiedKey(redisDb *db, robj *key) {
touchWatchedKey(db,key);
- if (server.ds_enabled)
- cacheScheduleIO(db,key,REDIS_IO_SAVE);
}
void signalFlushedDb(int dbid) {
signalFlushedDb(c->db->id);
dictEmpty(c->db->dict);
dictEmpty(c->db->expires);
- if (server.ds_enabled) dsFlushDb(c->db->id);
addReply(c,shared.ok);
}
signalFlushedDb(-1);
server.dirty += emptyDb();
addReply(c,shared.ok);
- if (server.bgsavechildpid != -1) {
- kill(server.bgsavechildpid,SIGKILL);
- rdbRemoveTempFile(server.bgsavechildpid);
+ if (server.rdb_child_pid != -1) {
+ kill(server.rdb_child_pid,SIGKILL);
+ rdbRemoveTempFile(server.rdb_child_pid);
+ }
+ if (server.saveparamslen > 0) {
+ /* Normally rdbSave() will reset dirty, but we don't want this here
+ * as otherwise FLUSHALL will not be replicated nor put into the AOF. */
+ int saved_dirty = server.dirty;
+ rdbSave(server.rdb_filename);
+ server.dirty = saved_dirty;
}
- if (server.ds_enabled)
- dsFlushDb(-1);
- else
- rdbSave(server.dbfilename);
server.dirty++;
}
int deleted = 0, j;
for (j = 1; j < c->argc; j++) {
- if (server.ds_enabled) {
- lookupKeyRead(c->db,c->argv[j]);
- /* FIXME: this can be optimized a lot, no real need to load
- * a possibly huge value. */
- }
if (dbDelete(c->db,c->argv[j])) {
signalModifiedKey(c->db,c->argv[j]);
server.dirty++;
deleted++;
- } else if (server.ds_enabled) {
- if (cacheKeyMayExist(c->db,c->argv[j]) &&
- dsExists(c->db,c->argv[j]))
- {
- cacheScheduleIO(c->db,c->argv[j],REDIS_IO_SAVE);
- deleted = 1;
- }
}
}
addReplyLongLong(c,deleted);
}
void selectCommand(redisClient *c) {
- int id = atoi(c->argv[1]->ptr);
+ long id;
- if (server.cluster_enabled && id != 0) {
- addReplyError(c,"SELECT is not allowed in cluster mode");
+ if (getLongFromObjectOrReply(c, c->argv[1], &id,
+ "invalid DB index") != REDIS_OK)
return;
- }
+
if (selectDb(c,id) == REDIS_ERR) {
addReplyError(c,"invalid DB index");
} else {
unsigned long numkeys = 0;
void *replylen = addDeferredMultiBulkLength(c);
- di = dictGetIterator(c->db->dict);
+ di = dictGetSafeIterator(c->db->dict);
allkeys = (pattern[0] == '*' && pattern[1] == '\0');
while((de = dictNext(di)) != NULL) {
- sds key = dictGetEntryKey(de);
+ sds key = dictGetKey(de);
robj *keyobj;
if (allkeys || stringmatchlen(pattern,plen,key,sdslen(key),0)) {
}
void shutdownCommand(redisClient *c) {
- if (prepareForShutdown() == REDIS_OK)
- exit(0);
+ int flags = 0;
+
+ if (c->argc > 2) {
+ addReply(c,shared.syntaxerr);
+ return;
+ } else if (c->argc == 2) {
+ if (!strcasecmp(c->argv[1]->ptr,"nosave")) {
+ flags |= REDIS_SHUTDOWN_NOSAVE;
+ } else if (!strcasecmp(c->argv[1]->ptr,"save")) {
+ flags |= REDIS_SHUTDOWN_SAVE;
+ } else {
+ addReply(c,shared.syntaxerr);
+ return;
+ }
+ }
+ if (prepareForShutdown(flags) == REDIS_OK) exit(0);
addReplyError(c,"Errors trying to SHUTDOWN. Check logs.");
}
void renameGenericCommand(redisClient *c, int nx) {
robj *o;
+ long long expire;
/* To use the same key as src and dst is probably an error */
if (sdscmp(c->argv[1]->ptr,c->argv[2]->ptr) == 0) {
return;
incrRefCount(o);
+ expire = getExpire(c->db,c->argv[1]);
if (lookupKeyWrite(c->db,c->argv[2]) != NULL) {
if (nx) {
decrRefCount(o);
addReply(c,shared.czero);
return;
}
- dbOverwrite(c->db,c->argv[2],o);
- } else {
- dbAdd(c->db,c->argv[2],o);
+ /* Overwrite: delete the old key before creating the new one with the same name. */
+ dbDelete(c->db,c->argv[2]);
}
+ dbAdd(c->db,c->argv[2],o);
+ if (expire != -1) setExpire(c->db,c->argv[2],expire);
dbDelete(c->db,c->argv[1]);
signalModifiedKey(c->db,c->argv[1]);
signalModifiedKey(c->db,c->argv[2]);
redisDb *src, *dst;
int srcid;
- if (server.cluster_enabled) {
- addReplyError(c,"MOVE is not allowed in cluster mode");
- return;
- }
-
/* Obtain source and target DB pointers */
src = c->db;
srcid = c->db->id;
int removeExpire(redisDb *db, robj *key) {
/* An expire may only be removed if there is a corresponding entry in the
* main dict. Otherwise, the key will never be freed. */
- redisAssert(dictFind(db->dict,key->ptr) != NULL);
+ redisAssertWithInfo(NULL,key,dictFind(db->dict,key->ptr) != NULL);
return dictDelete(db->expires,key->ptr) == DICT_OK;
}
-void setExpire(redisDb *db, robj *key, time_t when) {
- dictEntry *de;
+void setExpire(redisDb *db, robj *key, long long when) {
+ dictEntry *kde, *de;
/* Reuse the sds from the main dict in the expire dict */
- de = dictFind(db->dict,key->ptr);
- redisAssert(de != NULL);
- dictReplace(db->expires,dictGetEntryKey(de),(void*)when);
+ kde = dictFind(db->dict,key->ptr);
+ redisAssertWithInfo(NULL,key,kde != NULL);
+ de = dictReplaceRaw(db->expires,dictGetKey(kde));
+ dictSetSignedIntegerVal(de,when);
}
/* Return the expire time of the specified key, or -1 if no expire
* is associated with this key (i.e. the key is non volatile) */
-time_t getExpire(redisDb *db, robj *key) {
+long long getExpire(redisDb *db, robj *key) {
dictEntry *de;
/* No expire? return ASAP */
/* The entry was found in the expire dict, this means it should also
* be present in the main dict (safety check). */
- redisAssert(dictFind(db->dict,key->ptr) != NULL);
- return (time_t) dictGetEntryVal(de);
+ redisAssertWithInfo(NULL,key,dictFind(db->dict,key->ptr) != NULL);
+ return dictGetSignedIntegerVal(de);
}
/* Propagate expires into slaves and the AOF file.
void propagateExpire(redisDb *db, robj *key) {
robj *argv[2];
- argv[0] = createStringObject("DEL",3);
+ argv[0] = shared.del;
argv[1] = key;
- incrRefCount(key);
+ incrRefCount(argv[0]);
+ incrRefCount(argv[1]);
- if (server.appendonly)
+ if (server.aof_state != REDIS_AOF_OFF)
feedAppendOnlyFile(server.delCommand,db->id,argv,2);
if (listLength(server.slaves))
replicationFeedSlaves(server.slaves,db->id,argv,2);
}
int expireIfNeeded(redisDb *db, robj *key) {
- time_t when = getExpire(db,key);
+ long long when = getExpire(db,key);
if (when < 0) return 0; /* No expire for this key */
+ /* Don't expire anything while loading. It will be done later. */
+ if (server.loading) return 0;
+
/* If we are running in the context of a slave, return ASAP:
* the slave key expiration is controlled by the master that will
* send us synthesized DEL operations for expired keys.
* that is, 0 if we think the key should be still valid, 1 if
* we think the key is expired at this time. */
if (server.masterhost != NULL) {
- return time(NULL) > when;
+ return mstime() > when;
}
/* Return when this key has not expired */
- if (time(NULL) <= when) return 0;
+ if (mstime() <= when) return 0;
/* Delete the key */
server.stat_expiredkeys++;
* Expires Commands
*----------------------------------------------------------------------------*/
-void expireGenericCommand(redisClient *c, robj *key, robj *param, long offset) {
+/* This is the generic command implementation for EXPIRE, PEXPIRE, EXPIREAT
+ * and PEXPIREAT. Because the commad second argument may be relative or absolute
+ * the "basetime" argument is used to signal what the base time is (either 0
+ * for *AT variants of the command, or the current time for relative expires).
+ *
+ * unit is either UNIT_SECONDS or UNIT_MILLISECONDS, and is only used for
+ * the argv[2] parameter. The basetime is always specified in milliesconds. */
+void expireGenericCommand(redisClient *c, long long basetime, int unit) {
dictEntry *de;
- long seconds;
+ robj *key = c->argv[1], *param = c->argv[2];
+ long long when; /* unix time in milliseconds when the key will expire. */
- if (getLongFromObjectOrReply(c, param, &seconds, NULL) != REDIS_OK) return;
+ if (getLongLongFromObjectOrReply(c, param, &when, NULL) != REDIS_OK)
+ return;
- seconds -= offset;
+ if (unit == UNIT_SECONDS) when *= 1000;
+ when += basetime;
de = dictFind(c->db->dict,key->ptr);
if (de == NULL) {
addReply(c,shared.czero);
return;
}
- if (seconds <= 0) {
- if (dbDelete(c->db,key)) server.dirty++;
- addReply(c, shared.cone);
+ /* EXPIRE with negative TTL, or EXPIREAT with a timestamp into the past
+ * should never be executed as a DEL when load the AOF or in the context
+ * of a slave instance.
+ *
+ * Instead we take the other branch of the IF statement setting an expire
+ * (possibly in the past) and wait for an explicit DEL from the master. */
+ if (when <= mstime() && !server.loading && !server.masterhost) {
+ robj *aux;
+
+ redisAssertWithInfo(c,key,dbDelete(c->db,key));
+ server.dirty++;
+
+ /* Replicate/AOF this as an explicit DEL. */
+ aux = createStringObject("DEL",3);
+ rewriteClientCommandVector(c,2,aux,key);
+ decrRefCount(aux);
signalModifiedKey(c->db,key);
+ addReply(c, shared.cone);
return;
} else {
- time_t when = time(NULL)+seconds;
setExpire(c->db,key,when);
addReply(c,shared.cone);
signalModifiedKey(c->db,key);
}
void expireCommand(redisClient *c) {
- expireGenericCommand(c,c->argv[1],c->argv[2],0);
+ expireGenericCommand(c,mstime(),UNIT_SECONDS);
}
void expireatCommand(redisClient *c) {
- expireGenericCommand(c,c->argv[1],c->argv[2],time(NULL));
+ expireGenericCommand(c,0,UNIT_SECONDS);
}
-void ttlCommand(redisClient *c) {
- time_t expire, ttl = -1;
+void pexpireCommand(redisClient *c) {
+ expireGenericCommand(c,mstime(),UNIT_MILLISECONDS);
+}
+
+void pexpireatCommand(redisClient *c) {
+ expireGenericCommand(c,0,UNIT_MILLISECONDS);
+}
+
+void ttlGenericCommand(redisClient *c, int output_ms) {
+ long long expire, ttl = -1;
- if (server.ds_enabled) lookupKeyRead(c->db,c->argv[1]);
expire = getExpire(c->db,c->argv[1]);
if (expire != -1) {
- ttl = (expire-time(NULL));
+ ttl = expire-mstime();
if (ttl < 0) ttl = -1;
}
- addReplyLongLong(c,(long long)ttl);
+ if (ttl == -1) {
+ addReplyLongLong(c,-1);
+ } else {
+ addReplyLongLong(c,output_ms ? ttl : ((ttl+500)/1000));
+ }
+}
+
+void ttlCommand(redisClient *c) {
+ ttlGenericCommand(c, 0);
+}
+
+void pttlCommand(redisClient *c) {
+ ttlGenericCommand(c, 1);
}
void persistCommand(redisClient *c) {
*numkeys = num;
return keys;
}
-
-/* Slot to Key API. This is used by Redis Cluster in order to obtain in
- * a fast way a key that belongs to a specified hash slot. This is useful
- * while rehashing the cluster. */
-void SlotToKeyAdd(robj *key) {
- unsigned int hashslot = keyHashSlot(key->ptr,sdslen(key->ptr));
-
- zslInsert(server.cluster.slots_to_keys,hashslot,key);
- incrRefCount(key);
-}
-
-void SlotToKeyDel(robj *key) {
- unsigned int hashslot = keyHashSlot(key->ptr,sdslen(key->ptr));
-
- zslDelete(server.cluster.slots_to_keys,hashslot,key);
-}
-
-unsigned int GetKeysInSlot(unsigned int hashslot, robj **keys, unsigned int count) {
- zskiplistNode *n;
- zrangespec range;
- int j = 0;
-
- range.min = range.max = hashslot;
- range.minex = range.maxex = 0;
-
- n = zslFirstInRange(server.cluster.slots_to_keys, range);
- while(n && n->score == hashslot && count--) {
- keys[j++] = n->obj;
- n = n->level[0].forward;
- }
- return j;
-}
projects / redis.git / blobdiff