X-Git-Url: https://git.saurik.com/redis.git/blobdiff_plain/092dac2a64f912fe67be2e924c58979b200aa64b..123d7c93cb73a0d09f2a8b34322f1a0f4acc71ba:/redis.c diff --git a/redis.c b/redis.c index e6606f18..607c827a 100644 --- a/redis.c +++ b/redis.c @@ -27,14 +27,24 @@ * POSSIBILITY OF SUCH DAMAGE. */ -#define REDIS_VERSION "0.07" +#define REDIS_VERSION "0.900" + +#include "fmacros.h" +#include "config.h" #include #include #include #include #include +#define __USE_POSIX199309 #include + +#ifdef HAVE_BACKTRACE +#include +#include +#endif /* HAVE_BACKTRACE */ + #include #include #include @@ -46,13 +56,17 @@ #include #include #include +#include +#include "redis.h" #include "ae.h" /* Event driven programming library */ #include "sds.h" /* Dynamic safe strings */ #include "anet.h" /* Networking the easy way */ #include "dict.h" /* Hash tables */ #include "adlist.h" /* Linked lists */ #include "zmalloc.h" /* total memory usage aware version of malloc/free */ +#include "lzf.h" /* LZF compression library */ +#include "pqsort.h" /* Partial qsort for SORT+LIMIT */ /* Error codes */ #define REDIS_OK 0 @@ -61,40 +75,87 @@ /* Static server configuration */ #define REDIS_SERVERPORT 6379 /* TCP port */ #define REDIS_MAXIDLETIME (60*5) /* default client timeout */ -#define REDIS_QUERYBUF_LEN 1024 +#define REDIS_IOBUF_LEN 1024 #define REDIS_LOADBUF_LEN 1024 -#define REDIS_MAX_ARGS 16 +#define REDIS_STATIC_ARGS 4 #define REDIS_DEFAULT_DBNUM 16 #define REDIS_CONFIGLINE_MAX 1024 #define REDIS_OBJFREELIST_MAX 1000000 /* Max number of objects to cache */ #define REDIS_MAX_SYNC_TIME 60 /* Slave can't take more to sync */ +#define REDIS_EXPIRELOOKUPS_PER_CRON 100 /* try to expire 100 keys/second */ +#define REDIS_MAX_WRITE_PER_EVENT (1024*64) +#define REDIS_REQUEST_MAX_SIZE (1024*1024*256) /* max bytes in inline command */ /* Hash table parameters */ #define REDIS_HT_MINFILL 10 /* Minimal hash table fill 10% */ -#define REDIS_HT_MINSLOTS 16384 /* Never resize the HT under this */ /* Command flags */ -#define REDIS_CMD_BULK 1 -#define REDIS_CMD_INLINE 2 +#define REDIS_CMD_BULK 1 /* Bulk write command */ +#define REDIS_CMD_INLINE 2 /* Inline command */ +/* REDIS_CMD_DENYOOM reserves a longer comment: all the commands marked with + this flags will return an error when the 'maxmemory' option is set in the + config file and the server is using more than maxmemory bytes of memory. + In short this commands are denied on low memory conditions. */ +#define REDIS_CMD_DENYOOM 4 /* Object types */ #define REDIS_STRING 0 #define REDIS_LIST 1 #define REDIS_SET 2 #define REDIS_HASH 3 + +/* Object types only used for dumping to disk */ +#define REDIS_EXPIRETIME 253 #define REDIS_SELECTDB 254 #define REDIS_EOF 255 +/* Defines related to the dump file format. To store 32 bits lengths for short + * keys requires a lot of space, so we check the most significant 2 bits of + * the first byte to interpreter the length: + * + * 00|000000 => if the two MSB are 00 the len is the 6 bits of this byte + * 01|000000 00000000 => 01, the len is 14 byes, 6 bits + 8 bits of next byte + * 10|000000 [32 bit integer] => if it's 01, a full 32 bit len will follow + * 11|000000 this means: specially encoded object will follow. The six bits + * number specify the kind of object that follows. + * See the REDIS_RDB_ENC_* defines. + * + * Lenghts up to 63 are stored using a single byte, most DB keys, and may + * values, will fit inside. */ +#define REDIS_RDB_6BITLEN 0 +#define REDIS_RDB_14BITLEN 1 +#define REDIS_RDB_32BITLEN 2 +#define REDIS_RDB_ENCVAL 3 +#define REDIS_RDB_LENERR UINT_MAX + +/* When a length of a string object stored on disk has the first two bits + * set, the remaining two bits specify a special encoding for the object + * accordingly to the following defines: */ +#define REDIS_RDB_ENC_INT8 0 /* 8 bit signed integer */ +#define REDIS_RDB_ENC_INT16 1 /* 16 bit signed integer */ +#define REDIS_RDB_ENC_INT32 2 /* 32 bit signed integer */ +#define REDIS_RDB_ENC_LZF 3 /* string compressed with FASTLZ */ + /* Client flags */ #define REDIS_CLOSE 1 /* This client connection should be closed ASAP */ #define REDIS_SLAVE 2 /* This client is a slave server */ #define REDIS_MASTER 4 /* This client is a master server */ +#define REDIS_MONITOR 8 /* This client is a slave monitor, see MONITOR */ -/* Server replication state */ +/* Slave replication state - slave side */ #define REDIS_REPL_NONE 0 /* No active replication */ #define REDIS_REPL_CONNECT 1 /* Must connect to master */ #define REDIS_REPL_CONNECTED 2 /* Connected to master */ +/* Slave replication state - from the point of view of master + * Note that in SEND_BULK and ONLINE state the slave receives new updates + * in its output queue. In the WAIT_BGSAVE state instead the server is waiting + * to start the next background saving in order to send updates to it. */ +#define REDIS_REPL_WAIT_BGSAVE_START 3 /* master waits bgsave to start feeding it */ +#define REDIS_REPL_WAIT_BGSAVE_END 4 /* master waits bgsave to start bulk DB transmission */ +#define REDIS_REPL_SEND_BULK 5 /* master is sending the bulk DB */ +#define REDIS_REPL_ONLINE 6 /* bulk DB already transmitted, receive updates */ + /* List related stuff */ #define REDIS_HEAD 0 #define REDIS_TAIL 1 @@ -116,30 +177,42 @@ /* Anti-warning macro... */ #define REDIS_NOTUSED(V) ((void) V) + /*================================= Data types ============================== */ /* A redis object, that is a type able to hold a string / list / set */ typedef struct redisObject { - int type; void *ptr; + int type; int refcount; } robj; +typedef struct redisDb { + dict *dict; + dict *expires; + int id; +} redisDb; + /* With multiplexing we need to take per-clinet state. * Clients are taken in a liked list. */ typedef struct redisClient { int fd; - dict *dict; + redisDb *db; int dictid; sds querybuf; - robj *argv[REDIS_MAX_ARGS]; + robj **argv; int argc; - int bulklen; /* bulk read len. -1 if not in bulk read mode */ + int bulklen; /* bulk read len. -1 if not in bulk read mode */ list *reply; int sentlen; time_t lastinteraction; /* time of the last interaction, used for timeout */ - int flags; /* REDIS_CLOSE | REDIS_SLAVE */ - int slaveseldb; /* slave selected db, if this client is a slave */ + int flags; /* REDIS_CLOSE | REDIS_SLAVE | REDIS_MONITOR */ + int slaveseldb; /* slave selected db, if this client is a slave */ + int authenticated; /* when requirepass is non-NULL */ + int replstate; /* replication state if this is a slave */ + int repldbfd; /* replication DB file descriptor */ + long repldboff; /* replication DB file offset */ + off_t repldbsize; /* replication DB file size */ } redisClient; struct saveparam { @@ -151,16 +224,18 @@ struct saveparam { struct redisServer { int port; int fd; - dict **dict; + redisDb *db; + dict *sharingpool; + unsigned int sharingpoolsize; long long dirty; /* changes to DB from the last save */ list *clients; - list *slaves; + list *slaves, *monitors; char neterr[ANET_ERR_LEN]; aeEventLoop *el; int cronloops; /* number of times the cron function run */ list *objfreelist; /* A list of freed objects to avoid malloc() */ time_t lastsave; /* Unix time of last save succeeede */ - int usedmemory; /* Used memory in megabytes */ + size_t usedmemory; /* Used memory in megabytes */ /* Fields used only for stats */ time_t stat_starttime; /* server start time */ long long stat_numcommands; /* number of processed commands */ @@ -171,18 +246,24 @@ struct redisServer { int maxidletime; int dbnum; int daemonize; + char *pidfile; int bgsaveinprogress; + pid_t bgsavechildpid; struct saveparam *saveparams; int saveparamslen; char *logfile; char *bindaddr; char *dbfilename; + char *requirepass; + int shareobjects; /* Replication related */ int isslave; char *masterhost; int masterport; - redisClient *master; + redisClient *master; /* client that is master for this slave */ int replstate; + unsigned int maxclients; + unsigned int maxmemory; /* Sort parameters - qsort_r() is only available under BSD so we * have to take this state global, in order to pass it to sortCompare() */ int sort_desc; @@ -198,6 +279,11 @@ struct redisCommand { int flags; }; +struct redisFunctionSym { + char *name; + unsigned long pointer; +}; + typedef struct _redisSortObject { robj *obj; union { @@ -212,10 +298,10 @@ typedef struct _redisSortOperation { } redisSortOperation; struct sharedObjectsStruct { - robj *crlf, *ok, *err, *zerobulk, *nil, *zero, *one, *pong, *space, - *minus1, *minus2, *minus3, *minus4, - *wrongtypeerr, *nokeyerr, *wrongtypeerrbulk, *nokeyerrbulk, - *syntaxerr, *syntaxerrbulk, + robj *crlf, *ok, *err, *emptybulk, *czero, *cone, *pong, *space, + *colon, *nullbulk, *nullmultibulk, + *emptymultibulk, *wrongtypeerr, *nokeyerr, *syntaxerr, *sameobjecterr, + *outofrangeerr, *plus, *select0, *select1, *select2, *select3, *select4, *select5, *select6, *select7, *select8, *select9; } shared; @@ -228,15 +314,28 @@ static void freeSetObject(robj *o); static void decrRefCount(void *o); static robj *createObject(int type, void *ptr); static void freeClient(redisClient *c); -static int loadDb(char *filename); +static int rdbLoad(char *filename); static void addReply(redisClient *c, robj *obj); static void addReplySds(redisClient *c, sds s); static void incrRefCount(robj *o); -static int saveDbBackground(char *filename); +static int rdbSaveBackground(char *filename); static robj *createStringObject(char *ptr, size_t len); -static void replicationFeedSlaves(struct redisCommand *cmd, int dictid, robj **argv, int argc); +static void replicationFeedSlaves(list *slaves, struct redisCommand *cmd, int dictid, robj **argv, int argc); static int syncWithMaster(void); - +static robj *tryObjectSharing(robj *o); +static int removeExpire(redisDb *db, robj *key); +static int expireIfNeeded(redisDb *db, robj *key); +static int deleteIfVolatile(redisDb *db, robj *key); +static int deleteKey(redisDb *db, robj *key); +static time_t getExpire(redisDb *db, robj *key); +static int setExpire(redisDb *db, robj *key, time_t when); +static void updateSlavesWaitingBgsave(int bgsaveerr); +static void freeMemoryIfNeeded(void); +static int processCommand(redisClient *c); +static void setupSigSegvAction(void); +static void rdbRemoveTempFile(pid_t childpid); + +static void authCommand(redisClient *c); static void pingCommand(redisClient *c); static void echoCommand(redisClient *c); static void setCommand(redisClient *c); @@ -271,10 +370,16 @@ static void typeCommand(redisClient *c); static void lsetCommand(redisClient *c); static void saddCommand(redisClient *c); static void sremCommand(redisClient *c); +static void smoveCommand(redisClient *c); static void sismemberCommand(redisClient *c); static void scardCommand(redisClient *c); +static void spopCommand(redisClient *c); static void sinterCommand(redisClient *c); static void sinterstoreCommand(redisClient *c); +static void sunionCommand(redisClient *c); +static void sunionstoreCommand(redisClient *c); +static void sdiffCommand(redisClient *c); +static void sdiffstoreCommand(redisClient *c); static void syncCommand(redisClient *c); static void flushdbCommand(redisClient *c); static void flushallCommand(redisClient *c); @@ -282,46 +387,60 @@ static void sortCommand(redisClient *c); static void lremCommand(redisClient *c); static void infoCommand(redisClient *c); static void mgetCommand(redisClient *c); - +static void monitorCommand(redisClient *c); +static void expireCommand(redisClient *c); +static void getSetCommand(redisClient *c); +static void ttlCommand(redisClient *c); +static void slaveofCommand(redisClient *c); +static void debugCommand(redisClient *c); /*================================= Globals ================================= */ /* Global vars */ static struct redisServer server; /* server global state */ static struct redisCommand cmdTable[] = { {"get",getCommand,2,REDIS_CMD_INLINE}, - {"set",setCommand,3,REDIS_CMD_BULK}, - {"setnx",setnxCommand,3,REDIS_CMD_BULK}, - {"del",delCommand,2,REDIS_CMD_INLINE}, + {"set",setCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM}, + {"setnx",setnxCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM}, + {"del",delCommand,-2,REDIS_CMD_INLINE}, {"exists",existsCommand,2,REDIS_CMD_INLINE}, - {"incr",incrCommand,2,REDIS_CMD_INLINE}, - {"decr",decrCommand,2,REDIS_CMD_INLINE}, + {"incr",incrCommand,2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM}, + {"decr",decrCommand,2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM}, {"mget",mgetCommand,-2,REDIS_CMD_INLINE}, - {"rpush",rpushCommand,3,REDIS_CMD_BULK}, - {"lpush",lpushCommand,3,REDIS_CMD_BULK}, + {"rpush",rpushCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM}, + {"lpush",lpushCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM}, {"rpop",rpopCommand,2,REDIS_CMD_INLINE}, {"lpop",lpopCommand,2,REDIS_CMD_INLINE}, {"llen",llenCommand,2,REDIS_CMD_INLINE}, {"lindex",lindexCommand,3,REDIS_CMD_INLINE}, - {"lset",lsetCommand,4,REDIS_CMD_BULK}, + {"lset",lsetCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM}, {"lrange",lrangeCommand,4,REDIS_CMD_INLINE}, {"ltrim",ltrimCommand,4,REDIS_CMD_INLINE}, {"lrem",lremCommand,4,REDIS_CMD_BULK}, - {"sadd",saddCommand,3,REDIS_CMD_BULK}, + {"sadd",saddCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM}, {"srem",sremCommand,3,REDIS_CMD_BULK}, + {"smove",smoveCommand,4,REDIS_CMD_BULK}, {"sismember",sismemberCommand,3,REDIS_CMD_BULK}, {"scard",scardCommand,2,REDIS_CMD_INLINE}, - {"sinter",sinterCommand,-2,REDIS_CMD_INLINE}, - {"sinterstore",sinterstoreCommand,-3,REDIS_CMD_INLINE}, + {"spop",spopCommand,2,REDIS_CMD_INLINE}, + {"sinter",sinterCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM}, + {"sinterstore",sinterstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM}, + {"sunion",sunionCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM}, + {"sunionstore",sunionstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM}, + {"sdiff",sdiffCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM}, + {"sdiffstore",sdiffstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM}, {"smembers",sinterCommand,2,REDIS_CMD_INLINE}, - {"incrby",incrbyCommand,3,REDIS_CMD_INLINE}, - {"decrby",decrbyCommand,3,REDIS_CMD_INLINE}, + {"incrby",incrbyCommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM}, + {"decrby",decrbyCommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM}, + {"getset",getSetCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM}, {"randomkey",randomkeyCommand,1,REDIS_CMD_INLINE}, {"select",selectCommand,2,REDIS_CMD_INLINE}, {"move",moveCommand,3,REDIS_CMD_INLINE}, {"rename",renameCommand,3,REDIS_CMD_INLINE}, {"renamenx",renamenxCommand,3,REDIS_CMD_INLINE}, + {"expire",expireCommand,3,REDIS_CMD_INLINE}, {"keys",keysCommand,2,REDIS_CMD_INLINE}, {"dbsize",dbsizeCommand,1,REDIS_CMD_INLINE}, + {"auth",authCommand,2,REDIS_CMD_INLINE}, {"ping",pingCommand,1,REDIS_CMD_INLINE}, {"echo",echoCommand,2,REDIS_CMD_BULK}, {"save",saveCommand,1,REDIS_CMD_INLINE}, @@ -332,11 +451,14 @@ static struct redisCommand cmdTable[] = { {"sync",syncCommand,1,REDIS_CMD_INLINE}, {"flushdb",flushdbCommand,1,REDIS_CMD_INLINE}, {"flushall",flushallCommand,1,REDIS_CMD_INLINE}, - {"sort",sortCommand,-2,REDIS_CMD_INLINE}, + {"sort",sortCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM}, {"info",infoCommand,1,REDIS_CMD_INLINE}, + {"monitor",monitorCommand,1,REDIS_CMD_INLINE}, + {"ttl",ttlCommand,2,REDIS_CMD_INLINE}, + {"slaveof",slaveofCommand,3,REDIS_CMD_INLINE}, + {"debug",debugCommand,-2,REDIS_CMD_INLINE}, {NULL,NULL,0,0} }; - /*============================ Utility functions ============================ */ /* Glob-style pattern matching. */ @@ -462,8 +584,7 @@ int stringmatchlen(const char *pattern, int patternLen, return 0; } -void redisLog(int level, const char *fmt, ...) -{ +static void redisLog(int level, const char *fmt, ...) { va_list ap; FILE *fp; @@ -473,7 +594,12 @@ void redisLog(int level, const char *fmt, ...) va_start(ap, fmt); if (level >= server.verbosity) { char *c = ".-*"; - fprintf(fp,"%c ",c[level]); + char buf[64]; + time_t now; + + now = time(NULL); + strftime(buf,64,"%d %b %H:%M:%S",gmtime(&now)); + fprintf(fp,"%s %c ",buf,c[level]); vfprintf(fp, fmt, ap); fprintf(fp,"\n"); fflush(fp); @@ -553,27 +679,50 @@ static void oom(const char *msg) { } /* ====================== Redis server networking stuff ===================== */ -void closeTimedoutClients(void) { +static void closeTimedoutClients(void) { redisClient *c; - listIter *li; listNode *ln; time_t now = time(NULL); - li = listGetIterator(server.clients,AL_START_HEAD); - if (!li) return; - while ((ln = listNextElement(li)) != NULL) { + listRewind(server.clients); + while ((ln = listYield(server.clients)) != NULL) { c = listNodeValue(ln); if (!(c->flags & REDIS_SLAVE) && /* no timeout for slaves */ + !(c->flags & REDIS_MASTER) && /* no timeout for masters */ (now - c->lastinteraction > server.maxidletime)) { redisLog(REDIS_DEBUG,"Closing idle client"); freeClient(c); } } - listReleaseIterator(li); } -int serverCron(struct aeEventLoop *eventLoop, long long id, void *clientData) { - int j, size, used, loops = server.cronloops++; +static int htNeedsResize(dict *dict) { + long long size, used; + + size = dictSlots(dict); + used = dictSize(dict); + return (size && used && size > DICT_HT_INITIAL_SIZE && + (used*100/size < REDIS_HT_MINFILL)); +} + +/* If the percentage of used slots in the HT reaches REDIS_HT_MINFILL + * we resize the hash table to save memory */ +static void tryResizeHashTables(void) { + int j; + + for (j = 0; j < server.dbnum; j++) { + if (htNeedsResize(server.db[j].dict)) { + redisLog(REDIS_DEBUG,"The hash table %d is too sparse, resize it...",j); + dictResize(server.db[j].dict); + redisLog(REDIS_DEBUG,"Hash table %d resized.",j); + } + if (htNeedsResize(server.db[j].expires)) + dictResize(server.db[j].expires); + } +} + +static int serverCron(struct aeEventLoop *eventLoop, long long id, void *clientData) { + int j, loops = server.cronloops++; REDIS_NOTUSED(eventLoop); REDIS_NOTUSED(id); REDIS_NOTUSED(clientData); @@ -581,33 +730,38 @@ int serverCron(struct aeEventLoop *eventLoop, long long id, void *clientData) { /* Update the global state with the amount of used memory */ server.usedmemory = zmalloc_used_memory(); - /* If the percentage of used slots in the HT reaches REDIS_HT_MINFILL - * we resize the hash table to save memory */ + /* Show some info about non-empty databases */ for (j = 0; j < server.dbnum; j++) { - size = dictGetHashTableSize(server.dict[j]); - used = dictGetHashTableUsed(server.dict[j]); - if (!(loops % 5) && used > 0) { - redisLog(REDIS_DEBUG,"DB %d: %d keys in %d slots HT.",j,used,size); - // dictPrintStats(server.dict); - } - if (size && used && size > REDIS_HT_MINSLOTS && - (used*100/size < REDIS_HT_MINFILL)) { - redisLog(REDIS_NOTICE,"The hash table %d is too sparse, resize it...",j); - dictResize(server.dict[j]); - redisLog(REDIS_NOTICE,"Hash table %d resized.",j); + long long size, used, vkeys; + + size = dictSlots(server.db[j].dict); + used = dictSize(server.db[j].dict); + vkeys = dictSize(server.db[j].expires); + if (!(loops % 5) && (used || vkeys)) { + redisLog(REDIS_DEBUG,"DB %d: %lld keys (%lld volatile) in %lld slots HT.",j,used,vkeys,size); + /* dictPrintStats(server.dict); */ } } + /* We don't want to resize the hash tables while a bacground saving + * is in progress: the saving child is created using fork() that is + * implemented with a copy-on-write semantic in most modern systems, so + * if we resize the HT while there is the saving child at work actually + * a lot of memory movements in the parent will cause a lot of pages + * copied. */ + if (!server.bgsaveinprogress) tryResizeHashTables(); + /* Show information about connected clients */ if (!(loops % 5)) { - redisLog(REDIS_DEBUG,"%d clients connected (%d slaves), %d bytes in use", + redisLog(REDIS_DEBUG,"%d clients connected (%d slaves), %zu bytes in use, %d shared objects", listLength(server.clients)-listLength(server.slaves), listLength(server.slaves), - server.usedmemory); + server.usedmemory, + dictSize(server.sharingpool)); } /* Close connections of timedout clients */ - if (!(loops % 10)) + if (server.maxidletime && !(loops % 10)) closeTimedoutClients(); /* Check if a background saving in progress terminated */ @@ -615,16 +769,23 @@ int serverCron(struct aeEventLoop *eventLoop, long long id, void *clientData) { int statloc; if (wait4(-1,&statloc,WNOHANG,NULL)) { int exitcode = WEXITSTATUS(statloc); - if (exitcode == 0) { + int bysignal = WIFSIGNALED(statloc); + + if (!bysignal && exitcode == 0) { redisLog(REDIS_NOTICE, "Background saving terminated with success"); server.dirty = 0; server.lastsave = time(NULL); + } else if (!bysignal && exitcode != 0) { + redisLog(REDIS_WARNING, "Background saving error"); } else { redisLog(REDIS_WARNING, - "Background saving error"); + "Background saving terminated by signal"); + rdbRemoveTempFile(server.bgsavechildpid); } server.bgsaveinprogress = 0; + server.bgsavechildpid = -1; + updateSlavesWaitingBgsave(exitcode == 0 ? REDIS_OK : REDIS_ERR); } } else { /* If there is not a background saving in progress check if @@ -637,11 +798,35 @@ int serverCron(struct aeEventLoop *eventLoop, long long id, void *clientData) { now-server.lastsave > sp->seconds) { redisLog(REDIS_NOTICE,"%d changes in %d seconds. Saving...", sp->changes, sp->seconds); - saveDbBackground(server.dbfilename); + rdbSaveBackground(server.dbfilename); break; } } } + + /* Try to expire a few timed out keys */ + for (j = 0; j < server.dbnum; j++) { + redisDb *db = server.db+j; + int num = dictSize(db->expires); + + if (num) { + time_t now = time(NULL); + + if (num > REDIS_EXPIRELOOKUPS_PER_CRON) + num = REDIS_EXPIRELOOKUPS_PER_CRON; + while (num--) { + dictEntry *de; + time_t t; + + if ((de = dictGetRandomKey(db->expires)) == NULL) break; + t = (time_t) dictGetEntryVal(de); + if (now > t) { + deleteKey(db,dictGetEntryKey(de)); + } + } + } + } + /* Check if we should connect to a MASTER */ if (server.replstate == REDIS_REPL_CONNECT) { redisLog(REDIS_NOTICE,"Connecting to MASTER..."); @@ -656,29 +841,27 @@ static void createSharedObjects(void) { shared.crlf = createObject(REDIS_STRING,sdsnew("\r\n")); shared.ok = createObject(REDIS_STRING,sdsnew("+OK\r\n")); shared.err = createObject(REDIS_STRING,sdsnew("-ERR\r\n")); - shared.zerobulk = createObject(REDIS_STRING,sdsnew("0\r\n\r\n")); - shared.nil = createObject(REDIS_STRING,sdsnew("nil\r\n")); - shared.zero = createObject(REDIS_STRING,sdsnew("0\r\n")); - shared.one = createObject(REDIS_STRING,sdsnew("1\r\n")); + shared.emptybulk = createObject(REDIS_STRING,sdsnew("$0\r\n\r\n")); + shared.czero = createObject(REDIS_STRING,sdsnew(":0\r\n")); + shared.cone = createObject(REDIS_STRING,sdsnew(":1\r\n")); + shared.nullbulk = createObject(REDIS_STRING,sdsnew("$-1\r\n")); + shared.nullmultibulk = createObject(REDIS_STRING,sdsnew("*-1\r\n")); + shared.emptymultibulk = createObject(REDIS_STRING,sdsnew("*0\r\n")); /* no such key */ - shared.minus1 = createObject(REDIS_STRING,sdsnew("-1\r\n")); - /* operation against key holding a value of the wrong type */ - shared.minus2 = createObject(REDIS_STRING,sdsnew("-2\r\n")); - /* src and dest objects are the same */ - shared.minus3 = createObject(REDIS_STRING,sdsnew("-3\r\n")); - /* out of range argument */ - shared.minus4 = createObject(REDIS_STRING,sdsnew("-4\r\n")); shared.pong = createObject(REDIS_STRING,sdsnew("+PONG\r\n")); shared.wrongtypeerr = createObject(REDIS_STRING,sdsnew( "-ERR Operation against a key holding the wrong kind of value\r\n")); - shared.wrongtypeerrbulk = createObject(REDIS_STRING,sdscatprintf(sdsempty(),"%d\r\n%s",-sdslen(shared.wrongtypeerr->ptr)+2,shared.wrongtypeerr->ptr)); shared.nokeyerr = createObject(REDIS_STRING,sdsnew( "-ERR no such key\r\n")); - shared.nokeyerrbulk = createObject(REDIS_STRING,sdscatprintf(sdsempty(),"%d\r\n%s",-sdslen(shared.nokeyerr->ptr)+2,shared.nokeyerr->ptr)); shared.syntaxerr = createObject(REDIS_STRING,sdsnew( "-ERR syntax error\r\n")); - shared.syntaxerrbulk = createObject(REDIS_STRING,sdscatprintf(sdsempty(),"%d\r\n%s",-sdslen(shared.syntaxerr->ptr)+2,shared.syntaxerr->ptr)); + shared.sameobjecterr = createObject(REDIS_STRING,sdsnew( + "-ERR source and destination objects are the same\r\n")); + shared.outofrangeerr = createObject(REDIS_STRING,sdsnew( + "-ERR index out of range\r\n")); shared.space = createObject(REDIS_STRING,sdsnew(" ")); + shared.colon = createObject(REDIS_STRING,sdsnew(":")); + shared.plus = createObject(REDIS_STRING,sdsnew("+")); shared.select0 = createStringObject("select 0\r\n",10); shared.select1 = createStringObject("select 1\r\n",10); shared.select2 = createStringObject("select 2\r\n",10); @@ -715,7 +898,13 @@ static void initServerConfig() { server.bindaddr = NULL; server.glueoutputbuf = 1; server.daemonize = 0; + server.pidfile = "/var/run/redis.pid"; server.dbfilename = "dump.rdb"; + server.requirepass = NULL; + server.shareobjects = 0; + server.sharingpoolsize = 1024; + server.maxclients = 0; + server.maxmemory = 0; ResetServerSaveParams(); appendServerSaveParams(60*60,1); /* save after 1 hour and 1 change */ @@ -734,14 +923,17 @@ static void initServer() { signal(SIGHUP, SIG_IGN); signal(SIGPIPE, SIG_IGN); + setupSigSegvAction(); server.clients = listCreate(); server.slaves = listCreate(); + server.monitors = listCreate(); server.objfreelist = listCreate(); createSharedObjects(); server.el = aeCreateEventLoop(); - server.dict = zmalloc(sizeof(dict*)*server.dbnum); - if (!server.dict || !server.clients || !server.slaves || !server.el || !server.objfreelist) + server.db = zmalloc(sizeof(redisDb)*server.dbnum); + server.sharingpool = dictCreate(&setDictType,NULL); + if (!server.db || !server.clients || !server.slaves || !server.monitors || !server.el || !server.objfreelist) oom("server initialization"); /* Fatal OOM */ server.fd = anetTcpServer(server.neterr, server.port, server.bindaddr); if (server.fd == -1) { @@ -749,12 +941,13 @@ static void initServer() { exit(1); } for (j = 0; j < server.dbnum; j++) { - server.dict[j] = dictCreate(&hashDictType,NULL); - if (!server.dict[j]) - oom("dictCreate"); /* Fatal OOM */ + server.db[j].dict = dictCreate(&hashDictType,NULL); + server.db[j].expires = dictCreate(&setDictType,NULL); + server.db[j].id = j; } server.cronloops = 0; server.bgsaveinprogress = 0; + server.bgsavechildpid = -1; server.lastsave = time(NULL); server.dirty = 0; server.usedmemory = 0; @@ -765,25 +958,41 @@ static void initServer() { } /* Empty the whole database */ -static void emptyDb() { +static long long emptyDb() { int j; + long long removed = 0; + + for (j = 0; j < server.dbnum; j++) { + removed += dictSize(server.db[j].dict); + dictEmpty(server.db[j].dict); + dictEmpty(server.db[j].expires); + } + return removed; +} - for (j = 0; j < server.dbnum; j++) - dictEmpty(server.dict[j]); +static int yesnotoi(char *s) { + if (!strcasecmp(s,"yes")) return 1; + else if (!strcasecmp(s,"no")) return 0; + else return -1; } /* I agree, this is a very rudimental way to load a configuration... will improve later if the config gets more complex */ static void loadServerConfig(char *filename) { - FILE *fp = fopen(filename,"r"); + FILE *fp; char buf[REDIS_CONFIGLINE_MAX+1], *err = NULL; int linenum = 0; sds line = NULL; - - if (!fp) { - redisLog(REDIS_WARNING,"Fatal error, can't open config file"); - exit(1); + + if (filename[0] == '-' && filename[1] == '\0') + fp = stdin; + else { + if ((fp = fopen(filename,"r")) == NULL) { + redisLog(REDIS_WARNING,"Fatal error, can't open config file"); + exit(1); + } } + while(fgets(buf,REDIS_CONFIGLINE_MAX+1,fp) != NULL) { sds *argv; int argc, j; @@ -803,81 +1012,94 @@ static void loadServerConfig(char *filename) { sdstolower(argv[0]); /* Execute config directives */ - if (!strcmp(argv[0],"timeout") && argc == 2) { + if (!strcasecmp(argv[0],"timeout") && argc == 2) { server.maxidletime = atoi(argv[1]); - if (server.maxidletime < 1) { + if (server.maxidletime < 0) { err = "Invalid timeout value"; goto loaderr; } - } else if (!strcmp(argv[0],"port") && argc == 2) { + } else if (!strcasecmp(argv[0],"port") && argc == 2) { server.port = atoi(argv[1]); if (server.port < 1 || server.port > 65535) { err = "Invalid port"; goto loaderr; } - } else if (!strcmp(argv[0],"bind") && argc == 2) { + } else if (!strcasecmp(argv[0],"bind") && argc == 2) { server.bindaddr = zstrdup(argv[1]); - } else if (!strcmp(argv[0],"save") && argc == 3) { + } else if (!strcasecmp(argv[0],"save") && argc == 3) { int seconds = atoi(argv[1]); int changes = atoi(argv[2]); if (seconds < 1 || changes < 0) { err = "Invalid save parameters"; goto loaderr; } appendServerSaveParams(seconds,changes); - } else if (!strcmp(argv[0],"dir") && argc == 2) { + } else if (!strcasecmp(argv[0],"dir") && argc == 2) { if (chdir(argv[1]) == -1) { redisLog(REDIS_WARNING,"Can't chdir to '%s': %s", argv[1], strerror(errno)); exit(1); } - } else if (!strcmp(argv[0],"loglevel") && argc == 2) { - if (!strcmp(argv[1],"debug")) server.verbosity = REDIS_DEBUG; - else if (!strcmp(argv[1],"notice")) server.verbosity = REDIS_NOTICE; - else if (!strcmp(argv[1],"warning")) server.verbosity = REDIS_WARNING; + } else if (!strcasecmp(argv[0],"loglevel") && argc == 2) { + if (!strcasecmp(argv[1],"debug")) server.verbosity = REDIS_DEBUG; + else if (!strcasecmp(argv[1],"notice")) server.verbosity = REDIS_NOTICE; + else if (!strcasecmp(argv[1],"warning")) server.verbosity = REDIS_WARNING; else { err = "Invalid log level. Must be one of debug, notice, warning"; goto loaderr; } - } else if (!strcmp(argv[0],"logfile") && argc == 2) { - FILE *fp; + } else if (!strcasecmp(argv[0],"logfile") && argc == 2) { + FILE *logfp; server.logfile = zstrdup(argv[1]); - if (!strcmp(server.logfile,"stdout")) { + if (!strcasecmp(server.logfile,"stdout")) { zfree(server.logfile); server.logfile = NULL; } if (server.logfile) { /* Test if we are able to open the file. The server will not * be able to abort just for this problem later... */ - fp = fopen(server.logfile,"a"); - if (fp == NULL) { + logfp = fopen(server.logfile,"a"); + if (logfp == NULL) { err = sdscatprintf(sdsempty(), "Can't open the log file: %s", strerror(errno)); goto loaderr; } - fclose(fp); + fclose(logfp); } - } else if (!strcmp(argv[0],"databases") && argc == 2) { + } else if (!strcasecmp(argv[0],"databases") && argc == 2) { server.dbnum = atoi(argv[1]); if (server.dbnum < 1) { err = "Invalid number of databases"; goto loaderr; } - } else if (!strcmp(argv[0],"slaveof") && argc == 3) { + } else if (!strcasecmp(argv[0],"maxclients") && argc == 2) { + server.maxclients = atoi(argv[1]); + } else if (!strcasecmp(argv[0],"maxmemory") && argc == 2) { + server.maxmemory = atoi(argv[1]); + } else if (!strcasecmp(argv[0],"slaveof") && argc == 3) { server.masterhost = sdsnew(argv[1]); server.masterport = atoi(argv[2]); server.replstate = REDIS_REPL_CONNECT; - } else if (!strcmp(argv[0],"glueoutputbuf") && argc == 2) { - sdstolower(argv[1]); - if (!strcmp(argv[1],"yes")) server.glueoutputbuf = 1; - else if (!strcmp(argv[1],"no")) server.glueoutputbuf = 0; - else { + } else if (!strcasecmp(argv[0],"glueoutputbuf") && argc == 2) { + if ((server.glueoutputbuf = yesnotoi(argv[1])) == -1) { err = "argument must be 'yes' or 'no'"; goto loaderr; } - } else if (!strcmp(argv[0],"daemonize") && argc == 2) { - sdstolower(argv[1]); - if (!strcmp(argv[1],"yes")) server.daemonize = 1; - else if (!strcmp(argv[1],"no")) server.daemonize = 0; - else { + } else if (!strcasecmp(argv[0],"shareobjects") && argc == 2) { + if ((server.shareobjects = yesnotoi(argv[1])) == -1) { err = "argument must be 'yes' or 'no'"; goto loaderr; } + } else if (!strcasecmp(argv[0],"shareobjectspoolsize") && argc == 2) { + server.sharingpoolsize = atoi(argv[1]); + if (server.sharingpoolsize < 1) { + err = "invalid object sharing pool size"; goto loaderr; + } + } else if (!strcasecmp(argv[0],"daemonize") && argc == 2) { + if ((server.daemonize = yesnotoi(argv[1])) == -1) { + err = "argument must be 'yes' or 'no'"; goto loaderr; + } + } else if (!strcasecmp(argv[0],"requirepass") && argc == 2) { + server.requirepass = zstrdup(argv[1]); + } else if (!strcasecmp(argv[0],"pidfile") && argc == 2) { + server.pidfile = zstrdup(argv[1]); + } else if (!strcasecmp(argv[0],"dbfilename") && argc == 2) { + server.dbfilename = zstrdup(argv[1]); } else { err = "Bad directive or wrong number of arguments"; goto loaderr; } @@ -886,7 +1108,7 @@ static void loadServerConfig(char *filename) { zfree(argv); sdsfree(line); } - fclose(fp); + if (fp != stdin) fclose(fp); return; loaderr: @@ -918,26 +1140,30 @@ static void freeClient(redisClient *c) { assert(ln != NULL); listDelNode(server.clients,ln); if (c->flags & REDIS_SLAVE) { - ln = listSearchKey(server.slaves,c); + if (c->replstate == REDIS_REPL_SEND_BULK && c->repldbfd != -1) + close(c->repldbfd); + list *l = (c->flags & REDIS_MONITOR) ? server.monitors : server.slaves; + ln = listSearchKey(l,c); assert(ln != NULL); - listDelNode(server.slaves,ln); + listDelNode(l,ln); } if (c->flags & REDIS_MASTER) { server.master = NULL; server.replstate = REDIS_REPL_CONNECT; } + zfree(c->argv); zfree(c); } static void glueReplyBuffersIfNeeded(redisClient *c) { int totlen = 0; - listNode *ln = c->reply->head, *next; + listNode *ln; robj *o; - while(ln) { + listRewind(c->reply); + while((ln = listYield(c->reply))) { o = ln->value; totlen += sdslen(o->ptr); - ln = ln->next; /* This optimization makes more sense if we don't have to copy * too much data */ if (totlen > 1024) return; @@ -946,17 +1172,16 @@ static void glueReplyBuffersIfNeeded(redisClient *c) { char buf[1024]; int copylen = 0; - ln = c->reply->head; - while(ln) { - next = ln->next; + listRewind(c->reply); + while((ln = listYield(c->reply))) { o = ln->value; memcpy(buf+copylen,o->ptr,sdslen(o->ptr)); copylen += sdslen(o->ptr); listDelNode(c->reply,ln); - ln = next; } /* Now the output buffer is empty, add the new single element */ - addReplySds(c,sdsnewlen(buf,totlen)); + o = createObject(REDIS_STRING,sdsnewlen(buf,totlen)); + if (!listAddNodeTail(c->reply,o)) oom("listAddNodeTail"); } } @@ -979,9 +1204,10 @@ static void sendReplyToClient(aeEventLoop *el, int fd, void *privdata, int mask) } if (c->flags & REDIS_MASTER) { + /* Don't reply to a master */ nwritten = objlen - c->sentlen; } else { - nwritten = write(fd, o->ptr+c->sentlen, objlen - c->sentlen); + nwritten = write(fd, ((char*)o->ptr)+c->sentlen, objlen - c->sentlen); if (nwritten <= 0) break; } c->sentlen += nwritten; @@ -991,6 +1217,12 @@ static void sendReplyToClient(aeEventLoop *el, int fd, void *privdata, int mask) listDelNode(c->reply,listFirst(c->reply)); c->sentlen = 0; } + /* Note that we avoid to send more thank REDIS_MAX_WRITE_PER_EVENT + * bytes, in a single threaded server it's a good idea to server + * other clients as well, even if a very large request comes from + * super fast link that is always able to accept data (in real world + * terms think to 'KEYS *' against the loopback interfae) */ + if (totwritten > REDIS_MAX_WRITE_PER_EVENT) break; } if (nwritten == -1) { if (errno == EAGAIN) { @@ -1012,7 +1244,7 @@ static void sendReplyToClient(aeEventLoop *el, int fd, void *privdata, int mask) static struct redisCommand *lookupCommand(char *name) { int j = 0; while(cmdTable[j].name != NULL) { - if (!strcmp(name,cmdTable[j].name)) return &cmdTable[j]; + if (!strcasecmp(name,cmdTable[j].name)) return &cmdTable[j]; j++; } return NULL; @@ -1036,10 +1268,12 @@ static int processCommand(redisClient *c) { struct redisCommand *cmd; long long dirty; - sdstolower(c->argv[0]->ptr); + /* Free some memory if needed (maxmemory setting) */ + if (server.maxmemory) freeMemoryIfNeeded(); + /* The QUIT command is handled as a special case. Normal command * procs are unable to close the client connection safely */ - if (!strcmp(c->argv[0]->ptr,"quit")) { + if (!strcasecmp(c->argv[0]->ptr,"quit")) { freeClient(c); return 0; } @@ -1053,6 +1287,10 @@ static int processCommand(redisClient *c) { addReplySds(c,sdsnew("-ERR wrong number of arguments\r\n")); resetClient(c); return 1; + } else if (server.maxmemory && cmd->flags & REDIS_CMD_DENYOOM && zmalloc_used_memory() > server.maxmemory) { + addReplySds(c,sdsnew("-ERR command not allowed when used memory > 'maxmemory'\r\n")); + resetClient(c); + return 1; } else if (cmd->flags & REDIS_CMD_BULK && c->bulklen == -1) { int bulklen = atoi(c->argv[c->argc-1]->ptr); @@ -1075,11 +1313,26 @@ static int processCommand(redisClient *c) { return 1; } } + /* Let's try to share objects on the command arguments vector */ + if (server.shareobjects) { + int j; + for(j = 1; j < c->argc; j++) + c->argv[j] = tryObjectSharing(c->argv[j]); + } + /* Check if the user is authenticated */ + if (server.requirepass && !c->authenticated && cmd->proc != authCommand) { + addReplySds(c,sdsnew("-ERR operation not permitted\r\n")); + resetClient(c); + return 1; + } + /* Exec the command */ dirty = server.dirty; cmd->proc(c); if (server.dirty-dirty != 0 && listLength(server.slaves)) - replicationFeedSlaves(cmd,c->dictid,c->argv,c->argc); + replicationFeedSlaves(server.slaves,cmd,c->db->id,c->argv,c->argc); + if (listLength(server.monitors)) + replicationFeedSlaves(server.monitors,cmd,c->db->id,c->argv,c->argc); server.stat_numcommands++; /* Prepare the client for the next command */ @@ -1091,10 +1344,19 @@ static int processCommand(redisClient *c) { return 1; } -static void replicationFeedSlaves(struct redisCommand *cmd, int dictid, robj **argv, int argc) { - listNode *ln = server.slaves->head; - robj *outv[REDIS_MAX_ARGS*4]; /* enough room for args, spaces, newlines */ +static void replicationFeedSlaves(list *slaves, struct redisCommand *cmd, int dictid, robj **argv, int argc) { + listNode *ln; int outc = 0, j; + robj **outv; + /* (args*2)+1 is enough room for args, spaces, newlines */ + robj *static_outv[REDIS_STATIC_ARGS*2+1]; + + if (argc <= REDIS_STATIC_ARGS) { + outv = static_outv; + } else { + outv = zmalloc(sizeof(robj*)*(argc*2+1)); + if (!outv) oom("replicationFeedSlaves"); + } for (j = 0; j < argc; j++) { if (j != 0) outv[outc++] = shared.space; @@ -1110,8 +1372,18 @@ static void replicationFeedSlaves(struct redisCommand *cmd, int dictid, robj **a } outv[outc++] = shared.crlf; - while(ln) { + /* Increment all the refcounts at start and decrement at end in order to + * be sure to free objects if there is no slave in a replication state + * able to be feed with commands */ + for (j = 0; j < outc; j++) incrRefCount(outv[j]); + listRewind(slaves); + while((ln = listYield(slaves))) { redisClient *slave = ln->value; + + /* Don't feed slaves that are still waiting for BGSAVE to start */ + if (slave->replstate == REDIS_REPL_WAIT_BGSAVE_START) continue; + + /* Feed all the other slaves, MONITORs and so on */ if (slave->slaveseldb != dictid) { robj *selectcmd; @@ -1136,18 +1408,19 @@ static void replicationFeedSlaves(struct redisCommand *cmd, int dictid, robj **a slave->slaveseldb = dictid; } for (j = 0; j < outc; j++) addReply(slave,outv[j]); - ln = ln->next; } + for (j = 0; j < outc; j++) decrRefCount(outv[j]); + if (outv != static_outv) zfree(outv); } static void readQueryFromClient(aeEventLoop *el, int fd, void *privdata, int mask) { redisClient *c = (redisClient*) privdata; - char buf[REDIS_QUERYBUF_LEN]; + char buf[REDIS_IOBUF_LEN]; int nread; REDIS_NOTUSED(el); REDIS_NOTUSED(mask); - nread = read(fd, buf, REDIS_QUERYBUF_LEN); + nread = read(fd, buf, REDIS_IOBUF_LEN); if (nread == -1) { if (errno == EAGAIN) { nread = 0; @@ -1173,6 +1446,7 @@ again: /* Read the first line of the query */ char *p = strchr(c->querybuf,'\n'); size_t querylen; + if (p) { sds query, *argv; int argc, j; @@ -1195,9 +1469,14 @@ again: return; } argv = sdssplitlen(query,sdslen(query)," ",1,&argc); - sdsfree(query); if (argv == NULL) oom("sdssplitlen"); - for (j = 0; j < argc && j < REDIS_MAX_ARGS; j++) { + sdsfree(query); + + if (c->argv) zfree(c->argv); + c->argv = zmalloc(sizeof(robj*)*argc); + if (c->argv == NULL) oom("allocating arguments list for client"); + + for (j = 0; j < argc; j++) { if (sdslen(argv[j])) { c->argv[c->argc] = createObject(REDIS_STRING,argv[j]); c->argc++; @@ -1209,9 +1488,9 @@ again: /* Execute the command. If the client is still valid * after processCommand() return and there is something * on the query buffer try to process the next command. */ - if (processCommand(c) && sdslen(c->querybuf)) goto again; + if (c->argc && processCommand(c) && sdslen(c->querybuf)) goto again; return; - } else if (sdslen(c->querybuf) >= 1024) { + } else if (sdslen(c->querybuf) >= REDIS_REQUEST_MAX_SIZE) { redisLog(REDIS_DEBUG, "Client protocol error"); freeClient(c); return; @@ -1237,11 +1516,15 @@ again: static int selectDb(redisClient *c, int id) { if (id < 0 || id >= server.dbnum) return REDIS_ERR; - c->dict = server.dict[id]; - c->dictid = id; + c->db = &server.db[id]; return REDIS_OK; } +static void *dupClientReplyValue(void *o) { + incrRefCount((robj*)o); + return 0; +} + static redisClient *createClient(int fd) { redisClient *c = zmalloc(sizeof(*c)); @@ -1252,12 +1535,16 @@ static redisClient *createClient(int fd) { c->fd = fd; c->querybuf = sdsempty(); c->argc = 0; + c->argv = NULL; c->bulklen = -1; c->sentlen = 0; c->flags = 0; c->lastinteraction = time(NULL); + c->authenticated = 0; + c->replstate = REDIS_REPL_NONE; if ((c->reply = listCreate()) == NULL) oom("listCreate"); listSetFreeMethod(c->reply,decrRefCount); + listSetDupMethod(c->reply,dupClientReplyValue); if (aeCreateFileEvent(server.el, c->fd, AE_READABLE, readQueryFromClient, c, NULL) == AE_ERR) { freeClient(c); @@ -1269,6 +1556,8 @@ static redisClient *createClient(int fd) { static void addReply(redisClient *c, robj *obj) { if (listLength(c->reply) == 0 && + (c->replstate == REDIS_REPL_NONE || + c->replstate == REDIS_REPL_ONLINE) && aeCreateFileEvent(server.el, c->fd, AE_WRITABLE, sendReplyToClient, c, NULL) == AE_ERR) return; if (!listAddNodeTail(c->reply,obj)) oom("listAddNodeTail"); @@ -1284,6 +1573,7 @@ static void addReplySds(redisClient *c, sds s) { static void acceptHandler(aeEventLoop *el, int fd, void *privdata, int mask) { int cport, cfd; char cip[128]; + redisClient *c; REDIS_NOTUSED(el); REDIS_NOTUSED(mask); REDIS_NOTUSED(privdata); @@ -1294,11 +1584,23 @@ static void acceptHandler(aeEventLoop *el, int fd, void *privdata, int mask) { return; } redisLog(REDIS_DEBUG,"Accepted %s:%d", cip, cport); - if (createClient(cfd) == NULL) { + if ((c = createClient(cfd)) == NULL) { redisLog(REDIS_WARNING,"Error allocating resoures for the client"); close(cfd); /* May be already closed, just ingore errors */ return; } + /* If maxclient directive is set and this is one client more... close the + * connection. Note that we create the client instead to check before + * for this condition, since now the socket is already set in nonblocking + * mode and we can send an error for free using the Kernel I/O */ + if (server.maxclients && listLength(server.clients) > server.maxclients) { + char *err = "-ERR max number of clients reached\r\n"; + + /* That's a best effort error message, don't check write errors */ + (void) write(c->fd,err,strlen(err)); + freeClient(c); + return; + } server.stat_numconnections++; } @@ -1339,14 +1641,6 @@ static robj *createSetObject(void) { return createObject(REDIS_SET,d); } -#if 0 -static robj *createHashObject(void) { - dict *d = dictCreate(&hashDictType,NULL); - if (!d) oom("dictCreate"); - return createObject(REDIS_SET,d); -} -#endif - static void freeStringObject(robj *o) { sdsfree(o->ptr); } @@ -1365,10 +1659,19 @@ static void freeHashObject(robj *o) { static void incrRefCount(robj *o) { o->refcount++; +#ifdef DEBUG_REFCOUNT + if (o->type == REDIS_STRING) + printf("Increment '%s'(%p), now is: %d\n",o->ptr,o,o->refcount); +#endif } static void decrRefCount(void *obj) { robj *o = obj; + +#ifdef DEBUG_REFCOUNT + if (o->type == REDIS_STRING) + printf("Decrement '%s'(%p), now is: %d\n",o->ptr,o,o->refcount-1); +#endif if (--(o->refcount) == 0) { switch(o->type) { case REDIS_STRING: freeStringObject(o); break; @@ -1383,28 +1686,235 @@ static void decrRefCount(void *obj) { } } +/* Try to share an object against the shared objects pool */ +static robj *tryObjectSharing(robj *o) { + struct dictEntry *de; + unsigned long c; + + if (o == NULL || server.shareobjects == 0) return o; + + assert(o->type == REDIS_STRING); + de = dictFind(server.sharingpool,o); + if (de) { + robj *shared = dictGetEntryKey(de); + + c = ((unsigned long) dictGetEntryVal(de))+1; + dictGetEntryVal(de) = (void*) c; + incrRefCount(shared); + decrRefCount(o); + return shared; + } else { + /* Here we are using a stream algorihtm: Every time an object is + * shared we increment its count, everytime there is a miss we + * recrement the counter of a random object. If this object reaches + * zero we remove the object and put the current object instead. */ + if (dictSize(server.sharingpool) >= + server.sharingpoolsize) { + de = dictGetRandomKey(server.sharingpool); + assert(de != NULL); + c = ((unsigned long) dictGetEntryVal(de))-1; + dictGetEntryVal(de) = (void*) c; + if (c == 0) { + dictDelete(server.sharingpool,de->key); + } + } else { + c = 0; /* If the pool is empty we want to add this object */ + } + if (c == 0) { + int retval; + + retval = dictAdd(server.sharingpool,o,(void*)1); + assert(retval == DICT_OK); + incrRefCount(o); + } + return o; + } +} + +static robj *lookupKey(redisDb *db, robj *key) { + dictEntry *de = dictFind(db->dict,key); + return de ? dictGetEntryVal(de) : NULL; +} + +static robj *lookupKeyRead(redisDb *db, robj *key) { + expireIfNeeded(db,key); + return lookupKey(db,key); +} + +static robj *lookupKeyWrite(redisDb *db, robj *key) { + deleteIfVolatile(db,key); + return lookupKey(db,key); +} + +static int deleteKey(redisDb *db, robj *key) { + int retval; + + /* We need to protect key from destruction: after the first dictDelete() + * it may happen that 'key' is no longer valid if we don't increment + * it's count. This may happen when we get the object reference directly + * from the hash table with dictRandomKey() or dict iterators */ + incrRefCount(key); + if (dictSize(db->expires)) dictDelete(db->expires,key); + retval = dictDelete(db->dict,key); + decrRefCount(key); + + return retval == DICT_OK; +} + /*============================ DB saving/loading ============================ */ +static int rdbSaveType(FILE *fp, unsigned char type) { + if (fwrite(&type,1,1,fp) == 0) return -1; + return 0; +} + +static int rdbSaveTime(FILE *fp, time_t t) { + int32_t t32 = (int32_t) t; + if (fwrite(&t32,4,1,fp) == 0) return -1; + return 0; +} + +/* check rdbLoadLen() comments for more info */ +static int rdbSaveLen(FILE *fp, uint32_t len) { + unsigned char buf[2]; + + if (len < (1<<6)) { + /* Save a 6 bit len */ + buf[0] = (len&0xFF)|(REDIS_RDB_6BITLEN<<6); + if (fwrite(buf,1,1,fp) == 0) return -1; + } else if (len < (1<<14)) { + /* Save a 14 bit len */ + buf[0] = ((len>>8)&0xFF)|(REDIS_RDB_14BITLEN<<6); + buf[1] = len&0xFF; + if (fwrite(buf,2,1,fp) == 0) return -1; + } else { + /* Save a 32 bit len */ + buf[0] = (REDIS_RDB_32BITLEN<<6); + if (fwrite(buf,1,1,fp) == 0) return -1; + len = htonl(len); + if (fwrite(&len,4,1,fp) == 0) return -1; + } + return 0; +} + +/* String objects in the form "2391" "-100" without any space and with a + * range of values that can fit in an 8, 16 or 32 bit signed value can be + * encoded as integers to save space */ +static int rdbTryIntegerEncoding(sds s, unsigned char *enc) { + long long value; + char *endptr, buf[32]; + + /* Check if it's possible to encode this value as a number */ + value = strtoll(s, &endptr, 10); + if (endptr[0] != '\0') return 0; + snprintf(buf,32,"%lld",value); + + /* If the number converted back into a string is not identical + * then it's not possible to encode the string as integer */ + if (strlen(buf) != sdslen(s) || memcmp(buf,s,sdslen(s))) return 0; + + /* Finally check if it fits in our ranges */ + if (value >= -(1<<7) && value <= (1<<7)-1) { + enc[0] = (REDIS_RDB_ENCVAL<<6)|REDIS_RDB_ENC_INT8; + enc[1] = value&0xFF; + return 2; + } else if (value >= -(1<<15) && value <= (1<<15)-1) { + enc[0] = (REDIS_RDB_ENCVAL<<6)|REDIS_RDB_ENC_INT16; + enc[1] = value&0xFF; + enc[2] = (value>>8)&0xFF; + return 3; + } else if (value >= -((long long)1<<31) && value <= ((long long)1<<31)-1) { + enc[0] = (REDIS_RDB_ENCVAL<<6)|REDIS_RDB_ENC_INT32; + enc[1] = value&0xFF; + enc[2] = (value>>8)&0xFF; + enc[3] = (value>>16)&0xFF; + enc[4] = (value>>24)&0xFF; + return 5; + } else { + return 0; + } +} + +static int rdbSaveLzfStringObject(FILE *fp, robj *obj) { + unsigned int comprlen, outlen; + unsigned char byte; + void *out; + + /* We require at least four bytes compression for this to be worth it */ + outlen = sdslen(obj->ptr)-4; + if (outlen <= 0) return 0; + if ((out = zmalloc(outlen+1)) == NULL) return 0; + comprlen = lzf_compress(obj->ptr, sdslen(obj->ptr), out, outlen); + if (comprlen == 0) { + zfree(out); + return 0; + } + /* Data compressed! Let's save it on disk */ + byte = (REDIS_RDB_ENCVAL<<6)|REDIS_RDB_ENC_LZF; + if (fwrite(&byte,1,1,fp) == 0) goto writeerr; + if (rdbSaveLen(fp,comprlen) == -1) goto writeerr; + if (rdbSaveLen(fp,sdslen(obj->ptr)) == -1) goto writeerr; + if (fwrite(out,comprlen,1,fp) == 0) goto writeerr; + zfree(out); + return comprlen; + +writeerr: + zfree(out); + return -1; +} + +/* Save a string objet as [len][data] on disk. If the object is a string + * representation of an integer value we try to safe it in a special form */ +static int rdbSaveStringObject(FILE *fp, robj *obj) { + size_t len = sdslen(obj->ptr); + int enclen; + + /* Try integer encoding */ + if (len <= 11) { + unsigned char buf[5]; + if ((enclen = rdbTryIntegerEncoding(obj->ptr,buf)) > 0) { + if (fwrite(buf,enclen,1,fp) == 0) return -1; + return 0; + } + } + + /* Try LZF compression - under 20 bytes it's unable to compress even + * aaaaaaaaaaaaaaaaaa so skip it */ + if (1 && len > 20) { + int retval; + + retval = rdbSaveLzfStringObject(fp,obj); + if (retval == -1) return -1; + if (retval > 0) return 0; + /* retval == 0 means data can't be compressed, save the old way */ + } + + /* Store verbatim */ + if (rdbSaveLen(fp,len) == -1) return -1; + if (len && fwrite(obj->ptr,len,1,fp) == 0) return -1; + return 0; +} + /* Save the DB on disk. Return REDIS_ERR on error, REDIS_OK on success */ -static int saveDb(char *filename) { +static int rdbSave(char *filename) { dictIterator *di = NULL; dictEntry *de; - uint32_t len; - uint8_t type; FILE *fp; char tmpfile[256]; int j; + time_t now = time(NULL); - snprintf(tmpfile,256,"temp-%d.%ld.rdb",(int)time(NULL),(long int)random()); + snprintf(tmpfile,256,"temp-%d.rdb", (int) getpid()); fp = fopen(tmpfile,"w"); if (!fp) { redisLog(REDIS_WARNING, "Failed saving the DB: %s", strerror(errno)); return REDIS_ERR; } - if (fwrite("REDIS0000",9,1,fp) == 0) goto werr; + if (fwrite("REDIS0001",9,1,fp) == 0) goto werr; for (j = 0; j < server.dbnum; j++) { - dict *d = server.dict[j]; - if (dictGetHashTableUsed(d) == 0) continue; + redisDb *db = server.db+j; + dict *d = db->dict; + if (dictSize(d) == 0) continue; di = dictGetIterator(d); if (!di) { fclose(fp); @@ -1412,60 +1922,52 @@ static int saveDb(char *filename) { } /* Write the SELECT DB opcode */ - type = REDIS_SELECTDB; - len = htonl(j); - if (fwrite(&type,1,1,fp) == 0) goto werr; - if (fwrite(&len,4,1,fp) == 0) goto werr; + if (rdbSaveType(fp,REDIS_SELECTDB) == -1) goto werr; + if (rdbSaveLen(fp,j) == -1) goto werr; /* Iterate this DB writing every entry */ while((de = dictNext(di)) != NULL) { robj *key = dictGetEntryKey(de); robj *o = dictGetEntryVal(de); - - type = o->type; - len = htonl(sdslen(key->ptr)); - if (fwrite(&type,1,1,fp) == 0) goto werr; - if (fwrite(&len,4,1,fp) == 0) goto werr; - if (fwrite(key->ptr,sdslen(key->ptr),1,fp) == 0) goto werr; - if (type == REDIS_STRING) { + time_t expiretime = getExpire(db,key); + + /* Save the expire time */ + if (expiretime != -1) { + /* If this key is already expired skip it */ + if (expiretime < now) continue; + if (rdbSaveType(fp,REDIS_EXPIRETIME) == -1) goto werr; + if (rdbSaveTime(fp,expiretime) == -1) goto werr; + } + /* Save the key and associated value */ + if (rdbSaveType(fp,o->type) == -1) goto werr; + if (rdbSaveStringObject(fp,key) == -1) goto werr; + if (o->type == REDIS_STRING) { /* Save a string value */ - sds sval = o->ptr; - len = htonl(sdslen(sval)); - if (fwrite(&len,4,1,fp) == 0) goto werr; - if (sdslen(sval) && - fwrite(sval,sdslen(sval),1,fp) == 0) goto werr; - } else if (type == REDIS_LIST) { + if (rdbSaveStringObject(fp,o) == -1) goto werr; + } else if (o->type == REDIS_LIST) { /* Save a list value */ list *list = o->ptr; - listNode *ln = list->head; + listNode *ln; - len = htonl(listLength(list)); - if (fwrite(&len,4,1,fp) == 0) goto werr; - while(ln) { + listRewind(list); + if (rdbSaveLen(fp,listLength(list)) == -1) goto werr; + while((ln = listYield(list))) { robj *eleobj = listNodeValue(ln); - len = htonl(sdslen(eleobj->ptr)); - if (fwrite(&len,4,1,fp) == 0) goto werr; - if (sdslen(eleobj->ptr) && fwrite(eleobj->ptr,sdslen(eleobj->ptr),1,fp) == 0) - goto werr; - ln = ln->next; + + if (rdbSaveStringObject(fp,eleobj) == -1) goto werr; } - } else if (type == REDIS_SET) { + } else if (o->type == REDIS_SET) { /* Save a set value */ dict *set = o->ptr; dictIterator *di = dictGetIterator(set); dictEntry *de; if (!set) oom("dictGetIteraotr"); - len = htonl(dictGetHashTableUsed(set)); - if (fwrite(&len,4,1,fp) == 0) goto werr; + if (rdbSaveLen(fp,dictSize(set)) == -1) goto werr; while((de = dictNext(di)) != NULL) { - robj *eleobj; + robj *eleobj = dictGetEntryKey(de); - eleobj = dictGetEntryKey(de); - len = htonl(sdslen(eleobj->ptr)); - if (fwrite(&len,4,1,fp) == 0) goto werr; - if (sdslen(eleobj->ptr) && fwrite(eleobj->ptr,sdslen(eleobj->ptr),1,fp) == 0) - goto werr; + if (rdbSaveStringObject(fp,eleobj) == -1) goto werr; } dictReleaseIterator(di); } else { @@ -1475,8 +1977,9 @@ static int saveDb(char *filename) { dictReleaseIterator(di); } /* EOF opcode */ - type = REDIS_EOF; - if (fwrite(&type,1,1,fp) == 0) goto werr; + if (rdbSaveType(fp,REDIS_EOF) == -1) goto werr; + + /* Make sure data will not remain on the OS's output buffers */ fflush(fp); fsync(fileno(fp)); fclose(fp); @@ -1501,105 +2004,226 @@ werr: return REDIS_ERR; } -static int saveDbBackground(char *filename) { +static int rdbSaveBackground(char *filename) { pid_t childpid; if (server.bgsaveinprogress) return REDIS_ERR; if ((childpid = fork()) == 0) { /* Child */ close(server.fd); - if (saveDb(filename) == REDIS_OK) { + if (rdbSave(filename) == REDIS_OK) { exit(0); } else { exit(1); } } else { /* Parent */ + if (childpid == -1) { + redisLog(REDIS_WARNING,"Can't save in background: fork: %s", + strerror(errno)); + return REDIS_ERR; + } redisLog(REDIS_NOTICE,"Background saving started by pid %d",childpid); server.bgsaveinprogress = 1; + server.bgsavechildpid = childpid; return REDIS_OK; } return REDIS_OK; /* unreached */ } -static int loadDb(char *filename) { +static void rdbRemoveTempFile(pid_t childpid) { + char tmpfile[256]; + + snprintf(tmpfile,256,"temp-%d.rdb", (int) childpid); + unlink(tmpfile); +} + +static int rdbLoadType(FILE *fp) { + unsigned char type; + if (fread(&type,1,1,fp) == 0) return -1; + return type; +} + +static time_t rdbLoadTime(FILE *fp) { + int32_t t32; + if (fread(&t32,4,1,fp) == 0) return -1; + return (time_t) t32; +} + +/* Load an encoded length from the DB, see the REDIS_RDB_* defines on the top + * of this file for a description of how this are stored on disk. + * + * isencoded is set to 1 if the readed length is not actually a length but + * an "encoding type", check the above comments for more info */ +static uint32_t rdbLoadLen(FILE *fp, int rdbver, int *isencoded) { + unsigned char buf[2]; + uint32_t len; + + if (isencoded) *isencoded = 0; + if (rdbver == 0) { + if (fread(&len,4,1,fp) == 0) return REDIS_RDB_LENERR; + return ntohl(len); + } else { + int type; + + if (fread(buf,1,1,fp) == 0) return REDIS_RDB_LENERR; + type = (buf[0]&0xC0)>>6; + if (type == REDIS_RDB_6BITLEN) { + /* Read a 6 bit len */ + return buf[0]&0x3F; + } else if (type == REDIS_RDB_ENCVAL) { + /* Read a 6 bit len encoding type */ + if (isencoded) *isencoded = 1; + return buf[0]&0x3F; + } else if (type == REDIS_RDB_14BITLEN) { + /* Read a 14 bit len */ + if (fread(buf+1,1,1,fp) == 0) return REDIS_RDB_LENERR; + return ((buf[0]&0x3F)<<8)|buf[1]; + } else { + /* Read a 32 bit len */ + if (fread(&len,4,1,fp) == 0) return REDIS_RDB_LENERR; + return ntohl(len); + } + } +} + +static robj *rdbLoadIntegerObject(FILE *fp, int enctype) { + unsigned char enc[4]; + long long val; + + if (enctype == REDIS_RDB_ENC_INT8) { + if (fread(enc,1,1,fp) == 0) return NULL; + val = (signed char)enc[0]; + } else if (enctype == REDIS_RDB_ENC_INT16) { + uint16_t v; + if (fread(enc,2,1,fp) == 0) return NULL; + v = enc[0]|(enc[1]<<8); + val = (int16_t)v; + } else if (enctype == REDIS_RDB_ENC_INT32) { + uint32_t v; + if (fread(enc,4,1,fp) == 0) return NULL; + v = enc[0]|(enc[1]<<8)|(enc[2]<<16)|(enc[3]<<24); + val = (int32_t)v; + } else { + val = 0; /* anti-warning */ + assert(0!=0); + } + return createObject(REDIS_STRING,sdscatprintf(sdsempty(),"%lld",val)); +} + +static robj *rdbLoadLzfStringObject(FILE*fp, int rdbver) { + unsigned int len, clen; + unsigned char *c = NULL; + sds val = NULL; + + if ((clen = rdbLoadLen(fp,rdbver,NULL)) == REDIS_RDB_LENERR) return NULL; + if ((len = rdbLoadLen(fp,rdbver,NULL)) == REDIS_RDB_LENERR) return NULL; + if ((c = zmalloc(clen)) == NULL) goto err; + if ((val = sdsnewlen(NULL,len)) == NULL) goto err; + if (fread(c,clen,1,fp) == 0) goto err; + if (lzf_decompress(c,clen,val,len) == 0) goto err; + zfree(c); + return createObject(REDIS_STRING,val); +err: + zfree(c); + sdsfree(val); + return NULL; +} + +static robj *rdbLoadStringObject(FILE*fp, int rdbver) { + int isencoded; + uint32_t len; + sds val; + + len = rdbLoadLen(fp,rdbver,&isencoded); + if (isencoded) { + switch(len) { + case REDIS_RDB_ENC_INT8: + case REDIS_RDB_ENC_INT16: + case REDIS_RDB_ENC_INT32: + return tryObjectSharing(rdbLoadIntegerObject(fp,len)); + case REDIS_RDB_ENC_LZF: + return tryObjectSharing(rdbLoadLzfStringObject(fp,rdbver)); + default: + assert(0!=0); + } + } + + if (len == REDIS_RDB_LENERR) return NULL; + val = sdsnewlen(NULL,len); + if (len && fread(val,len,1,fp) == 0) { + sdsfree(val); + return NULL; + } + return tryObjectSharing(createObject(REDIS_STRING,val)); +} + +static int rdbLoad(char *filename) { FILE *fp; - char buf[REDIS_LOADBUF_LEN]; /* Try to use this buffer instead of */ - char vbuf[REDIS_LOADBUF_LEN]; /* malloc() when the element is small */ - char *key = NULL, *val = NULL; - uint32_t klen,vlen,dbid; - uint8_t type; - int retval; - dict *d = server.dict[0]; + robj *keyobj = NULL; + uint32_t dbid; + int type, retval, rdbver; + dict *d = server.db[0].dict; + redisDb *db = server.db+0; + char buf[1024]; + time_t expiretime = -1, now = time(NULL); fp = fopen(filename,"r"); if (!fp) return REDIS_ERR; if (fread(buf,9,1,fp) == 0) goto eoferr; - if (memcmp(buf,"REDIS0000",9) != 0) { + buf[9] = '\0'; + if (memcmp(buf,"REDIS",5) != 0) { fclose(fp); redisLog(REDIS_WARNING,"Wrong signature trying to load DB from file"); return REDIS_ERR; } + rdbver = atoi(buf+5); + if (rdbver > 1) { + fclose(fp); + redisLog(REDIS_WARNING,"Can't handle RDB format version %d",rdbver); + return REDIS_ERR; + } while(1) { robj *o; /* Read type. */ - if (fread(&type,1,1,fp) == 0) goto eoferr; + if ((type = rdbLoadType(fp)) == -1) goto eoferr; + if (type == REDIS_EXPIRETIME) { + if ((expiretime = rdbLoadTime(fp)) == -1) goto eoferr; + /* We read the time so we need to read the object type again */ + if ((type = rdbLoadType(fp)) == -1) goto eoferr; + } if (type == REDIS_EOF) break; /* Handle SELECT DB opcode as a special case */ if (type == REDIS_SELECTDB) { - if (fread(&dbid,4,1,fp) == 0) goto eoferr; - dbid = ntohl(dbid); + if ((dbid = rdbLoadLen(fp,rdbver,NULL)) == REDIS_RDB_LENERR) + goto eoferr; if (dbid >= (unsigned)server.dbnum) { - redisLog(REDIS_WARNING,"FATAL: Data file was created with a Redis server compiled to handle more than %d databases. Exiting\n", server.dbnum); + redisLog(REDIS_WARNING,"FATAL: Data file was created with a Redis server configured to handle more than %d databases. Exiting\n", server.dbnum); exit(1); } - d = server.dict[dbid]; + db = server.db+dbid; + d = db->dict; continue; } /* Read key */ - if (fread(&klen,4,1,fp) == 0) goto eoferr; - klen = ntohl(klen); - if (klen <= REDIS_LOADBUF_LEN) { - key = buf; - } else { - key = zmalloc(klen); - if (!key) oom("Loading DB from file"); - } - if (fread(key,klen,1,fp) == 0) goto eoferr; + if ((keyobj = rdbLoadStringObject(fp,rdbver)) == NULL) goto eoferr; if (type == REDIS_STRING) { /* Read string value */ - if (fread(&vlen,4,1,fp) == 0) goto eoferr; - vlen = ntohl(vlen); - if (vlen <= REDIS_LOADBUF_LEN) { - val = vbuf; - } else { - val = zmalloc(vlen); - if (!val) oom("Loading DB from file"); - } - if (vlen && fread(val,vlen,1,fp) == 0) goto eoferr; - o = createObject(REDIS_STRING,sdsnewlen(val,vlen)); + if ((o = rdbLoadStringObject(fp,rdbver)) == NULL) goto eoferr; } else if (type == REDIS_LIST || type == REDIS_SET) { /* Read list/set value */ uint32_t listlen; - if (fread(&listlen,4,1,fp) == 0) goto eoferr; - listlen = ntohl(listlen); + + if ((listlen = rdbLoadLen(fp,rdbver,NULL)) == REDIS_RDB_LENERR) + goto eoferr; o = (type == REDIS_LIST) ? createListObject() : createSetObject(); /* Load every single element of the list/set */ while(listlen--) { robj *ele; - if (fread(&vlen,4,1,fp) == 0) goto eoferr; - vlen = ntohl(vlen); - if (vlen <= REDIS_LOADBUF_LEN) { - val = vbuf; - } else { - val = zmalloc(vlen); - if (!val) oom("Loading DB from file"); - } - if (vlen && fread(val,vlen,1,fp) == 0) goto eoferr; - ele = createObject(REDIS_STRING,sdsnewlen(val,vlen)); + if ((ele = rdbLoadStringObject(fp,rdbver)) == NULL) goto eoferr; if (type == REDIS_LIST) { if (!listAddNodeTail((list*)o->ptr,ele)) oom("listAddNodeTail"); @@ -1607,43 +2231,53 @@ static int loadDb(char *filename) { if (dictAdd((dict*)o->ptr,ele,NULL) == DICT_ERR) oom("dictAdd"); } - /* free the temp buffer if needed */ - if (val != vbuf) zfree(val); - val = NULL; } } else { assert(0 != 0); } /* Add the new object in the hash table */ - retval = dictAdd(d,createStringObject(key,klen),o); + retval = dictAdd(d,keyobj,o); if (retval == DICT_ERR) { - redisLog(REDIS_WARNING,"Loading DB, duplicated key found! Unrecoverable error, exiting now."); + redisLog(REDIS_WARNING,"Loading DB, duplicated key (%s) found! Unrecoverable error, exiting now.", keyobj->ptr); exit(1); } - /* Iteration cleanup */ - if (key != buf) zfree(key); - if (val != vbuf) zfree(val); - key = val = NULL; + /* Set the expire time if needed */ + if (expiretime != -1) { + setExpire(db,keyobj,expiretime); + /* Delete this key if already expired */ + if (expiretime < now) deleteKey(db,keyobj); + expiretime = -1; + } + keyobj = o = NULL; } fclose(fp); return REDIS_OK; eoferr: /* unexpected end of file is handled here with a fatal exit */ - if (key != buf) zfree(key); - if (val != vbuf) zfree(val); - redisLog(REDIS_WARNING,"Short read loading DB. Unrecoverable error, exiting now."); + if (keyobj) decrRefCount(keyobj); + redisLog(REDIS_WARNING,"Short read or OOM loading DB. Unrecoverable error, exiting now."); exit(1); return REDIS_ERR; /* Just to avoid warning */ } /*================================== Commands =============================== */ +static void authCommand(redisClient *c) { + if (!server.requirepass || !strcmp(c->argv[1]->ptr, server.requirepass)) { + c->authenticated = 1; + addReply(c,shared.ok); + } else { + c->authenticated = 0; + addReply(c,shared.err); + } +} + static void pingCommand(redisClient *c) { addReply(c,shared.pong); } static void echoCommand(redisClient *c) { - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n", + addReplySds(c,sdscatprintf(sdsempty(),"$%d\r\n", (int)sdslen(c->argv[1]->ptr))); addReply(c,c->argv[1]); addReply(c,shared.crlf); @@ -1654,13 +2288,13 @@ static void echoCommand(redisClient *c) { static void setGenericCommand(redisClient *c, int nx) { int retval; - retval = dictAdd(c->dict,c->argv[1],c->argv[2]); + retval = dictAdd(c->db->dict,c->argv[1],c->argv[2]); if (retval == DICT_ERR) { if (!nx) { - dictReplace(c->dict,c->argv[1],c->argv[2]); + dictReplace(c->db->dict,c->argv[1],c->argv[2]); incrRefCount(c->argv[2]); } else { - addReply(c,shared.zero); + addReply(c,shared.czero); return; } } else { @@ -1668,52 +2302,59 @@ static void setGenericCommand(redisClient *c, int nx) { incrRefCount(c->argv[2]); } server.dirty++; - addReply(c, nx ? shared.one : shared.ok); + removeExpire(c->db,c->argv[1]); + addReply(c, nx ? shared.cone : shared.ok); } static void setCommand(redisClient *c) { - return setGenericCommand(c,0); + setGenericCommand(c,0); } static void setnxCommand(redisClient *c) { - return setGenericCommand(c,1); + setGenericCommand(c,1); } static void getCommand(redisClient *c) { - dictEntry *de; - - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - addReply(c,shared.nil); + robj *o = lookupKeyRead(c->db,c->argv[1]); + + if (o == NULL) { + addReply(c,shared.nullbulk); } else { - robj *o = dictGetEntryVal(de); - if (o->type != REDIS_STRING) { - addReply(c,shared.wrongtypeerrbulk); + addReply(c,shared.wrongtypeerr); } else { - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",(int)sdslen(o->ptr))); + addReplySds(c,sdscatprintf(sdsempty(),"$%d\r\n",(int)sdslen(o->ptr))); addReply(c,o); addReply(c,shared.crlf); } } } +static void getSetCommand(redisClient *c) { + getCommand(c); + if (dictAdd(c->db->dict,c->argv[1],c->argv[2]) == DICT_ERR) { + dictReplace(c->db->dict,c->argv[1],c->argv[2]); + } else { + incrRefCount(c->argv[1]); + } + incrRefCount(c->argv[2]); + server.dirty++; + removeExpire(c->db,c->argv[1]); +} + static void mgetCommand(redisClient *c) { - dictEntry *de; int j; - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",c->argc-1)); + addReplySds(c,sdscatprintf(sdsempty(),"*%d\r\n",c->argc-1)); for (j = 1; j < c->argc; j++) { - de = dictFind(c->dict,c->argv[j]); - if (de == NULL) { - addReply(c,shared.minus1); + robj *o = lookupKeyRead(c->db,c->argv[j]); + if (o == NULL) { + addReply(c,shared.nullbulk); } else { - robj *o = dictGetEntryVal(de); - if (o->type != REDIS_STRING) { - addReply(c,shared.minus1); + addReply(c,shared.nullbulk); } else { - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",(int)sdslen(o->ptr))); + addReplySds(c,sdscatprintf(sdsempty(),"$%d\r\n",(int)sdslen(o->ptr))); addReply(c,o); addReply(c,shared.crlf); } @@ -1721,18 +2362,15 @@ static void mgetCommand(redisClient *c) { } } -static void incrDecrCommand(redisClient *c, int incr) { - dictEntry *de; +static void incrDecrCommand(redisClient *c, long long incr) { long long value; int retval; robj *o; - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { + o = lookupKeyWrite(c->db,c->argv[1]); + if (o == NULL) { value = 0; } else { - robj *o = dictGetEntryVal(de); - if (o->type != REDIS_STRING) { value = 0; } else { @@ -1744,61 +2382,70 @@ static void incrDecrCommand(redisClient *c, int incr) { value += incr; o = createObject(REDIS_STRING,sdscatprintf(sdsempty(),"%lld",value)); - retval = dictAdd(c->dict,c->argv[1],o); + retval = dictAdd(c->db->dict,c->argv[1],o); if (retval == DICT_ERR) { - dictReplace(c->dict,c->argv[1],o); + dictReplace(c->db->dict,c->argv[1],o); + removeExpire(c->db,c->argv[1]); } else { incrRefCount(c->argv[1]); } server.dirty++; + addReply(c,shared.colon); addReply(c,o); addReply(c,shared.crlf); } static void incrCommand(redisClient *c) { - return incrDecrCommand(c,1); + incrDecrCommand(c,1); } static void decrCommand(redisClient *c) { - return incrDecrCommand(c,-1); + incrDecrCommand(c,-1); } static void incrbyCommand(redisClient *c) { - int incr = atoi(c->argv[2]->ptr); - return incrDecrCommand(c,incr); + long long incr = strtoll(c->argv[2]->ptr, NULL, 10); + incrDecrCommand(c,incr); } static void decrbyCommand(redisClient *c) { - int incr = atoi(c->argv[2]->ptr); - return incrDecrCommand(c,-incr); + long long incr = strtoll(c->argv[2]->ptr, NULL, 10); + incrDecrCommand(c,-incr); } /* ========================= Type agnostic commands ========================= */ static void delCommand(redisClient *c) { - if (dictDelete(c->dict,c->argv[1]) == DICT_OK) { - server.dirty++; - addReply(c,shared.one); - } else { - addReply(c,shared.zero); + int deleted = 0, j; + + for (j = 1; j < c->argc; j++) { + if (deleteKey(c->db,c->argv[j])) { + server.dirty++; + deleted++; + } + } + switch(deleted) { + case 0: + addReply(c,shared.czero); + break; + case 1: + addReply(c,shared.cone); + break; + default: + addReplySds(c,sdscatprintf(sdsempty(),":%d\r\n",deleted)); + break; } } static void existsCommand(redisClient *c) { - dictEntry *de; - - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) - addReply(c,shared.zero); - else - addReply(c,shared.one); + addReply(c,lookupKeyRead(c->db,c->argv[1]) ? shared.cone : shared.czero); } static void selectCommand(redisClient *c) { int id = atoi(c->argv[1]->ptr); if (selectDb(c,id) == REDIS_ERR) { - addReplySds(c,"-ERR invalid DB index\r\n"); + addReplySds(c,sdsnew("-ERR invalid DB index\r\n")); } else { addReply(c,shared.ok); } @@ -1806,11 +2453,16 @@ static void selectCommand(redisClient *c) { static void randomkeyCommand(redisClient *c) { dictEntry *de; - - de = dictGetRandomKey(c->dict); + + while(1) { + de = dictGetRandomKey(c->db->dict); + if (!de || expireIfNeeded(c->db,dictGetEntryKey(de)) == 0) break; + } if (de == NULL) { + addReply(c,shared.plus); addReply(c,shared.crlf); } else { + addReply(c,shared.plus); addReply(c,dictGetEntryKey(de)); addReply(c,shared.crlf); } @@ -1824,51 +2476,52 @@ static void keysCommand(redisClient *c) { int numkeys = 0, keyslen = 0; robj *lenobj = createObject(REDIS_STRING,NULL); - di = dictGetIterator(c->dict); + di = dictGetIterator(c->db->dict); if (!di) oom("dictGetIterator"); addReply(c,lenobj); decrRefCount(lenobj); while((de = dictNext(di)) != NULL) { robj *keyobj = dictGetEntryKey(de); + sds key = keyobj->ptr; if ((pattern[0] == '*' && pattern[1] == '\0') || stringmatchlen(pattern,plen,key,sdslen(key),0)) { - if (numkeys != 0) - addReply(c,shared.space); - addReply(c,keyobj); - numkeys++; - keyslen += sdslen(key); + if (expireIfNeeded(c->db,keyobj) == 0) { + if (numkeys != 0) + addReply(c,shared.space); + addReply(c,keyobj); + numkeys++; + keyslen += sdslen(key); + } } } dictReleaseIterator(di); - lenobj->ptr = sdscatprintf(sdsempty(),"%lu\r\n",keyslen+(numkeys ? (numkeys-1) : 0)); + lenobj->ptr = sdscatprintf(sdsempty(),"$%lu\r\n",keyslen+(numkeys ? (numkeys-1) : 0)); addReply(c,shared.crlf); } static void dbsizeCommand(redisClient *c) { addReplySds(c, - sdscatprintf(sdsempty(),"%lu\r\n",dictGetHashTableUsed(c->dict))); + sdscatprintf(sdsempty(),":%lu\r\n",dictSize(c->db->dict))); } static void lastsaveCommand(redisClient *c) { addReplySds(c, - sdscatprintf(sdsempty(),"%lu\r\n",server.lastsave)); + sdscatprintf(sdsempty(),":%lu\r\n",server.lastsave)); } static void typeCommand(redisClient *c) { - dictEntry *de; + robj *o; char *type; - - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - type = "none"; + + o = lookupKeyRead(c->db,c->argv[1]); + if (o == NULL) { + type = "+none"; } else { - robj *o = dictGetEntryVal(de); - switch(o->type) { - case REDIS_STRING: type = "string"; break; - case REDIS_LIST: type = "list"; break; - case REDIS_SET: type = "set"; break; + case REDIS_STRING: type = "+string"; break; + case REDIS_LIST: type = "+list"; break; + case REDIS_SET: type = "+set"; break; default: type = "unknown"; break; } } @@ -1877,7 +2530,11 @@ static void typeCommand(redisClient *c) { } static void saveCommand(redisClient *c) { - if (saveDb(server.dbfilename) == REDIS_OK) { + if (server.bgsaveinprogress) { + addReplySds(c,sdsnew("-ERR background save in progress\r\n")); + return; + } + if (rdbSave(server.dbfilename) == REDIS_OK) { addReply(c,shared.ok); } else { addReply(c,shared.err); @@ -1889,7 +2546,7 @@ static void bgsaveCommand(redisClient *c) { addReplySds(c,sdsnew("-ERR background save already in progress\r\n")); return; } - if (saveDbBackground(server.dbfilename) == REDIS_OK) { + if (rdbSaveBackground(server.dbfilename) == REDIS_OK) { addReply(c,shared.ok); } else { addReply(c,shared.err); @@ -1898,51 +2555,60 @@ static void bgsaveCommand(redisClient *c) { static void shutdownCommand(redisClient *c) { redisLog(REDIS_WARNING,"User requested shutdown, saving DB..."); - if (saveDb(server.dbfilename) == REDIS_OK) { - redisLog(REDIS_WARNING,"Server exit now, bye bye..."); + /* Kill the saving child if there is a background saving in progress. + We want to avoid race conditions, for instance our saving child may + overwrite the synchronous saving did by SHUTDOWN. */ + if (server.bgsaveinprogress) { + redisLog(REDIS_WARNING,"There is a live saving child. Killing it!"); + kill(server.bgsavechildpid,SIGKILL); + rdbRemoveTempFile(server.bgsavechildpid); + } + /* SYNC SAVE */ + if (rdbSave(server.dbfilename) == REDIS_OK) { + if (server.daemonize) + unlink(server.pidfile); + redisLog(REDIS_WARNING,"%zu bytes used at exit",zmalloc_used_memory()); + redisLog(REDIS_WARNING,"Server exit now, bye bye..."); exit(1); } else { + /* Ooops.. error saving! The best we can do is to continue operating. + * Note that if there was a background saving process, in the next + * cron() Redis will be notified that the background saving aborted, + * handling special stuff like slaves pending for synchronization... */ redisLog(REDIS_WARNING,"Error trying to save the DB, can't exit"); addReplySds(c,sdsnew("-ERR can't quit, problems saving the DB\r\n")); } } static void renameGenericCommand(redisClient *c, int nx) { - dictEntry *de; robj *o; /* To use the same key as src and dst is probably an error */ if (sdscmp(c->argv[1]->ptr,c->argv[2]->ptr) == 0) { - if (nx) - addReply(c,shared.minus3); - else - addReplySds(c,sdsnew("-ERR src and dest key are the same\r\n")); + addReply(c,shared.sameobjecterr); return; } - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - if (nx) - addReply(c,shared.minus1); - else - addReply(c,shared.nokeyerr); + o = lookupKeyWrite(c->db,c->argv[1]); + if (o == NULL) { + addReply(c,shared.nokeyerr); return; } - o = dictGetEntryVal(de); incrRefCount(o); - if (dictAdd(c->dict,c->argv[2],o) == DICT_ERR) { + deleteIfVolatile(c->db,c->argv[2]); + if (dictAdd(c->db->dict,c->argv[2],o) == DICT_ERR) { if (nx) { decrRefCount(o); - addReply(c,shared.zero); + addReply(c,shared.czero); return; } - dictReplace(c->dict,c->argv[2],o); + dictReplace(c->db->dict,c->argv[2],o); } else { incrRefCount(c->argv[2]); } - dictDelete(c->dict,c->argv[1]); + deleteKey(c->db,c->argv[1]); server.dirty++; - addReply(c,nx ? shared.one : shared.ok); + addReply(c,nx ? shared.cone : shared.ok); } static void renameCommand(redisClient *c) { @@ -1954,60 +2620,56 @@ static void renamenxCommand(redisClient *c) { } static void moveCommand(redisClient *c) { - dictEntry *de; - robj *o, *key; - dict *src, *dst; + robj *o; + redisDb *src, *dst; int srcid; /* Obtain source and target DB pointers */ - src = c->dict; - srcid = c->dictid; + src = c->db; + srcid = c->db->id; if (selectDb(c,atoi(c->argv[2]->ptr)) == REDIS_ERR) { - addReply(c,shared.minus4); + addReply(c,shared.outofrangeerr); return; } - dst = c->dict; - c->dict = src; - c->dictid = srcid; + dst = c->db; + selectDb(c,srcid); /* Back to the source DB */ /* If the user is moving using as target the same * DB as the source DB it is probably an error. */ if (src == dst) { - addReply(c,shared.minus3); + addReply(c,shared.sameobjecterr); return; } /* Check if the element exists and get a reference */ - de = dictFind(c->dict,c->argv[1]); - if (!de) { - addReply(c,shared.zero); + o = lookupKeyWrite(c->db,c->argv[1]); + if (!o) { + addReply(c,shared.czero); return; } /* Try to add the element to the target DB */ - key = dictGetEntryKey(de); - o = dictGetEntryVal(de); - if (dictAdd(dst,key,o) == DICT_ERR) { - addReply(c,shared.zero); + deleteIfVolatile(dst,c->argv[1]); + if (dictAdd(dst->dict,c->argv[1],o) == DICT_ERR) { + addReply(c,shared.czero); return; } - incrRefCount(key); + incrRefCount(c->argv[1]); incrRefCount(o); /* OK! key moved, free the entry in the source DB */ - dictDelete(src,c->argv[1]); + deleteKey(src,c->argv[1]); server.dirty++; - addReply(c,shared.one); + addReply(c,shared.cone); } /* =================================== Lists ================================ */ static void pushGenericCommand(redisClient *c, int where) { robj *lobj; - dictEntry *de; list *list; - - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { + + lobj = lookupKeyWrite(c->db,c->argv[1]); + if (lobj == NULL) { lobj = createListObject(); list = lobj->ptr; if (where == REDIS_HEAD) { @@ -2015,11 +2677,10 @@ static void pushGenericCommand(redisClient *c, int where) { } else { if (!listAddNodeTail(list,c->argv[2])) oom("listAddNodeTail"); } - dictAdd(c->dict,c->argv[1],lobj); + dictAdd(c->db->dict,c->argv[1],lobj); incrRefCount(c->argv[1]); incrRefCount(c->argv[2]); } else { - lobj = dictGetEntryVal(de); if (lobj->type != REDIS_LIST) { addReply(c,shared.wrongtypeerr); return; @@ -2045,46 +2706,43 @@ static void rpushCommand(redisClient *c) { } static void llenCommand(redisClient *c) { - dictEntry *de; + robj *o; list *l; - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - addReply(c,shared.zero); + o = lookupKeyRead(c->db,c->argv[1]); + if (o == NULL) { + addReply(c,shared.czero); return; } else { - robj *o = dictGetEntryVal(de); if (o->type != REDIS_LIST) { - addReply(c,shared.minus2); + addReply(c,shared.wrongtypeerr); } else { l = o->ptr; - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",listLength(l))); + addReplySds(c,sdscatprintf(sdsempty(),":%d\r\n",listLength(l))); } } } static void lindexCommand(redisClient *c) { - dictEntry *de; + robj *o; int index = atoi(c->argv[2]->ptr); - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - addReply(c,shared.nil); + o = lookupKeyRead(c->db,c->argv[1]); + if (o == NULL) { + addReply(c,shared.nullbulk); } else { - robj *o = dictGetEntryVal(de); - if (o->type != REDIS_LIST) { - addReply(c,shared.wrongtypeerrbulk); + addReply(c,shared.wrongtypeerr); } else { list *list = o->ptr; listNode *ln; ln = listIndex(list, index); if (ln == NULL) { - addReply(c,shared.nil); + addReply(c,shared.nullbulk); } else { robj *ele = listNodeValue(ln); - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",(int)sdslen(ele->ptr))); + addReplySds(c,sdscatprintf(sdsempty(),"$%d\r\n",(int)sdslen(ele->ptr))); addReply(c,ele); addReply(c,shared.crlf); } @@ -2093,15 +2751,13 @@ static void lindexCommand(redisClient *c) { } static void lsetCommand(redisClient *c) { - dictEntry *de; + robj *o; int index = atoi(c->argv[2]->ptr); - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { + o = lookupKeyWrite(c->db,c->argv[1]); + if (o == NULL) { addReply(c,shared.nokeyerr); } else { - robj *o = dictGetEntryVal(de); - if (o->type != REDIS_LIST) { addReply(c,shared.wrongtypeerr); } else { @@ -2110,7 +2766,7 @@ static void lsetCommand(redisClient *c) { ln = listIndex(list, index); if (ln == NULL) { - addReplySds(c,sdsnew("-ERR index out of range\r\n")); + addReply(c,shared.outofrangeerr); } else { robj *ele = listNodeValue(ln); @@ -2125,16 +2781,14 @@ static void lsetCommand(redisClient *c) { } static void popGenericCommand(redisClient *c, int where) { - dictEntry *de; - - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - addReply(c,shared.nil); + robj *o; + + o = lookupKeyWrite(c->db,c->argv[1]); + if (o == NULL) { + addReply(c,shared.nullbulk); } else { - robj *o = dictGetEntryVal(de); - if (o->type != REDIS_LIST) { - addReply(c,shared.wrongtypeerrbulk); + addReply(c,shared.wrongtypeerr); } else { list *list = o->ptr; listNode *ln; @@ -2145,10 +2799,10 @@ static void popGenericCommand(redisClient *c, int where) { ln = listLast(list); if (ln == NULL) { - addReply(c,shared.nil); + addReply(c,shared.nullbulk); } else { robj *ele = listNodeValue(ln); - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",(int)sdslen(ele->ptr))); + addReplySds(c,sdscatprintf(sdsempty(),"$%d\r\n",(int)sdslen(ele->ptr))); addReply(c,ele); addReply(c,shared.crlf); listDelNode(list,ln); @@ -2167,18 +2821,16 @@ static void rpopCommand(redisClient *c) { } static void lrangeCommand(redisClient *c) { - dictEntry *de; + robj *o; int start = atoi(c->argv[2]->ptr); int end = atoi(c->argv[3]->ptr); - - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - addReply(c,shared.nil); + + o = lookupKeyRead(c->db,c->argv[1]); + if (o == NULL) { + addReply(c,shared.nullmultibulk); } else { - robj *o = dictGetEntryVal(de); - if (o->type != REDIS_LIST) { - addReply(c,shared.wrongtypeerrbulk); + addReply(c,shared.wrongtypeerr); } else { list *list = o->ptr; listNode *ln; @@ -2195,7 +2847,7 @@ static void lrangeCommand(redisClient *c) { /* indexes sanity checks */ if (start > end || start >= llen) { /* Out of range start or start > end result in empty list */ - addReply(c,shared.zero); + addReply(c,shared.emptymultibulk); return; } if (end >= llen) end = llen-1; @@ -2203,10 +2855,10 @@ static void lrangeCommand(redisClient *c) { /* Return the result in form of a multi-bulk reply */ ln = listIndex(list, start); - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",rangelen)); + addReplySds(c,sdscatprintf(sdsempty(),"*%d\r\n",rangelen)); for (j = 0; j < rangelen; j++) { ele = listNodeValue(ln); - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",(int)sdslen(ele->ptr))); + addReplySds(c,sdscatprintf(sdsempty(),"$%d\r\n",(int)sdslen(ele->ptr))); addReply(c,ele); addReply(c,shared.crlf); ln = ln->next; @@ -2216,16 +2868,14 @@ static void lrangeCommand(redisClient *c) { } static void ltrimCommand(redisClient *c) { - dictEntry *de; + robj *o; int start = atoi(c->argv[2]->ptr); int end = atoi(c->argv[3]->ptr); - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { + o = lookupKeyWrite(c->db,c->argv[1]); + if (o == NULL) { addReply(c,shared.nokeyerr); } else { - robj *o = dictGetEntryVal(de); - if (o->type != REDIS_LIST) { addReply(c,shared.wrongtypeerr); } else { @@ -2260,23 +2910,21 @@ static void ltrimCommand(redisClient *c) { ln = listLast(list); listDelNode(list,ln); } - addReply(c,shared.ok); server.dirty++; + addReply(c,shared.ok); } } } static void lremCommand(redisClient *c) { - dictEntry *de; + robj *o; - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - addReply(c,shared.minus1); + o = lookupKeyWrite(c->db,c->argv[1]); + if (o == NULL) { + addReply(c,shared.czero); } else { - robj *o = dictGetEntryVal(de); - if (o->type != REDIS_LIST) { - addReply(c,shared.minus2); + addReply(c,shared.wrongtypeerr); } else { list *list = o->ptr; listNode *ln, *next; @@ -2290,8 +2938,9 @@ static void lremCommand(redisClient *c) { } ln = fromtail ? list->tail : list->head; while (ln) { - next = fromtail ? ln->prev : ln->next; robj *ele = listNodeValue(ln); + + next = fromtail ? ln->prev : ln->next; if (sdscmp(ele->ptr,c->argv[3]->ptr) == 0) { listDelNode(list,ln); server.dirty++; @@ -2300,7 +2949,7 @@ static void lremCommand(redisClient *c) { } ln = next; } - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",removed)); + addReplySds(c,sdscatprintf(sdsempty(),":%d\r\n",removed)); } } } @@ -2308,90 +2957,145 @@ static void lremCommand(redisClient *c) { /* ==================================== Sets ================================ */ static void saddCommand(redisClient *c) { - dictEntry *de; robj *set; - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { + set = lookupKeyWrite(c->db,c->argv[1]); + if (set == NULL) { set = createSetObject(); - dictAdd(c->dict,c->argv[1],set); + dictAdd(c->db->dict,c->argv[1],set); incrRefCount(c->argv[1]); } else { - set = dictGetEntryVal(de); if (set->type != REDIS_SET) { - addReply(c,shared.minus2); + addReply(c,shared.wrongtypeerr); return; } } if (dictAdd(set->ptr,c->argv[2],NULL) == DICT_OK) { incrRefCount(c->argv[2]); server.dirty++; - addReply(c,shared.one); + addReply(c,shared.cone); } else { - addReply(c,shared.zero); + addReply(c,shared.czero); } } static void sremCommand(redisClient *c) { - dictEntry *de; + robj *set; - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - addReply(c,shared.zero); + set = lookupKeyWrite(c->db,c->argv[1]); + if (set == NULL) { + addReply(c,shared.czero); } else { - robj *set; - - set = dictGetEntryVal(de); if (set->type != REDIS_SET) { - addReply(c,shared.minus2); + addReply(c,shared.wrongtypeerr); return; } if (dictDelete(set->ptr,c->argv[2]) == DICT_OK) { server.dirty++; - addReply(c,shared.one); + if (htNeedsResize(set->ptr)) dictResize(set->ptr); + addReply(c,shared.cone); } else { - addReply(c,shared.zero); + addReply(c,shared.czero); } } } +static void smoveCommand(redisClient *c) { + robj *srcset, *dstset; + + srcset = lookupKeyWrite(c->db,c->argv[1]); + dstset = lookupKeyWrite(c->db,c->argv[2]); + + /* If the source key does not exist return 0, if it's of the wrong type + * raise an error */ + if (srcset == NULL || srcset->type != REDIS_SET) { + addReply(c, srcset ? shared.wrongtypeerr : shared.czero); + return; + } + /* Error if the destination key is not a set as well */ + if (dstset && dstset->type != REDIS_SET) { + addReply(c,shared.wrongtypeerr); + return; + } + /* Remove the element from the source set */ + if (dictDelete(srcset->ptr,c->argv[3]) == DICT_ERR) { + /* Key not found in the src set! return zero */ + addReply(c,shared.czero); + return; + } + server.dirty++; + /* Add the element to the destination set */ + if (!dstset) { + dstset = createSetObject(); + dictAdd(c->db->dict,c->argv[2],dstset); + incrRefCount(c->argv[2]); + } + if (dictAdd(dstset->ptr,c->argv[3],NULL) == DICT_OK) + incrRefCount(c->argv[3]); + addReply(c,shared.cone); +} + static void sismemberCommand(redisClient *c) { - dictEntry *de; + robj *set; - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - addReply(c,shared.zero); + set = lookupKeyRead(c->db,c->argv[1]); + if (set == NULL) { + addReply(c,shared.czero); } else { - robj *set; - - set = dictGetEntryVal(de); if (set->type != REDIS_SET) { - addReply(c,shared.minus2); + addReply(c,shared.wrongtypeerr); return; } if (dictFind(set->ptr,c->argv[2])) - addReply(c,shared.one); + addReply(c,shared.cone); else - addReply(c,shared.zero); + addReply(c,shared.czero); } } static void scardCommand(redisClient *c) { - dictEntry *de; + robj *o; dict *s; - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - addReply(c,shared.zero); + o = lookupKeyRead(c->db,c->argv[1]); + if (o == NULL) { + addReply(c,shared.czero); return; } else { - robj *o = dictGetEntryVal(de); if (o->type != REDIS_SET) { - addReply(c,shared.minus2); + addReply(c,shared.wrongtypeerr); } else { s = o->ptr; - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n", - dictGetHashTableUsed(s))); + addReplySds(c,sdscatprintf(sdsempty(),":%d\r\n", + dictSize(s))); + } + } +} + +static void spopCommand(redisClient *c) { + robj *set; + dictEntry *de; + + set = lookupKeyWrite(c->db,c->argv[1]); + if (set == NULL) { + addReply(c,shared.nullbulk); + } else { + if (set->type != REDIS_SET) { + addReply(c,shared.wrongtypeerr); + return; + } + de = dictGetRandomKey(set->ptr); + if (de == NULL) { + addReply(c,shared.nullbulk); + } else { + robj *ele = dictGetEntryKey(de); + + addReplySds(c,sdscatprintf(sdsempty(),"$%d\r\n",sdslen(ele->ptr))); + addReply(c,ele); + addReply(c,shared.crlf); + dictDelete(set->ptr,ele); + if (htNeedsResize(set->ptr)) dictResize(set->ptr); + server.dirty++; } } } @@ -2399,7 +3103,7 @@ static void scardCommand(redisClient *c) { static int qsortCompareSetsByCardinality(const void *s1, const void *s2) { dict **d1 = (void*) s1, **d2 = (void*) s2; - return dictGetHashTableUsed(*d1)-dictGetHashTableUsed(*d2); + return dictSize(*d1)-dictSize(*d2); } static void sinterGenericCommand(redisClient *c, robj **setskeys, int setsnum, robj *dstkey) { @@ -2409,21 +3113,26 @@ static void sinterGenericCommand(redisClient *c, robj **setskeys, int setsnum, r robj *lenobj = NULL, *dstset = NULL; int j, cardinality = 0; - if (!dv) oom("sinterCommand"); + if (!dv) oom("sinterGenericCommand"); for (j = 0; j < setsnum; j++) { robj *setobj; - dictEntry *de; - - de = dictFind(c->dict,setskeys[j]); - if (!de) { + + setobj = dstkey ? + lookupKeyWrite(c->db,setskeys[j]) : + lookupKeyRead(c->db,setskeys[j]); + if (!setobj) { zfree(dv); - addReply(c,dstkey ? shared.nokeyerr : shared.nil); + if (dstkey) { + deleteKey(c->db,dstkey); + addReply(c,shared.ok); + } else { + addReply(c,shared.nullmultibulk); + } return; } - setobj = dictGetEntryVal(de); if (setobj->type != REDIS_SET) { zfree(dv); - addReply(c,dstkey ? shared.wrongtypeerr : shared.wrongtypeerrbulk); + addReply(c,shared.wrongtypeerr); return; } dv[j] = setobj->ptr; @@ -2445,9 +3154,6 @@ static void sinterGenericCommand(redisClient *c, robj **setskeys, int setsnum, r /* If we have a target key where to store the resulting set * create this key with an empty set inside */ dstset = createSetObject(); - dictDelete(c->dict,dstkey); - dictAdd(c->dict,dstkey,dstset); - incrRefCount(dstkey); } /* Iterate all the elements of the first (smallest) set, and test @@ -2465,7 +3171,7 @@ static void sinterGenericCommand(redisClient *c, robj **setskeys, int setsnum, r continue; /* at least one set does not contain the member */ ele = dictGetEntryKey(de); if (!dstkey) { - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",sdslen(ele->ptr))); + addReplySds(c,sdscatprintf(sdsempty(),"$%d\r\n",sdslen(ele->ptr))); addReply(c,ele); addReply(c,shared.crlf); cardinality++; @@ -2476,10 +3182,20 @@ static void sinterGenericCommand(redisClient *c, robj **setskeys, int setsnum, r } dictReleaseIterator(di); - if (!dstkey) - lenobj->ptr = sdscatprintf(sdsempty(),"%d\r\n",cardinality); - else - addReply(c,shared.ok); + if (dstkey) { + /* Store the resulting set into the target */ + deleteKey(c->db,dstkey); + dictAdd(c->db->dict,dstkey,dstset); + incrRefCount(dstkey); + } + + if (!dstkey) { + lenobj->ptr = sdscatprintf(sdsempty(),"*%d\r\n",cardinality); + } else { + addReplySds(c,sdscatprintf(sdsempty(),":%d\r\n", + dictSize((dict*)dstset->ptr))); + server.dirty++; + } zfree(dv); } @@ -2491,19 +3207,135 @@ static void sinterstoreCommand(redisClient *c) { sinterGenericCommand(c,c->argv+2,c->argc-2,c->argv[1]); } +#define REDIS_OP_UNION 0 +#define REDIS_OP_DIFF 1 + +static void sunionDiffGenericCommand(redisClient *c, robj **setskeys, int setsnum, robj *dstkey, int op) { + dict **dv = zmalloc(sizeof(dict*)*setsnum); + dictIterator *di; + dictEntry *de; + robj *dstset = NULL; + int j, cardinality = 0; + + if (!dv) oom("sunionDiffGenericCommand"); + for (j = 0; j < setsnum; j++) { + robj *setobj; + + setobj = dstkey ? + lookupKeyWrite(c->db,setskeys[j]) : + lookupKeyRead(c->db,setskeys[j]); + if (!setobj) { + dv[j] = NULL; + continue; + } + if (setobj->type != REDIS_SET) { + zfree(dv); + addReply(c,shared.wrongtypeerr); + return; + } + dv[j] = setobj->ptr; + } + + /* We need a temp set object to store our union. If the dstkey + * is not NULL (that is, we are inside an SUNIONSTORE operation) then + * this set object will be the resulting object to set into the target key*/ + dstset = createSetObject(); + + /* Iterate all the elements of all the sets, add every element a single + * time to the result set */ + for (j = 0; j < setsnum; j++) { + if (op == REDIS_OP_DIFF && j == 0 && !dv[j]) break; /* result set is empty */ + if (!dv[j]) continue; /* non existing keys are like empty sets */ + + di = dictGetIterator(dv[j]); + if (!di) oom("dictGetIterator"); + + while((de = dictNext(di)) != NULL) { + robj *ele; + + /* dictAdd will not add the same element multiple times */ + ele = dictGetEntryKey(de); + if (op == REDIS_OP_UNION || j == 0) { + if (dictAdd(dstset->ptr,ele,NULL) == DICT_OK) { + incrRefCount(ele); + cardinality++; + } + } else if (op == REDIS_OP_DIFF) { + if (dictDelete(dstset->ptr,ele) == DICT_OK) { + cardinality--; + } + } + } + dictReleaseIterator(di); + + if (op == REDIS_OP_DIFF && cardinality == 0) break; /* result set is empty */ + } + + /* Output the content of the resulting set, if not in STORE mode */ + if (!dstkey) { + addReplySds(c,sdscatprintf(sdsempty(),"*%d\r\n",cardinality)); + di = dictGetIterator(dstset->ptr); + if (!di) oom("dictGetIterator"); + while((de = dictNext(di)) != NULL) { + robj *ele; + + ele = dictGetEntryKey(de); + addReplySds(c,sdscatprintf(sdsempty(), + "$%d\r\n",sdslen(ele->ptr))); + addReply(c,ele); + addReply(c,shared.crlf); + } + dictReleaseIterator(di); + } else { + /* If we have a target key where to store the resulting set + * create this key with the result set inside */ + deleteKey(c->db,dstkey); + dictAdd(c->db->dict,dstkey,dstset); + incrRefCount(dstkey); + } + + /* Cleanup */ + if (!dstkey) { + decrRefCount(dstset); + } else { + addReplySds(c,sdscatprintf(sdsempty(),":%d\r\n", + dictSize((dict*)dstset->ptr))); + server.dirty++; + } + zfree(dv); +} + +static void sunionCommand(redisClient *c) { + sunionDiffGenericCommand(c,c->argv+1,c->argc-1,NULL,REDIS_OP_UNION); +} + +static void sunionstoreCommand(redisClient *c) { + sunionDiffGenericCommand(c,c->argv+2,c->argc-2,c->argv[1],REDIS_OP_UNION); +} + +static void sdiffCommand(redisClient *c) { + sunionDiffGenericCommand(c,c->argv+1,c->argc-1,NULL,REDIS_OP_DIFF); +} + +static void sdiffstoreCommand(redisClient *c) { + sunionDiffGenericCommand(c,c->argv+2,c->argc-2,c->argv[1],REDIS_OP_DIFF); +} + static void flushdbCommand(redisClient *c) { - dictEmpty(c->dict); + server.dirty += dictSize(c->db->dict); + dictEmpty(c->db->dict); + dictEmpty(c->db->expires); addReply(c,shared.ok); - saveDb(server.dbfilename); } static void flushallCommand(redisClient *c) { - emptyDb(); + server.dirty += emptyDb(); addReply(c,shared.ok); - saveDb(server.dbfilename); + rdbSave(server.dbfilename); + server.dirty++; } -redisSortOperation *createSortOperation(int type, robj *pattern) { +static redisSortOperation *createSortOperation(int type, robj *pattern) { redisSortOperation *so = zmalloc(sizeof(*so)); if (!so) oom("createSortOperation"); so->type = type; @@ -2513,12 +3345,11 @@ redisSortOperation *createSortOperation(int type, robj *pattern) { /* Return the value associated to the key with a name obtained * substituting the first occurence of '*' in 'pattern' with 'subst' */ -robj *lookupKeyByPattern(dict *dict, robj *pattern, robj *subst) { +static robj *lookupKeyByPattern(redisDb *db, robj *pattern, robj *subst) { char *p; sds spat, ssub; robj keyobj; int prefixlen, sublen, postfixlen; - dictEntry *de; /* Expoit the internal sds representation to create a sds string allocated on the stack in order to make this function faster */ struct { long len; @@ -2526,7 +3357,6 @@ robj *lookupKeyByPattern(dict *dict, robj *pattern, robj *subst) { char buf[REDIS_SORTKEY_MAX+1]; } keyname; - spat = pattern->ptr; ssub = subst->ptr; if (sdslen(spat)+sdslen(ssub)-1 > REDIS_SORTKEY_MAX) return NULL; @@ -2546,10 +3376,8 @@ robj *lookupKeyByPattern(dict *dict, robj *pattern, robj *subst) { keyobj.type = REDIS_STRING; keyobj.ptr = ((char*)&keyname)+(sizeof(long)*2); - de = dictFind(dict,&keyobj); - // printf("lookup '%s' => %p\n", keyname.buf,de); - if (!de) return NULL; - return dictGetEntryVal(de); + /* printf("lookup '%s' => %p\n", keyname.buf,de); */ + return lookupKeyRead(db,&keyobj); } /* sortCompare() is used by qsort in sortCommand(). Given that qsort_r with @@ -2594,7 +3422,6 @@ static int sortCompare(const void *s1, const void *s2) { /* The SORT command is the most complex command in Redis. Warning: this code * is optimized for speed and a bit less for readability */ static void sortCommand(redisClient *c) { - dictEntry *de; list *operations; int outputlen = 0; int desc = 0, alpha = 0; @@ -2605,14 +3432,13 @@ static void sortCommand(redisClient *c) { redisSortObject *vector; /* Resulting vector to sort */ /* Lookup the key to sort. It must be of the right types */ - de = dictFind(c->dict,c->argv[1]); - if (de == NULL) { - addReply(c,shared.nokeyerrbulk); + sortval = lookupKeyRead(c->db,c->argv[1]); + if (sortval == NULL) { + addReply(c,shared.nokeyerr); return; } - sortval = dictGetEntryVal(de); if (sortval->type != REDIS_SET && sortval->type != REDIS_LIST) { - addReply(c,shared.wrongtypeerrbulk); + addReply(c,shared.wrongtypeerr); return; } @@ -2666,7 +3492,7 @@ static void sortCommand(redisClient *c) { } else { decrRefCount(sortval); listRelease(operations); - addReply(c,shared.syntaxerrbulk); + addReply(c,shared.syntaxerr); return; } j++; @@ -2675,19 +3501,20 @@ static void sortCommand(redisClient *c) { /* Load the sorting vector with all the objects to sort */ vectorlen = (sortval->type == REDIS_LIST) ? listLength((list*)sortval->ptr) : - dictGetHashTableUsed((dict*)sortval->ptr); + dictSize((dict*)sortval->ptr); vector = zmalloc(sizeof(redisSortObject)*vectorlen); if (!vector) oom("allocating objects vector for SORT"); j = 0; if (sortval->type == REDIS_LIST) { list *list = sortval->ptr; - listNode *ln = list->head; - while(ln) { + listNode *ln; + + listRewind(list); + while((ln = listYield(list))) { robj *ele = ln->value; vector[j].obj = ele; vector[j].u.score = 0; vector[j].u.cmpobj = NULL; - ln = ln->next; j++; } } else { @@ -2713,7 +3540,7 @@ static void sortCommand(redisClient *c) { if (sortby) { robj *byval; - byval = lookupKeyByPattern(c->dict,sortby,vector[j].obj); + byval = lookupKeyByPattern(c->db,sortby,vector[j].obj); if (!byval || byval->type != REDIS_STRING) continue; if (alpha) { vector[j].u.cmpobj = byval; @@ -2741,31 +3568,35 @@ static void sortCommand(redisClient *c) { server.sort_desc = desc; server.sort_alpha = alpha; server.sort_bypattern = sortby ? 1 : 0; - qsort(vector,vectorlen,sizeof(redisSortObject),sortCompare); + if (sortby && (start != 0 || end != vectorlen-1)) + pqsort(vector,vectorlen,sizeof(redisSortObject),sortCompare, start,end); + else + qsort(vector,vectorlen,sizeof(redisSortObject),sortCompare); } /* Send command output to the output buffer, performing the specified * GET/DEL/INCR/DECR operations if any. */ outputlen = getop ? getop*(end-start+1) : end-start+1; - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",outputlen)); + addReplySds(c,sdscatprintf(sdsempty(),"*%d\r\n",outputlen)); for (j = start; j <= end; j++) { - listNode *ln = operations->head; + listNode *ln; if (!getop) { - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n", + addReplySds(c,sdscatprintf(sdsempty(),"$%d\r\n", sdslen(vector[j].obj->ptr))); addReply(c,vector[j].obj); addReply(c,shared.crlf); } - while(ln) { + listRewind(operations); + while((ln = listYield(operations))) { redisSortOperation *sop = ln->value; - robj *val = lookupKeyByPattern(c->dict,sop->pattern, + robj *val = lookupKeyByPattern(c->db,sop->pattern, vector[j].obj); if (sop->type == REDIS_SORT_GET) { if (!val || val->type != REDIS_STRING) { - addReply(c,shared.minus1); + addReply(c,shared.nullbulk); } else { - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n", + addReplySds(c,sdscatprintf(sdsempty(),"$%d\r\n", sdslen(val->ptr))); addReply(c,val); addReply(c,shared.crlf); @@ -2773,7 +3604,6 @@ static void sortCommand(redisClient *c) { } else if (sop->type == REDIS_SORT_DEL) { /* TODO */ } - ln = ln->next; } } @@ -2790,54 +3620,171 @@ static void sortCommand(redisClient *c) { static void infoCommand(redisClient *c) { sds info; time_t uptime = time(NULL)-server.stat_starttime; + int j; info = sdscatprintf(sdsempty(), "redis_version:%s\r\n" + "uptime_in_seconds:%d\r\n" + "uptime_in_days:%d\r\n" "connected_clients:%d\r\n" "connected_slaves:%d\r\n" - "used_memory:%d\r\n" + "used_memory:%zu\r\n" "changes_since_last_save:%lld\r\n" + "bgsave_in_progress:%d\r\n" "last_save_time:%d\r\n" "total_connections_received:%lld\r\n" "total_commands_processed:%lld\r\n" - "uptime_in_seconds:%d\r\n" - "uptime_in_days:%d\r\n" + "role:%s\r\n" ,REDIS_VERSION, + uptime, + uptime/(3600*24), listLength(server.clients)-listLength(server.slaves), listLength(server.slaves), server.usedmemory, server.dirty, + server.bgsaveinprogress, server.lastsave, server.stat_numconnections, server.stat_numcommands, - uptime, - uptime/(3600*24) + server.masterhost == NULL ? "master" : "slave" ); - addReplySds(c,sdscatprintf(sdsempty(),"%d\r\n",sdslen(info))); + if (server.masterhost) { + info = sdscatprintf(info, + "master_host:%s\r\n" + "master_port:%d\r\n" + "master_link_status:%s\r\n" + "master_last_io_seconds_ago:%d\r\n" + ,server.masterhost, + server.masterport, + (server.replstate == REDIS_REPL_CONNECTED) ? + "up" : "down", + (int)(time(NULL)-server.master->lastinteraction) + ); + } + for (j = 0; j < server.dbnum; j++) { + long long keys, vkeys; + + keys = dictSize(server.db[j].dict); + vkeys = dictSize(server.db[j].expires); + if (keys || vkeys) { + info = sdscatprintf(info, "db%d: keys=%lld,expires=%lld\r\n", + j, keys, vkeys); + } + } + addReplySds(c,sdscatprintf(sdsempty(),"$%d\r\n",sdslen(info))); addReplySds(c,info); addReply(c,shared.crlf); } -/* =============================== Replication ============================= */ +static void monitorCommand(redisClient *c) { + /* ignore MONITOR if aleady slave or in monitor mode */ + if (c->flags & REDIS_SLAVE) return; -/* Send the whole output buffer syncronously to the slave. This a general operation in theory, but it is actually useful only for replication. */ -static int flushClientOutput(redisClient *c) { - int retval; - time_t start = time(NULL); + c->flags |= (REDIS_SLAVE|REDIS_MONITOR); + c->slaveseldb = 0; + if (!listAddNodeTail(server.monitors,c)) oom("listAddNodeTail"); + addReply(c,shared.ok); +} - while(listLength(c->reply)) { - if (time(NULL)-start > 5) return REDIS_ERR; /* 5 seconds timeout */ - retval = aeWait(c->fd,AE_WRITABLE,1000); - if (retval == -1) { - return REDIS_ERR; - } else if (retval & AE_WRITABLE) { - sendReplyToClient(NULL, c->fd, c, AE_WRITABLE); +/* ================================= Expire ================================= */ +static int removeExpire(redisDb *db, robj *key) { + if (dictDelete(db->expires,key) == DICT_OK) { + return 1; + } else { + return 0; + } +} + +static int setExpire(redisDb *db, robj *key, time_t when) { + if (dictAdd(db->expires,key,(void*)when) == DICT_ERR) { + return 0; + } else { + incrRefCount(key); + return 1; + } +} + +/* 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) */ +static time_t getExpire(redisDb *db, robj *key) { + dictEntry *de; + + /* No expire? return ASAP */ + if (dictSize(db->expires) == 0 || + (de = dictFind(db->expires,key)) == NULL) return -1; + + return (time_t) dictGetEntryVal(de); +} + +static int expireIfNeeded(redisDb *db, robj *key) { + time_t when; + dictEntry *de; + + /* No expire? return ASAP */ + if (dictSize(db->expires) == 0 || + (de = dictFind(db->expires,key)) == NULL) return 0; + + /* Lookup the expire */ + when = (time_t) dictGetEntryVal(de); + if (time(NULL) <= when) return 0; + + /* Delete the key */ + dictDelete(db->expires,key); + return dictDelete(db->dict,key) == DICT_OK; +} + +static int deleteIfVolatile(redisDb *db, robj *key) { + dictEntry *de; + + /* No expire? return ASAP */ + if (dictSize(db->expires) == 0 || + (de = dictFind(db->expires,key)) == NULL) return 0; + + /* Delete the key */ + server.dirty++; + dictDelete(db->expires,key); + return dictDelete(db->dict,key) == DICT_OK; +} + +static void expireCommand(redisClient *c) { + dictEntry *de; + int seconds = atoi(c->argv[2]->ptr); + + de = dictFind(c->db->dict,c->argv[1]); + if (de == NULL) { + addReply(c,shared.czero); + return; + } + if (seconds <= 0) { + addReply(c, shared.czero); + return; + } else { + time_t when = time(NULL)+seconds; + if (setExpire(c->db,c->argv[1],when)) { + addReply(c,shared.cone); + server.dirty++; + } else { + addReply(c,shared.czero); } + return; } - return REDIS_OK; } -static int syncWrite(int fd, void *ptr, ssize_t size, int timeout) { +static void ttlCommand(redisClient *c) { + time_t expire; + int ttl = -1; + + expire = getExpire(c->db,c->argv[1]); + if (expire != -1) { + ttl = (int) (expire-time(NULL)); + if (ttl < 0) ttl = -1; + } + addReplySds(c,sdscatprintf(sdsempty(),":%d\r\n",ttl)); +} + +/* =============================== Replication ============================= */ + +static int syncWrite(int fd, char *ptr, ssize_t size, int timeout) { ssize_t nwritten, ret = size; time_t start = time(NULL); @@ -2857,7 +3804,7 @@ static int syncWrite(int fd, void *ptr, ssize_t size, int timeout) { return ret; } -static int syncRead(int fd, void *ptr, ssize_t size, int timeout) { +static int syncRead(int fd, char *ptr, ssize_t size, int timeout) { ssize_t nread, totread = 0; time_t start = time(NULL); @@ -2900,44 +3847,171 @@ static int syncReadLine(int fd, char *ptr, ssize_t size, int timeout) { } static void syncCommand(redisClient *c) { - struct stat sb; - int fd = -1, len; - time_t start = time(NULL); - char sizebuf[32]; - - redisLog(REDIS_NOTICE,"Slave ask for syncronization"); - if (flushClientOutput(c) == REDIS_ERR || saveDb(server.dbfilename) != REDIS_OK) - goto closeconn; - - fd = open(server.dbfilename, O_RDONLY); - if (fd == -1 || fstat(fd,&sb) == -1) goto closeconn; - len = sb.st_size; - - snprintf(sizebuf,32,"%d\r\n",len); - if (syncWrite(c->fd,sizebuf,strlen(sizebuf),5) == -1) goto closeconn; - while(len) { - char buf[1024]; - int nread; + /* ignore SYNC if aleady slave or in monitor mode */ + if (c->flags & REDIS_SLAVE) return; + + /* SYNC can't be issued when the server has pending data to send to + * the client about already issued commands. We need a fresh reply + * buffer registering the differences between the BGSAVE and the current + * dataset, so that we can copy to other slaves if needed. */ + if (listLength(c->reply) != 0) { + addReplySds(c,sdsnew("-ERR SYNC is invalid with pending input\r\n")); + return; + } - if (time(NULL)-start > REDIS_MAX_SYNC_TIME) goto closeconn; - nread = read(fd,buf,1024); - if (nread == -1) goto closeconn; - len -= nread; - if (syncWrite(c->fd,buf,nread,5) == -1) goto closeconn; + redisLog(REDIS_NOTICE,"Slave ask for synchronization"); + /* Here we need to check if there is a background saving operation + * in progress, or if it is required to start one */ + if (server.bgsaveinprogress) { + /* Ok a background save is in progress. Let's check if it is a good + * one for replication, i.e. if there is another slave that is + * registering differences since the server forked to save */ + redisClient *slave; + listNode *ln; + + listRewind(server.slaves); + while((ln = listYield(server.slaves))) { + slave = ln->value; + if (slave->replstate == REDIS_REPL_WAIT_BGSAVE_END) break; + } + if (ln) { + /* Perfect, the server is already registering differences for + * another slave. Set the right state, and copy the buffer. */ + listRelease(c->reply); + c->reply = listDup(slave->reply); + if (!c->reply) oom("listDup copying slave reply list"); + c->replstate = REDIS_REPL_WAIT_BGSAVE_END; + redisLog(REDIS_NOTICE,"Waiting for end of BGSAVE for SYNC"); + } else { + /* No way, we need to wait for the next BGSAVE in order to + * register differences */ + c->replstate = REDIS_REPL_WAIT_BGSAVE_START; + redisLog(REDIS_NOTICE,"Waiting for next BGSAVE for SYNC"); + } + } else { + /* Ok we don't have a BGSAVE in progress, let's start one */ + redisLog(REDIS_NOTICE,"Starting BGSAVE for SYNC"); + if (rdbSaveBackground(server.dbfilename) != REDIS_OK) { + redisLog(REDIS_NOTICE,"Replication failed, can't BGSAVE"); + addReplySds(c,sdsnew("-ERR Unalbe to perform background save\r\n")); + return; + } + c->replstate = REDIS_REPL_WAIT_BGSAVE_END; } - if (syncWrite(c->fd,"\r\n",2,5) == -1) goto closeconn; - close(fd); + c->repldbfd = -1; c->flags |= REDIS_SLAVE; c->slaveseldb = 0; if (!listAddNodeTail(server.slaves,c)) oom("listAddNodeTail"); - redisLog(REDIS_NOTICE,"Syncronization with slave succeeded"); return; +} -closeconn: - if (fd != -1) close(fd); - c->flags |= REDIS_CLOSE; - redisLog(REDIS_WARNING,"Syncronization with slave failed"); - return; +static void sendBulkToSlave(aeEventLoop *el, int fd, void *privdata, int mask) { + redisClient *slave = privdata; + REDIS_NOTUSED(el); + REDIS_NOTUSED(mask); + char buf[REDIS_IOBUF_LEN]; + ssize_t nwritten, buflen; + + if (slave->repldboff == 0) { + /* Write the bulk write count before to transfer the DB. In theory here + * we don't know how much room there is in the output buffer of the + * socket, but in pratice SO_SNDLOWAT (the minimum count for output + * operations) will never be smaller than the few bytes we need. */ + sds bulkcount; + + bulkcount = sdscatprintf(sdsempty(),"$%lld\r\n",(unsigned long long) + slave->repldbsize); + if (write(fd,bulkcount,sdslen(bulkcount)) != (signed)sdslen(bulkcount)) + { + sdsfree(bulkcount); + freeClient(slave); + return; + } + sdsfree(bulkcount); + } + lseek(slave->repldbfd,slave->repldboff,SEEK_SET); + buflen = read(slave->repldbfd,buf,REDIS_IOBUF_LEN); + if (buflen <= 0) { + redisLog(REDIS_WARNING,"Read error sending DB to slave: %s", + (buflen == 0) ? "premature EOF" : strerror(errno)); + freeClient(slave); + return; + } + if ((nwritten = write(fd,buf,buflen)) == -1) { + redisLog(REDIS_DEBUG,"Write error sending DB to slave: %s", + strerror(errno)); + freeClient(slave); + return; + } + slave->repldboff += nwritten; + if (slave->repldboff == slave->repldbsize) { + close(slave->repldbfd); + slave->repldbfd = -1; + aeDeleteFileEvent(server.el,slave->fd,AE_WRITABLE); + slave->replstate = REDIS_REPL_ONLINE; + if (aeCreateFileEvent(server.el, slave->fd, AE_WRITABLE, + sendReplyToClient, slave, NULL) == AE_ERR) { + freeClient(slave); + return; + } + addReplySds(slave,sdsempty()); + redisLog(REDIS_NOTICE,"Synchronization with slave succeeded"); + } +} + +/* This function is called at the end of every backgrond saving. + * The argument bgsaveerr is REDIS_OK if the background saving succeeded + * otherwise REDIS_ERR is passed to the function. + * + * The goal of this function is to handle slaves waiting for a successful + * background saving in order to perform non-blocking synchronization. */ +static void updateSlavesWaitingBgsave(int bgsaveerr) { + listNode *ln; + int startbgsave = 0; + + listRewind(server.slaves); + while((ln = listYield(server.slaves))) { + redisClient *slave = ln->value; + + if (slave->replstate == REDIS_REPL_WAIT_BGSAVE_START) { + startbgsave = 1; + slave->replstate = REDIS_REPL_WAIT_BGSAVE_END; + } else if (slave->replstate == REDIS_REPL_WAIT_BGSAVE_END) { + struct redis_stat buf; + + if (bgsaveerr != REDIS_OK) { + freeClient(slave); + redisLog(REDIS_WARNING,"SYNC failed. BGSAVE child returned an error"); + continue; + } + if ((slave->repldbfd = open(server.dbfilename,O_RDONLY)) == -1 || + redis_fstat(slave->repldbfd,&buf) == -1) { + freeClient(slave); + redisLog(REDIS_WARNING,"SYNC failed. Can't open/stat DB after BGSAVE: %s", strerror(errno)); + continue; + } + slave->repldboff = 0; + slave->repldbsize = buf.st_size; + slave->replstate = REDIS_REPL_SEND_BULK; + aeDeleteFileEvent(server.el,slave->fd,AE_WRITABLE); + if (aeCreateFileEvent(server.el, slave->fd, AE_WRITABLE, sendBulkToSlave, slave, NULL) == AE_ERR) { + freeClient(slave); + continue; + } + } + } + if (startbgsave) { + if (rdbSaveBackground(server.dbfilename) != REDIS_OK) { + listRewind(server.slaves); + redisLog(REDIS_WARNING,"SYNC failed. BGSAVE failed"); + while((ln = listYield(server.slaves))) { + redisClient *slave = ln->value; + + if (slave->replstate == REDIS_REPL_WAIT_BGSAVE_START) + freeClient(slave); + } + } + } } static int syncWithMaster(void) { @@ -2959,13 +4033,13 @@ static int syncWithMaster(void) { return REDIS_ERR; } /* Read the bulk write count */ - if (syncReadLine(fd,buf,1024,5) == -1) { + if (syncReadLine(fd,buf,1024,3600) == -1) { close(fd); redisLog(REDIS_WARNING,"I/O error reading bulk count from MASTER: %s", strerror(errno)); return REDIS_ERR; } - dumpsize = atoi(buf); + dumpsize = atoi(buf+1); redisLog(REDIS_NOTICE,"Receiving %d bytes data dump from MASTER",dumpsize); /* Read the bulk write data on a temp file */ snprintf(tmpfile,256,"temp-%d.%ld.rdb",(int)time(NULL),(long int)random()); @@ -3003,7 +4077,7 @@ static int syncWithMaster(void) { return REDIS_ERR; } emptyDb(); - if (loadDb(server.dbfilename) != REDIS_OK) { + if (rdbLoad(server.dbfilename) != REDIS_OK) { redisLog(REDIS_WARNING,"Failed trying to load the MASTER synchronization DB from disk"); close(fd); return REDIS_ERR; @@ -3014,8 +4088,341 @@ static int syncWithMaster(void) { return REDIS_OK; } +static void slaveofCommand(redisClient *c) { + if (!strcasecmp(c->argv[1]->ptr,"no") && + !strcasecmp(c->argv[2]->ptr,"one")) { + if (server.masterhost) { + sdsfree(server.masterhost); + server.masterhost = NULL; + if (server.master) freeClient(server.master); + server.replstate = REDIS_REPL_NONE; + redisLog(REDIS_NOTICE,"MASTER MODE enabled (user request)"); + } + } else { + sdsfree(server.masterhost); + server.masterhost = sdsdup(c->argv[1]->ptr); + server.masterport = atoi(c->argv[2]->ptr); + if (server.master) freeClient(server.master); + server.replstate = REDIS_REPL_CONNECT; + redisLog(REDIS_NOTICE,"SLAVE OF %s:%d enabled (user request)", + server.masterhost, server.masterport); + } + addReply(c,shared.ok); +} + +/* ============================ Maxmemory directive ======================== */ + +/* This function gets called when 'maxmemory' is set on the config file to limit + * the max memory used by the server, and we are out of memory. + * This function will try to, in order: + * + * - Free objects from the free list + * - Try to remove keys with an EXPIRE set + * + * It is not possible to free enough memory to reach used-memory < maxmemory + * the server will start refusing commands that will enlarge even more the + * memory usage. + */ +static void freeMemoryIfNeeded(void) { + while (server.maxmemory && zmalloc_used_memory() > server.maxmemory) { + if (listLength(server.objfreelist)) { + robj *o; + + listNode *head = listFirst(server.objfreelist); + o = listNodeValue(head); + listDelNode(server.objfreelist,head); + zfree(o); + } else { + int j, k, freed = 0; + + for (j = 0; j < server.dbnum; j++) { + int minttl = -1; + robj *minkey = NULL; + struct dictEntry *de; + + if (dictSize(server.db[j].expires)) { + freed = 1; + /* From a sample of three keys drop the one nearest to + * the natural expire */ + for (k = 0; k < 3; k++) { + time_t t; + + de = dictGetRandomKey(server.db[j].expires); + t = (time_t) dictGetEntryVal(de); + if (minttl == -1 || t < minttl) { + minkey = dictGetEntryKey(de); + minttl = t; + } + } + deleteKey(server.db+j,minkey); + } + } + if (!freed) return; /* nothing to free... */ + } + } +} + +/* ================================= Debugging ============================== */ + +static void debugCommand(redisClient *c) { + if (!strcasecmp(c->argv[1]->ptr,"segfault")) { + *((char*)-1) = 'x'; + } else if (!strcasecmp(c->argv[1]->ptr,"object") && c->argc == 3) { + dictEntry *de = dictFind(c->db->dict,c->argv[2]); + robj *key, *val; + + if (!de) { + addReply(c,shared.nokeyerr); + return; + } + key = dictGetEntryKey(de); + val = dictGetEntryVal(de); + addReplySds(c,sdscatprintf(sdsempty(), + "+Key at:%p refcount:%d, value at:%p refcount:%d\r\n", + key, key->refcount, val, val->refcount)); + } else { + addReplySds(c,sdsnew( + "-ERR Syntax error, try DEBUG [SEGFAULT|OBJECT ]\r\n")); + } +} + +#ifdef HAVE_BACKTRACE +static struct redisFunctionSym symsTable[] = { +{"freeStringObject", (unsigned long)freeStringObject}, +{"freeListObject", (unsigned long)freeListObject}, +{"freeSetObject", (unsigned long)freeSetObject}, +{"decrRefCount", (unsigned long)decrRefCount}, +{"createObject", (unsigned long)createObject}, +{"freeClient", (unsigned long)freeClient}, +{"rdbLoad", (unsigned long)rdbLoad}, +{"addReply", (unsigned long)addReply}, +{"addReplySds", (unsigned long)addReplySds}, +{"incrRefCount", (unsigned long)incrRefCount}, +{"rdbSaveBackground", (unsigned long)rdbSaveBackground}, +{"createStringObject", (unsigned long)createStringObject}, +{"replicationFeedSlaves", (unsigned long)replicationFeedSlaves}, +{"syncWithMaster", (unsigned long)syncWithMaster}, +{"tryObjectSharing", (unsigned long)tryObjectSharing}, +{"removeExpire", (unsigned long)removeExpire}, +{"expireIfNeeded", (unsigned long)expireIfNeeded}, +{"deleteIfVolatile", (unsigned long)deleteIfVolatile}, +{"deleteKey", (unsigned long)deleteKey}, +{"getExpire", (unsigned long)getExpire}, +{"setExpire", (unsigned long)setExpire}, +{"updateSlavesWaitingBgsave", (unsigned long)updateSlavesWaitingBgsave}, +{"freeMemoryIfNeeded", (unsigned long)freeMemoryIfNeeded}, +{"authCommand", (unsigned long)authCommand}, +{"pingCommand", (unsigned long)pingCommand}, +{"echoCommand", (unsigned long)echoCommand}, +{"setCommand", (unsigned long)setCommand}, +{"setnxCommand", (unsigned long)setnxCommand}, +{"getCommand", (unsigned long)getCommand}, +{"delCommand", (unsigned long)delCommand}, +{"existsCommand", (unsigned long)existsCommand}, +{"incrCommand", (unsigned long)incrCommand}, +{"decrCommand", (unsigned long)decrCommand}, +{"incrbyCommand", (unsigned long)incrbyCommand}, +{"decrbyCommand", (unsigned long)decrbyCommand}, +{"selectCommand", (unsigned long)selectCommand}, +{"randomkeyCommand", (unsigned long)randomkeyCommand}, +{"keysCommand", (unsigned long)keysCommand}, +{"dbsizeCommand", (unsigned long)dbsizeCommand}, +{"lastsaveCommand", (unsigned long)lastsaveCommand}, +{"saveCommand", (unsigned long)saveCommand}, +{"bgsaveCommand", (unsigned long)bgsaveCommand}, +{"shutdownCommand", (unsigned long)shutdownCommand}, +{"moveCommand", (unsigned long)moveCommand}, +{"renameCommand", (unsigned long)renameCommand}, +{"renamenxCommand", (unsigned long)renamenxCommand}, +{"lpushCommand", (unsigned long)lpushCommand}, +{"rpushCommand", (unsigned long)rpushCommand}, +{"lpopCommand", (unsigned long)lpopCommand}, +{"rpopCommand", (unsigned long)rpopCommand}, +{"llenCommand", (unsigned long)llenCommand}, +{"lindexCommand", (unsigned long)lindexCommand}, +{"lrangeCommand", (unsigned long)lrangeCommand}, +{"ltrimCommand", (unsigned long)ltrimCommand}, +{"typeCommand", (unsigned long)typeCommand}, +{"lsetCommand", (unsigned long)lsetCommand}, +{"saddCommand", (unsigned long)saddCommand}, +{"sremCommand", (unsigned long)sremCommand}, +{"smoveCommand", (unsigned long)smoveCommand}, +{"sismemberCommand", (unsigned long)sismemberCommand}, +{"scardCommand", (unsigned long)scardCommand}, +{"spopCommand", (unsigned long)spopCommand}, +{"sinterCommand", (unsigned long)sinterCommand}, +{"sinterstoreCommand", (unsigned long)sinterstoreCommand}, +{"sunionCommand", (unsigned long)sunionCommand}, +{"sunionstoreCommand", (unsigned long)sunionstoreCommand}, +{"sdiffCommand", (unsigned long)sdiffCommand}, +{"sdiffstoreCommand", (unsigned long)sdiffstoreCommand}, +{"syncCommand", (unsigned long)syncCommand}, +{"flushdbCommand", (unsigned long)flushdbCommand}, +{"flushallCommand", (unsigned long)flushallCommand}, +{"sortCommand", (unsigned long)sortCommand}, +{"lremCommand", (unsigned long)lremCommand}, +{"infoCommand", (unsigned long)infoCommand}, +{"mgetCommand", (unsigned long)mgetCommand}, +{"monitorCommand", (unsigned long)monitorCommand}, +{"expireCommand", (unsigned long)expireCommand}, +{"getSetCommand", (unsigned long)getSetCommand}, +{"ttlCommand", (unsigned long)ttlCommand}, +{"slaveofCommand", (unsigned long)slaveofCommand}, +{"debugCommand", (unsigned long)debugCommand}, +{"processCommand", (unsigned long)processCommand}, +{"setupSigSegvAction", (unsigned long)setupSigSegvAction}, +{"readQueryFromClient", (unsigned long)readQueryFromClient}, +{"rdbRemoveTempFile", (unsigned long)rdbRemoveTempFile}, +{NULL,0} +}; + +/* This function try to convert a pointer into a function name. It's used in + * oreder to provide a backtrace under segmentation fault that's able to + * display functions declared as static (otherwise the backtrace is useless). */ +static char *findFuncName(void *pointer, unsigned long *offset){ + int i, ret = -1; + unsigned long off, minoff = 0; + + /* Try to match against the Symbol with the smallest offset */ + for (i=0; symsTable[i].pointer; i++) { + unsigned long lp = (unsigned long) pointer; + + if (lp != (unsigned long)-1 && lp >= symsTable[i].pointer) { + off=lp-symsTable[i].pointer; + if (ret < 0 || off < minoff) { + minoff=off; + ret=i; + } + } + } + if (ret == -1) return NULL; + *offset = minoff; + return symsTable[ret].name; +} + +static void *getMcontextEip(ucontext_t *uc) { +#if defined(__FreeBSD__) + return (void*) uc->uc_mcontext.mc_eip; +#elif defined(__dietlibc__) + return (void*) uc->uc_mcontext.eip; +#elif defined(__APPLE__) && !defined(MAC_OS_X_VERSION_10_6) + return (void*) uc->uc_mcontext->__ss.__eip; +#elif defined(__APPLE__) && defined(MAC_OS_X_VERSION_10_6) + #ifdef _STRUCT_X86_THREAD_STATE64 + return (void*) uc->uc_mcontext->__ss.__rip; + #else + return (void*) uc->uc_mcontext->__ss.__eip; + #endif +#elif defined(__i386__) || defined(__X86_64__) /* Linux x86 */ + return (void*) uc->uc_mcontext.gregs[REG_EIP]; +#elif defined(__ia64__) /* Linux IA64 */ + return (void*) uc->uc_mcontext.sc_ip; +#else + return NULL; +#endif +} + +static void segvHandler(int sig, siginfo_t *info, void *secret) { + void *trace[100]; + char **messages = NULL; + int i, trace_size = 0; + unsigned long offset=0; + time_t uptime = time(NULL)-server.stat_starttime; + ucontext_t *uc = (ucontext_t*) secret; + REDIS_NOTUSED(info); + + redisLog(REDIS_WARNING, + "======= Ooops! Redis %s got signal: -%d- =======", REDIS_VERSION, sig); + redisLog(REDIS_WARNING, "%s", sdscatprintf(sdsempty(), + "redis_version:%s; " + "uptime_in_seconds:%d; " + "connected_clients:%d; " + "connected_slaves:%d; " + "used_memory:%zu; " + "changes_since_last_save:%lld; " + "bgsave_in_progress:%d; " + "last_save_time:%d; " + "total_connections_received:%lld; " + "total_commands_processed:%lld; " + "role:%s;" + ,REDIS_VERSION, + uptime, + listLength(server.clients)-listLength(server.slaves), + listLength(server.slaves), + server.usedmemory, + server.dirty, + server.bgsaveinprogress, + server.lastsave, + server.stat_numconnections, + server.stat_numcommands, + server.masterhost == NULL ? "master" : "slave" + )); + + trace_size = backtrace(trace, 100); + /* overwrite sigaction with caller's address */ + if (getMcontextEip(uc) != NULL) { + trace[1] = getMcontextEip(uc); + } + messages = backtrace_symbols(trace, trace_size); + + for (i=1; i STDERR_FILENO) close(fd); } /* Try to write the pid file */ - fp = fopen("/var/run/redis.pid","w"); + fp = fopen(server.pidfile,"w"); if (fp) { fprintf(fp,"%d\n",getpid()); fclose(fp); @@ -3048,15 +4455,20 @@ int main(int argc, char **argv) { } else if (argc > 2) { fprintf(stderr,"Usage: ./redis-server [/path/to/redis.conf]\n"); exit(1); + } else { + redisLog(REDIS_WARNING,"Warning: no config file specified, using the default config. In order to specify a config file use 'redis-server /path/to/redis.conf'"); } initServer(); if (server.daemonize) daemonize(); redisLog(REDIS_NOTICE,"Server started, Redis version " REDIS_VERSION); - if (loadDb(server.dbfilename) == REDIS_OK) +#ifdef __linux__ + linuxOvercommitMemoryWarning(); +#endif + if (rdbLoad(server.dbfilename) == REDIS_OK) redisLog(REDIS_NOTICE,"DB loaded from disk"); if (aeCreateFileEvent(server.el, server.fd, AE_READABLE, acceptHandler, NULL, NULL) == AE_ERR) oom("creating file event"); - redisLog(REDIS_NOTICE,"The server is now ready to accept connections"); + redisLog(REDIS_NOTICE,"The server is now ready to accept connections on port %d", server.port); aeMain(server.el); aeDeleteEventLoop(server.el); return 0;