]> git.saurik.com Git - redis.git/blob - src/redis.h
added new RDB codes for ziplist encoded lists and intset encodeed sets
[redis.git] / src / redis.h
1 #ifndef __REDIS_H
2 #define __REDIS_H
3
4 #include "fmacros.h"
5 #include "config.h"
6
7 #if defined(__sun)
8 #include "solarisfixes.h"
9 #endif
10
11 #include <stdio.h>
12 #include <stdlib.h>
13 #include <string.h>
14 #include <time.h>
15 #include <limits.h>
16 #include <unistd.h>
17 #include <errno.h>
18 #include <inttypes.h>
19 #include <pthread.h>
20 #include <syslog.h>
21
22 #include "ae.h" /* Event driven programming library */
23 #include "sds.h" /* Dynamic safe strings */
24 #include "dict.h" /* Hash tables */
25 #include "adlist.h" /* Linked lists */
26 #include "zmalloc.h" /* total memory usage aware version of malloc/free */
27 #include "anet.h" /* Networking the easy way */
28 #include "zipmap.h" /* Compact string -> string data structure */
29 #include "ziplist.h" /* Compact list data structure */
30 #include "intset.h" /* Compact integer set structure */
31 #include "version.h"
32
33 /* Error codes */
34 #define REDIS_OK 0
35 #define REDIS_ERR -1
36
37 /* Static server configuration */
38 #define REDIS_SERVERPORT 6379 /* TCP port */
39 #define REDIS_MAXIDLETIME (60*5) /* default client timeout */
40 #define REDIS_IOBUF_LEN 1024
41 #define REDIS_LOADBUF_LEN 1024
42 #define REDIS_STATIC_ARGS 8
43 #define REDIS_DEFAULT_DBNUM 16
44 #define REDIS_CONFIGLINE_MAX 1024
45 #define REDIS_MAX_SYNC_TIME 60 /* Slave can't take more to sync */
46 #define REDIS_EXPIRELOOKUPS_PER_CRON 10 /* lookup 10 expires per loop */
47 #define REDIS_MAX_WRITE_PER_EVENT (1024*64)
48 #define REDIS_REQUEST_MAX_SIZE (1024*1024*256) /* max bytes in inline command */
49 #define REDIS_SHARED_INTEGERS 10000
50 #define REDIS_REPLY_CHUNK_BYTES (5*1500) /* 5 TCP packets with default MTU */
51 #define REDIS_MAX_LOGMSG_LEN 1024 /* Default maximum length of syslog messages */
52
53 /* Hash table parameters */
54 #define REDIS_HT_MINFILL 10 /* Minimal hash table fill 10% */
55
56 /* Command flags:
57 * REDIS_CMD_DENYOOM:
58 * Commands marked with this flag will return an error when 'maxmemory' is
59 * set and the server is using more than 'maxmemory' bytes of memory.
60 * In short: commands with this flag are denied on low memory conditions.
61 * REDIS_CMD_FORCE_REPLICATION:
62 * Force replication even if dirty is 0. */
63 #define REDIS_CMD_DENYOOM 4
64 #define REDIS_CMD_FORCE_REPLICATION 8
65
66 /* Object types */
67 #define REDIS_STRING 0
68 #define REDIS_LIST 1
69 #define REDIS_SET 2
70 #define REDIS_ZSET 3
71 #define REDIS_HASH 4
72 #define REDIS_VMPOINTER 8
73 /* Object types only used for persistence in .rdb files */
74 #define REDIS_HASH_ZIPMAP 9
75 #define REDIS_LIST_ZIPLIST 10
76 #define REDIS_SET_INTSET 11
77
78 /* Objects encoding. Some kind of objects like Strings and Hashes can be
79 * internally represented in multiple ways. The 'encoding' field of the object
80 * is set to one of this fields for this object. */
81 #define REDIS_ENCODING_RAW 0 /* Raw representation */
82 #define REDIS_ENCODING_INT 1 /* Encoded as integer */
83 #define REDIS_ENCODING_HT 2 /* Encoded as hash table */
84 #define REDIS_ENCODING_ZIPMAP 3 /* Encoded as zipmap */
85 #define REDIS_ENCODING_LINKEDLIST 4 /* Encoded as regular linked list */
86 #define REDIS_ENCODING_ZIPLIST 5 /* Encoded as ziplist */
87 #define REDIS_ENCODING_INTSET 6 /* Encoded as intset */
88
89 /* Object types only used for dumping to disk */
90 #define REDIS_EXPIRETIME 253
91 #define REDIS_SELECTDB 254
92 #define REDIS_EOF 255
93
94 /* Defines related to the dump file format. To store 32 bits lengths for short
95 * keys requires a lot of space, so we check the most significant 2 bits of
96 * the first byte to interpreter the length:
97 *
98 * 00|000000 => if the two MSB are 00 the len is the 6 bits of this byte
99 * 01|000000 00000000 => 01, the len is 14 byes, 6 bits + 8 bits of next byte
100 * 10|000000 [32 bit integer] => if it's 01, a full 32 bit len will follow
101 * 11|000000 this means: specially encoded object will follow. The six bits
102 * number specify the kind of object that follows.
103 * See the REDIS_RDB_ENC_* defines.
104 *
105 * Lenghts up to 63 are stored using a single byte, most DB keys, and may
106 * values, will fit inside. */
107 #define REDIS_RDB_6BITLEN 0
108 #define REDIS_RDB_14BITLEN 1
109 #define REDIS_RDB_32BITLEN 2
110 #define REDIS_RDB_ENCVAL 3
111 #define REDIS_RDB_LENERR UINT_MAX
112
113 /* When a length of a string object stored on disk has the first two bits
114 * set, the remaining two bits specify a special encoding for the object
115 * accordingly to the following defines: */
116 #define REDIS_RDB_ENC_INT8 0 /* 8 bit signed integer */
117 #define REDIS_RDB_ENC_INT16 1 /* 16 bit signed integer */
118 #define REDIS_RDB_ENC_INT32 2 /* 32 bit signed integer */
119 #define REDIS_RDB_ENC_LZF 3 /* string compressed with FASTLZ */
120
121 /* Scheduled IO opeations flags. */
122 #define REDIS_IO_LOAD 1
123 #define REDIS_IO_SAVE 2
124 #define REDIS_IO_LOADINPROG 4
125 #define REDIS_IO_SAVEINPROG 8
126
127 /* Generic IO flags */
128 #define REDIS_IO_ONLYLOADS 1
129 #define REDIS_IO_ASAP 2
130
131 #define REDIS_MAX_COMPLETED_JOBS_PROCESSED 1
132 #define REDIS_THREAD_STACK_SIZE (1024*1024*4)
133
134 /* Client flags */
135 #define REDIS_SLAVE 1 /* This client is a slave server */
136 #define REDIS_MASTER 2 /* This client is a master server */
137 #define REDIS_MONITOR 4 /* This client is a slave monitor, see MONITOR */
138 #define REDIS_MULTI 8 /* This client is in a MULTI context */
139 #define REDIS_BLOCKED 16 /* The client is waiting in a blocking operation */
140 #define REDIS_IO_WAIT 32 /* The client is waiting for Virtual Memory I/O */
141 #define REDIS_DIRTY_CAS 64 /* Watched keys modified. EXEC will fail. */
142 #define REDIS_CLOSE_AFTER_REPLY 128 /* Close after writing entire reply. */
143 #define REDIS_UNBLOCKED 256 /* This client was unblocked and is stored in
144 server.unblocked_clients */
145
146 /* Client request types */
147 #define REDIS_REQ_INLINE 1
148 #define REDIS_REQ_MULTIBULK 2
149
150 /* Slave replication state - slave side */
151 #define REDIS_REPL_NONE 0 /* No active replication */
152 #define REDIS_REPL_CONNECT 1 /* Must connect to master */
153 #define REDIS_REPL_TRANSFER 2 /* Receiving .rdb from master */
154 #define REDIS_REPL_CONNECTED 3 /* Connected to master */
155
156 /* Slave replication state - from the point of view of master
157 * Note that in SEND_BULK and ONLINE state the slave receives new updates
158 * in its output queue. In the WAIT_BGSAVE state instead the server is waiting
159 * to start the next background saving in order to send updates to it. */
160 #define REDIS_REPL_WAIT_BGSAVE_START 3 /* master waits bgsave to start feeding it */
161 #define REDIS_REPL_WAIT_BGSAVE_END 4 /* master waits bgsave to start bulk DB transmission */
162 #define REDIS_REPL_SEND_BULK 5 /* master is sending the bulk DB */
163 #define REDIS_REPL_ONLINE 6 /* bulk DB already transmitted, receive updates */
164
165 /* List related stuff */
166 #define REDIS_HEAD 0
167 #define REDIS_TAIL 1
168
169 /* Sort operations */
170 #define REDIS_SORT_GET 0
171 #define REDIS_SORT_ASC 1
172 #define REDIS_SORT_DESC 2
173 #define REDIS_SORTKEY_MAX 1024
174
175 /* Log levels */
176 #define REDIS_DEBUG 0
177 #define REDIS_VERBOSE 1
178 #define REDIS_NOTICE 2
179 #define REDIS_WARNING 3
180
181 /* Anti-warning macro... */
182 #define REDIS_NOTUSED(V) ((void) V)
183
184 #define ZSKIPLIST_MAXLEVEL 32 /* Should be enough for 2^32 elements */
185 #define ZSKIPLIST_P 0.25 /* Skiplist P = 1/4 */
186
187 /* Append only defines */
188 #define APPENDFSYNC_NO 0
189 #define APPENDFSYNC_ALWAYS 1
190 #define APPENDFSYNC_EVERYSEC 2
191
192 /* Zip structure related defaults */
193 #define REDIS_HASH_MAX_ZIPMAP_ENTRIES 512
194 #define REDIS_HASH_MAX_ZIPMAP_VALUE 64
195 #define REDIS_LIST_MAX_ZIPLIST_ENTRIES 512
196 #define REDIS_LIST_MAX_ZIPLIST_VALUE 64
197 #define REDIS_SET_MAX_INTSET_ENTRIES 512
198
199 /* Sets operations codes */
200 #define REDIS_OP_UNION 0
201 #define REDIS_OP_DIFF 1
202 #define REDIS_OP_INTER 2
203
204 /* Redis maxmemory strategies */
205 #define REDIS_MAXMEMORY_VOLATILE_LRU 0
206 #define REDIS_MAXMEMORY_VOLATILE_TTL 1
207 #define REDIS_MAXMEMORY_VOLATILE_RANDOM 2
208 #define REDIS_MAXMEMORY_ALLKEYS_LRU 3
209 #define REDIS_MAXMEMORY_ALLKEYS_RANDOM 4
210 #define REDIS_MAXMEMORY_NO_EVICTION 5
211
212 /* Diskstore background saving thread states */
213 #define REDIS_BGSAVE_THREAD_UNACTIVE 0
214 #define REDIS_BGSAVE_THREAD_ACTIVE 1
215 #define REDIS_BGSAVE_THREAD_DONE_OK 2
216 #define REDIS_BGSAVE_THREAD_DONE_ERR 3
217
218 /* We can print the stacktrace, so our assert is defined this way: */
219 #define redisAssert(_e) ((_e)?(void)0 : (_redisAssert(#_e,__FILE__,__LINE__),_exit(1)))
220 #define redisPanic(_e) _redisPanic(#_e,__FILE__,__LINE__),_exit(1)
221 void _redisAssert(char *estr, char *file, int line);
222 void _redisPanic(char *msg, char *file, int line);
223
224 /*-----------------------------------------------------------------------------
225 * Data types
226 *----------------------------------------------------------------------------*/
227
228 /* A redis object, that is a type able to hold a string / list / set */
229
230 /* The actual Redis Object */
231 #define REDIS_LRU_CLOCK_MAX ((1<<21)-1) /* Max value of obj->lru */
232 #define REDIS_LRU_CLOCK_RESOLUTION 10 /* LRU clock resolution in seconds */
233 typedef struct redisObject {
234 unsigned type:4;
235 unsigned notused:2; /* Not used */
236 unsigned encoding:4;
237 unsigned lru:22; /* lru time (relative to server.lruclock) */
238 int refcount;
239 void *ptr;
240 /* VM fields are only allocated if VM is active, otherwise the
241 * object allocation function will just allocate
242 * sizeof(redisObjct) minus sizeof(redisObjectVM), so using
243 * Redis without VM active will not have any overhead. */
244 } robj;
245
246 /* The VM pointer structure - identifies an object in the swap file.
247 *
248 * This object is stored in place of the value
249 * object in the main key->value hash table representing a database.
250 * Note that the first fields (type, storage) are the same as the redisObject
251 * structure so that vmPointer strucuters can be accessed even when casted
252 * as redisObject structures.
253 *
254 * This is useful as we don't know if a value object is or not on disk, but we
255 * are always able to read obj->storage to check this. For vmPointer
256 * structures "type" is set to REDIS_VMPOINTER (even if without this field
257 * is still possible to check the kind of object from the value of 'storage').*/
258 typedef struct vmPointer {
259 unsigned type:4;
260 unsigned storage:2; /* REDIS_VM_SWAPPED or REDIS_VM_LOADING */
261 unsigned notused:26;
262 unsigned int vtype; /* type of the object stored in the swap file */
263 off_t page; /* the page at witch the object is stored on disk */
264 off_t usedpages; /* number of pages used on disk */
265 } vmpointer;
266
267 /* Macro used to initalize a Redis object allocated on the stack.
268 * Note that this macro is taken near the structure definition to make sure
269 * we'll update it when the structure is changed, to avoid bugs like
270 * bug #85 introduced exactly in this way. */
271 #define initStaticStringObject(_var,_ptr) do { \
272 _var.refcount = 1; \
273 _var.type = REDIS_STRING; \
274 _var.encoding = REDIS_ENCODING_RAW; \
275 _var.ptr = _ptr; \
276 } while(0);
277
278 typedef struct redisDb {
279 dict *dict; /* The keyspace for this DB */
280 dict *expires; /* Timeout of keys with a timeout set */
281 dict *blocking_keys; /* Keys with clients waiting for data (BLPOP) */
282 dict *io_keys; /* Keys with clients waiting for DS I/O */
283 dict *io_negcache; /* Negative caching for disk store */
284 dict *io_queued; /* Queued IO operations hash table */
285 dict *watched_keys; /* WATCHED keys for MULTI/EXEC CAS */
286 int id;
287 } redisDb;
288
289 /* Client MULTI/EXEC state */
290 typedef struct multiCmd {
291 robj **argv;
292 int argc;
293 struct redisCommand *cmd;
294 } multiCmd;
295
296 typedef struct multiState {
297 multiCmd *commands; /* Array of MULTI commands */
298 int count; /* Total number of MULTI commands */
299 } multiState;
300
301 typedef struct blockingState {
302 robj **keys; /* The key we are waiting to terminate a blocking
303 * operation such as BLPOP. Otherwise NULL. */
304 int count; /* Number of blocking keys */
305 time_t timeout; /* Blocking operation timeout. If UNIX current time
306 * is >= timeout then the operation timed out. */
307 robj *target; /* The key that should receive the element,
308 * for BRPOPLPUSH. */
309 } blockingState;
310
311 /* With multiplexing we need to take per-clinet state.
312 * Clients are taken in a liked list. */
313 typedef struct redisClient {
314 int fd;
315 redisDb *db;
316 int dictid;
317 sds querybuf;
318 int argc;
319 robj **argv;
320 int reqtype;
321 int multibulklen; /* number of multi bulk arguments left to read */
322 long bulklen; /* length of bulk argument in multi bulk request */
323 list *reply;
324 int sentlen;
325 time_t lastinteraction; /* time of the last interaction, used for timeout */
326 int flags; /* REDIS_SLAVE | REDIS_MONITOR | REDIS_MULTI ... */
327 int slaveseldb; /* slave selected db, if this client is a slave */
328 int authenticated; /* when requirepass is non-NULL */
329 int replstate; /* replication state if this is a slave */
330 int repldbfd; /* replication DB file descriptor */
331 long repldboff; /* replication DB file offset */
332 off_t repldbsize; /* replication DB file size */
333 multiState mstate; /* MULTI/EXEC state */
334 blockingState bpop; /* blocking state */
335 list *io_keys; /* Keys this client is waiting to be loaded from the
336 * swap file in order to continue. */
337 list *watched_keys; /* Keys WATCHED for MULTI/EXEC CAS */
338 dict *pubsub_channels; /* channels a client is interested in (SUBSCRIBE) */
339 list *pubsub_patterns; /* patterns a client is interested in (SUBSCRIBE) */
340
341 /* Response buffer */
342 int bufpos;
343 char buf[REDIS_REPLY_CHUNK_BYTES];
344 } redisClient;
345
346 struct saveparam {
347 time_t seconds;
348 int changes;
349 };
350
351 struct sharedObjectsStruct {
352 robj *crlf, *ok, *err, *emptybulk, *czero, *cone, *cnegone, *pong, *space,
353 *colon, *nullbulk, *nullmultibulk, *queued,
354 *emptymultibulk, *wrongtypeerr, *nokeyerr, *syntaxerr, *sameobjecterr,
355 *outofrangeerr, *loadingerr, *plus,
356 *select0, *select1, *select2, *select3, *select4,
357 *select5, *select6, *select7, *select8, *select9,
358 *messagebulk, *pmessagebulk, *subscribebulk, *unsubscribebulk, *mbulk3,
359 *mbulk4, *psubscribebulk, *punsubscribebulk,
360 *integers[REDIS_SHARED_INTEGERS];
361 };
362
363 /* Global server state structure */
364 struct redisServer {
365 /* General */
366 pthread_t mainthread;
367 redisDb *db;
368 dict *commands; /* Command table hahs table */
369 aeEventLoop *el;
370 /* Networking */
371 int port;
372 char *bindaddr;
373 char *unixsocket;
374 int ipfd;
375 int sofd;
376 list *clients;
377 list *slaves, *monitors;
378 char neterr[ANET_ERR_LEN];
379 /* RDB / AOF loading information */
380 int loading;
381 off_t loading_total_bytes;
382 off_t loading_loaded_bytes;
383 time_t loading_start_time;
384 /* Fast pointers to often looked up command */
385 struct redisCommand *delCommand, *multiCommand;
386 int cronloops; /* number of times the cron function run */
387 time_t lastsave; /* Unix time of last save succeeede */
388 /* Fields used only for stats */
389 time_t stat_starttime; /* server start time */
390 long long stat_numcommands; /* number of processed commands */
391 long long stat_numconnections; /* number of connections received */
392 long long stat_expiredkeys; /* number of expired keys */
393 long long stat_evictedkeys; /* number of evicted keys (maxmemory) */
394 long long stat_keyspace_hits; /* number of successful lookups of keys */
395 long long stat_keyspace_misses; /* number of failed lookups of keys */
396 /* Configuration */
397 int verbosity;
398 int maxidletime;
399 int dbnum;
400 int daemonize;
401 int appendonly;
402 int appendfsync;
403 int no_appendfsync_on_rewrite;
404 int shutdown_asap;
405 int activerehashing;
406 char *requirepass;
407 /* Persistence */
408 long long dirty; /* changes to DB from the last save */
409 long long dirty_before_bgsave; /* used to restore dirty on failed BGSAVE */
410 time_t lastfsync;
411 int appendfd;
412 int appendseldb;
413 char *pidfile;
414 pid_t bgsavechildpid;
415 pid_t bgrewritechildpid;
416 int bgsavethread_state;
417 pthread_mutex_t bgsavethread_mutex;
418 pthread_t bgsavethread;
419 sds bgrewritebuf; /* buffer taken by parent during oppend only rewrite */
420 sds aofbuf; /* AOF buffer, written before entering the event loop */
421 struct saveparam *saveparams;
422 int saveparamslen;
423 char *dbfilename;
424 int rdbcompression;
425 char *appendfilename;
426 /* Logging */
427 char *logfile;
428 int syslog_enabled;
429 char *syslog_ident;
430 int syslog_facility;
431 /* Replication related */
432 int isslave;
433 /* Slave specific fields */
434 char *masterauth;
435 char *masterhost;
436 int masterport;
437 redisClient *master; /* client that is master for this slave */
438 int replstate; /* replication status if the instance is a slave */
439 off_t repl_transfer_left; /* bytes left reading .rdb */
440 int repl_transfer_s; /* slave -> master SYNC socket */
441 int repl_transfer_fd; /* slave -> master SYNC temp file descriptor */
442 char *repl_transfer_tmpfile; /* slave-> master SYNC temp file name */
443 time_t repl_transfer_lastio; /* unix time of the latest read, for timeout */
444 int repl_serve_stale_data; /* Serve stale data when link is down? */
445 /* Limits */
446 unsigned int maxclients;
447 unsigned long long maxmemory;
448 int maxmemory_policy;
449 int maxmemory_samples;
450 /* Blocked clients */
451 unsigned int bpop_blocked_clients;
452 unsigned int cache_blocked_clients;
453 list *unblocked_clients; /* list of clients to unblock before next loop */
454 list *cache_io_queue; /* IO operations queue */
455 int cache_flush_delay; /* seconds to wait before flushing keys */
456 /* Sort parameters - qsort_r() is only available under BSD so we
457 * have to take this state global, in order to pass it to sortCompare() */
458 int sort_desc;
459 int sort_alpha;
460 int sort_bypattern;
461 /* Virtual memory configuration */
462 int ds_enabled; /* backend disk in redis.conf */
463 char *ds_path; /* location of the disk store on disk */
464 unsigned long long cache_max_memory;
465 /* Zip structure config */
466 size_t hash_max_zipmap_entries;
467 size_t hash_max_zipmap_value;
468 size_t list_max_ziplist_entries;
469 size_t list_max_ziplist_value;
470 size_t set_max_intset_entries;
471 time_t unixtime; /* Unix time sampled every second. */
472 /* Virtual memory I/O threads stuff */
473 /* An I/O thread process an element taken from the io_jobs queue and
474 * put the result of the operation in the io_done list. While the
475 * job is being processed, it's put on io_processing queue. */
476 list *io_newjobs; /* List of VM I/O jobs yet to be processed */
477 list *io_processing; /* List of VM I/O jobs being processed */
478 list *io_processed; /* List of VM I/O jobs already processed */
479 list *io_ready_clients; /* Clients ready to be unblocked. All keys loaded */
480 pthread_mutex_t io_mutex; /* lock to access io_jobs/io_done/io_thread_job */
481 pthread_cond_t io_condvar; /* I/O threads conditional variable */
482 pthread_attr_t io_threads_attr; /* attributes for threads creation */
483 int io_active_threads; /* Number of running I/O threads */
484 int vm_max_threads; /* Max number of I/O threads running at the same time */
485 /* Our main thread is blocked on the event loop, locking for sockets ready
486 * to be read or written, so when a threaded I/O operation is ready to be
487 * processed by the main thread, the I/O thread will use a unix pipe to
488 * awake the main thread. The followings are the two pipe FDs. */
489 int io_ready_pipe_read;
490 int io_ready_pipe_write;
491 /* Virtual memory stats */
492 unsigned long long vm_stats_used_pages;
493 unsigned long long vm_stats_swapped_objects;
494 unsigned long long vm_stats_swapouts;
495 unsigned long long vm_stats_swapins;
496 /* Pubsub */
497 dict *pubsub_channels; /* Map channels to list of subscribed clients */
498 list *pubsub_patterns; /* A list of pubsub_patterns */
499 /* Misc */
500 unsigned lruclock:22; /* clock incrementing every minute, for LRU */
501 unsigned lruclock_padding:10;
502 };
503
504 typedef struct pubsubPattern {
505 redisClient *client;
506 robj *pattern;
507 } pubsubPattern;
508
509 typedef void redisCommandProc(redisClient *c);
510 typedef void redisVmPreloadProc(redisClient *c, struct redisCommand *cmd, int argc, robj **argv);
511 struct redisCommand {
512 char *name;
513 redisCommandProc *proc;
514 int arity;
515 int flags;
516 /* Use a function to determine which keys need to be loaded
517 * in the background prior to executing this command. Takes precedence
518 * over vm_firstkey and others, ignored when NULL */
519 redisVmPreloadProc *vm_preload_proc;
520 /* What keys should be loaded in background when calling this command? */
521 int vm_firstkey; /* The first argument that's a key (0 = no keys) */
522 int vm_lastkey; /* THe last argument that's a key */
523 int vm_keystep; /* The step between first and last key */
524 long long microseconds, calls;
525 };
526
527 struct redisFunctionSym {
528 char *name;
529 unsigned long pointer;
530 };
531
532 typedef struct _redisSortObject {
533 robj *obj;
534 union {
535 double score;
536 robj *cmpobj;
537 } u;
538 } redisSortObject;
539
540 typedef struct _redisSortOperation {
541 int type;
542 robj *pattern;
543 } redisSortOperation;
544
545 /* ZSETs use a specialized version of Skiplists */
546 typedef struct zskiplistNode {
547 robj *obj;
548 double score;
549 struct zskiplistNode *backward;
550 struct zskiplistLevel {
551 struct zskiplistNode *forward;
552 unsigned int span;
553 } level[];
554 } zskiplistNode;
555
556 typedef struct zskiplist {
557 struct zskiplistNode *header, *tail;
558 unsigned long length;
559 int level;
560 } zskiplist;
561
562 typedef struct zset {
563 dict *dict;
564 zskiplist *zsl;
565 } zset;
566
567 /* DIsk store threaded I/O request message */
568 #define REDIS_IOJOB_LOAD 0
569 #define REDIS_IOJOB_SAVE 1
570
571 typedef struct iojob {
572 int type; /* Request type, REDIS_IOJOB_* */
573 redisDb *db;/* Redis database */
574 robj *key; /* This I/O request is about this key */
575 robj *val; /* the value to swap for REDIS_IOJOB_SAVE, otherwise this
576 * field is populated by the I/O thread for REDIS_IOJOB_LOAD. */
577 time_t expire; /* Expire time for this key on REDIS_IOJOB_LOAD */
578 } iojob;
579
580 /* IO operations scheduled -- check dscache.c for more info */
581 typedef struct ioop {
582 int type;
583 redisDb *db;
584 robj *key;
585 time_t ctime; /* This is the creation time of the entry. */
586 } ioop;
587
588 /* Structure to hold list iteration abstraction. */
589 typedef struct {
590 robj *subject;
591 unsigned char encoding;
592 unsigned char direction; /* Iteration direction */
593 unsigned char *zi;
594 listNode *ln;
595 } listTypeIterator;
596
597 /* Structure for an entry while iterating over a list. */
598 typedef struct {
599 listTypeIterator *li;
600 unsigned char *zi; /* Entry in ziplist */
601 listNode *ln; /* Entry in linked list */
602 } listTypeEntry;
603
604 /* Structure to hold set iteration abstraction. */
605 typedef struct {
606 robj *subject;
607 int encoding;
608 int ii; /* intset iterator */
609 dictIterator *di;
610 } setTypeIterator;
611
612 /* Structure to hold hash iteration abstration. Note that iteration over
613 * hashes involves both fields and values. Because it is possible that
614 * not both are required, store pointers in the iterator to avoid
615 * unnecessary memory allocation for fields/values. */
616 typedef struct {
617 int encoding;
618 unsigned char *zi;
619 unsigned char *zk, *zv;
620 unsigned int zklen, zvlen;
621
622 dictIterator *di;
623 dictEntry *de;
624 } hashTypeIterator;
625
626 #define REDIS_HASH_KEY 1
627 #define REDIS_HASH_VALUE 2
628
629 /*-----------------------------------------------------------------------------
630 * Extern declarations
631 *----------------------------------------------------------------------------*/
632
633 extern struct redisServer server;
634 extern struct sharedObjectsStruct shared;
635 extern dictType setDictType;
636 extern dictType zsetDictType;
637 extern double R_Zero, R_PosInf, R_NegInf, R_Nan;
638 dictType hashDictType;
639
640 /*-----------------------------------------------------------------------------
641 * Functions prototypes
642 *----------------------------------------------------------------------------*/
643
644 /* Utils */
645 long long ustime(void);
646
647 /* networking.c -- Networking and Client related operations */
648 redisClient *createClient(int fd);
649 void closeTimedoutClients(void);
650 void freeClient(redisClient *c);
651 void resetClient(redisClient *c);
652 void sendReplyToClient(aeEventLoop *el, int fd, void *privdata, int mask);
653 void addReply(redisClient *c, robj *obj);
654 void *addDeferredMultiBulkLength(redisClient *c);
655 void setDeferredMultiBulkLength(redisClient *c, void *node, long length);
656 void addReplySds(redisClient *c, sds s);
657 void processInputBuffer(redisClient *c);
658 void acceptTcpHandler(aeEventLoop *el, int fd, void *privdata, int mask);
659 void acceptUnixHandler(aeEventLoop *el, int fd, void *privdata, int mask);
660 void readQueryFromClient(aeEventLoop *el, int fd, void *privdata, int mask);
661 void addReplyBulk(redisClient *c, robj *obj);
662 void addReplyBulkCString(redisClient *c, char *s);
663 void addReplyBulkCBuffer(redisClient *c, void *p, size_t len);
664 void addReplyBulkLongLong(redisClient *c, long long ll);
665 void acceptHandler(aeEventLoop *el, int fd, void *privdata, int mask);
666 void addReply(redisClient *c, robj *obj);
667 void addReplySds(redisClient *c, sds s);
668 void addReplyError(redisClient *c, char *err);
669 void addReplyStatus(redisClient *c, char *status);
670 void addReplyDouble(redisClient *c, double d);
671 void addReplyLongLong(redisClient *c, long long ll);
672 void addReplyMultiBulkLen(redisClient *c, long length);
673 void *dupClientReplyValue(void *o);
674 void getClientsMaxBuffers(unsigned long *longest_output_list,
675 unsigned long *biggest_input_buffer);
676
677 #ifdef __GNUC__
678 void addReplyErrorFormat(redisClient *c, const char *fmt, ...)
679 __attribute__((format(printf, 2, 3)));
680 void addReplyStatusFormat(redisClient *c, const char *fmt, ...)
681 __attribute__((format(printf, 2, 3)));
682 #else
683 void addReplyErrorFormat(redisClient *c, const char *fmt, ...);
684 void addReplyStatusFormat(redisClient *c, const char *fmt, ...);
685 #endif
686
687 /* List data type */
688 void listTypeTryConversion(robj *subject, robj *value);
689 void listTypePush(robj *subject, robj *value, int where);
690 robj *listTypePop(robj *subject, int where);
691 unsigned long listTypeLength(robj *subject);
692 listTypeIterator *listTypeInitIterator(robj *subject, int index, unsigned char direction);
693 void listTypeReleaseIterator(listTypeIterator *li);
694 int listTypeNext(listTypeIterator *li, listTypeEntry *entry);
695 robj *listTypeGet(listTypeEntry *entry);
696 void listTypeInsert(listTypeEntry *entry, robj *value, int where);
697 int listTypeEqual(listTypeEntry *entry, robj *o);
698 void listTypeDelete(listTypeEntry *entry);
699 void listTypeConvert(robj *subject, int enc);
700 void unblockClientWaitingData(redisClient *c);
701 int handleClientsWaitingListPush(redisClient *c, robj *key, robj *ele);
702 void popGenericCommand(redisClient *c, int where);
703
704 /* MULTI/EXEC/WATCH... */
705 void unwatchAllKeys(redisClient *c);
706 void initClientMultiState(redisClient *c);
707 void freeClientMultiState(redisClient *c);
708 void queueMultiCommand(redisClient *c, struct redisCommand *cmd);
709 void touchWatchedKey(redisDb *db, robj *key);
710 void touchWatchedKeysOnFlush(int dbid);
711
712 /* Redis object implementation */
713 void decrRefCount(void *o);
714 void incrRefCount(robj *o);
715 void freeStringObject(robj *o);
716 void freeListObject(robj *o);
717 void freeSetObject(robj *o);
718 void freeZsetObject(robj *o);
719 void freeHashObject(robj *o);
720 robj *createObject(int type, void *ptr);
721 robj *createStringObject(char *ptr, size_t len);
722 robj *dupStringObject(robj *o);
723 robj *tryObjectEncoding(robj *o);
724 robj *getDecodedObject(robj *o);
725 size_t stringObjectLen(robj *o);
726 robj *createStringObjectFromLongLong(long long value);
727 robj *createListObject(void);
728 robj *createZiplistObject(void);
729 robj *createSetObject(void);
730 robj *createIntsetObject(void);
731 robj *createHashObject(void);
732 robj *createZsetObject(void);
733 int getLongFromObjectOrReply(redisClient *c, robj *o, long *target, const char *msg);
734 int checkType(redisClient *c, robj *o, int type);
735 int getLongLongFromObjectOrReply(redisClient *c, robj *o, long long *target, const char *msg);
736 int getDoubleFromObjectOrReply(redisClient *c, robj *o, double *target, const char *msg);
737 int getLongLongFromObject(robj *o, long long *target);
738 char *strEncoding(int encoding);
739 int compareStringObjects(robj *a, robj *b);
740 int equalStringObjects(robj *a, robj *b);
741 unsigned long estimateObjectIdleTime(robj *o);
742
743 /* Synchronous I/O with timeout */
744 int syncWrite(int fd, char *ptr, ssize_t size, int timeout);
745 int syncRead(int fd, char *ptr, ssize_t size, int timeout);
746 int syncReadLine(int fd, char *ptr, ssize_t size, int timeout);
747 int fwriteBulkString(FILE *fp, char *s, unsigned long len);
748 int fwriteBulkDouble(FILE *fp, double d);
749 int fwriteBulkLongLong(FILE *fp, long long l);
750 int fwriteBulkObject(FILE *fp, robj *obj);
751
752 /* Replication */
753 void replicationFeedSlaves(list *slaves, int dictid, robj **argv, int argc);
754 void replicationFeedMonitors(list *monitors, int dictid, robj **argv, int argc);
755 int syncWithMaster(void);
756 void updateSlavesWaitingBgsave(int bgsaveerr);
757 void replicationCron(void);
758
759 /* Generic persistence functions */
760 void startLoading(FILE *fp);
761 void loadingProgress(off_t pos);
762 void stopLoading(void);
763
764 /* RDB persistence */
765 int rdbLoad(char *filename);
766 int rdbSaveBackground(char *filename);
767 void rdbRemoveTempFile(pid_t childpid);
768 int rdbSave(char *filename);
769 int rdbSaveObject(FILE *fp, robj *o);
770 off_t rdbSavedObjectLen(robj *o);
771 off_t rdbSavedObjectPages(robj *o);
772 robj *rdbLoadObject(int type, FILE *fp);
773 void backgroundSaveDoneHandler(int exitcode, int bysignal);
774 int rdbSaveKeyValuePair(FILE *fp, robj *key, robj *val, time_t expireitme, time_t now);
775 int rdbLoadType(FILE *fp);
776 time_t rdbLoadTime(FILE *fp);
777 robj *rdbLoadStringObject(FILE *fp);
778 int rdbSaveType(FILE *fp, unsigned char type);
779 int rdbSaveLen(FILE *fp, uint32_t len);
780
781 /* AOF persistence */
782 void flushAppendOnlyFile(void);
783 void feedAppendOnlyFile(struct redisCommand *cmd, int dictid, robj **argv, int argc);
784 void aofRemoveTempFile(pid_t childpid);
785 int rewriteAppendOnlyFileBackground(void);
786 int loadAppendOnlyFile(char *filename);
787 void stopAppendOnly(void);
788 int startAppendOnly(void);
789 void backgroundRewriteDoneHandler(int exitcode, int bysignal);
790
791 /* Sorted sets data type */
792 zskiplist *zslCreate(void);
793 void zslFree(zskiplist *zsl);
794 zskiplistNode *zslInsert(zskiplist *zsl, double score, robj *obj);
795
796 /* Core functions */
797 void freeMemoryIfNeeded(void);
798 int processCommand(redisClient *c);
799 void setupSigSegvAction(void);
800 struct redisCommand *lookupCommand(sds name);
801 struct redisCommand *lookupCommandByCString(char *s);
802 void call(redisClient *c, struct redisCommand *cmd);
803 int prepareForShutdown();
804 void redisLog(int level, const char *fmt, ...);
805 void usage();
806 void updateDictResizePolicy(void);
807 int htNeedsResize(dict *dict);
808 void oom(const char *msg);
809 void populateCommandTable(void);
810 void resetCommandTableStats(void);
811
812 /* Disk store */
813 int dsOpen(void);
814 int dsClose(void);
815 int dsSet(redisDb *db, robj *key, robj *val, time_t expire);
816 robj *dsGet(redisDb *db, robj *key, time_t *expire);
817 int dsDel(redisDb *db, robj *key);
818 int dsExists(redisDb *db, robj *key);
819 void dsFlushDb(int dbid);
820 int dsRdbSaveBackground(char *filename);
821 int dsRdbSave(char *filename);
822
823 /* Disk Store Cache */
824 void dsInit(void);
825 void vmThreadedIOCompletedJob(aeEventLoop *el, int fd, void *privdata, int mask);
826 void lockThreadedIO(void);
827 void unlockThreadedIO(void);
828 void freeIOJob(iojob *j);
829 void queueIOJob(iojob *j);
830 void waitEmptyIOJobsQueue(void);
831 void processAllPendingIOJobs(void);
832 void zunionInterBlockClientOnSwappedKeys(redisClient *c, struct redisCommand *cmd, int argc, robj **argv);
833 void execBlockClientOnSwappedKeys(redisClient *c, struct redisCommand *cmd, int argc, robj **argv);
834 int blockClientOnSwappedKeys(redisClient *c, struct redisCommand *cmd);
835 int dontWaitForSwappedKey(redisClient *c, robj *key);
836 void handleClientsBlockedOnSwappedKey(redisDb *db, robj *key);
837 int cacheFreeOneEntry(void);
838 void cacheScheduleIOAddFlag(redisDb *db, robj *key, long flag);
839 void cacheScheduleIODelFlag(redisDb *db, robj *key, long flag);
840 int cacheScheduleIOGetFlags(redisDb *db, robj *key);
841 void cacheScheduleIO(redisDb *db, robj *key, int type);
842 void cacheCron(void);
843 int cacheKeyMayExist(redisDb *db, robj *key);
844 void cacheSetKeyMayExist(redisDb *db, robj *key);
845 void cacheSetKeyDoesNotExist(redisDb *db, robj *key);
846 void cacheForcePointInTime(void);
847
848 /* Set data type */
849 robj *setTypeCreate(robj *value);
850 int setTypeAdd(robj *subject, robj *value);
851 int setTypeRemove(robj *subject, robj *value);
852 int setTypeIsMember(robj *subject, robj *value);
853 setTypeIterator *setTypeInitIterator(robj *subject);
854 void setTypeReleaseIterator(setTypeIterator *si);
855 int setTypeNext(setTypeIterator *si, robj **objele, int64_t *llele);
856 robj *setTypeNextObject(setTypeIterator *si);
857 int setTypeRandomElement(robj *setobj, robj **objele, int64_t *llele);
858 unsigned long setTypeSize(robj *subject);
859 void setTypeConvert(robj *subject, int enc);
860
861 /* Hash data type */
862 void convertToRealHash(robj *o);
863 void hashTypeTryConversion(robj *subject, robj **argv, int start, int end);
864 void hashTypeTryObjectEncoding(robj *subject, robj **o1, robj **o2);
865 int hashTypeGet(robj *o, robj *key, robj **objval, unsigned char **v, unsigned int *vlen);
866 robj *hashTypeGetObject(robj *o, robj *key);
867 int hashTypeExists(robj *o, robj *key);
868 int hashTypeSet(robj *o, robj *key, robj *value);
869 int hashTypeDelete(robj *o, robj *key);
870 unsigned long hashTypeLength(robj *o);
871 hashTypeIterator *hashTypeInitIterator(robj *subject);
872 void hashTypeReleaseIterator(hashTypeIterator *hi);
873 int hashTypeNext(hashTypeIterator *hi);
874 int hashTypeCurrent(hashTypeIterator *hi, int what, robj **objval, unsigned char **v, unsigned int *vlen);
875 robj *hashTypeCurrentObject(hashTypeIterator *hi, int what);
876 robj *hashTypeLookupWriteOrCreate(redisClient *c, robj *key);
877
878 /* Pub / Sub */
879 int pubsubUnsubscribeAllChannels(redisClient *c, int notify);
880 int pubsubUnsubscribeAllPatterns(redisClient *c, int notify);
881 void freePubsubPattern(void *p);
882 int listMatchPubsubPattern(void *a, void *b);
883
884 /* Utility functions */
885 int stringmatchlen(const char *pattern, int patternLen,
886 const char *string, int stringLen, int nocase);
887 int stringmatch(const char *pattern, const char *string, int nocase);
888 long long memtoll(const char *p, int *err);
889 int ll2string(char *s, size_t len, long long value);
890 int isStringRepresentableAsLong(sds s, long *longval);
891 int isStringRepresentableAsLongLong(sds s, long long *longval);
892 int isObjectRepresentableAsLongLong(robj *o, long long *llongval);
893
894 /* Configuration */
895 void loadServerConfig(char *filename);
896 void appendServerSaveParams(time_t seconds, int changes);
897 void resetServerSaveParams();
898
899 /* db.c -- Keyspace access API */
900 int removeExpire(redisDb *db, robj *key);
901 void propagateExpire(redisDb *db, robj *key);
902 int expireIfNeeded(redisDb *db, robj *key);
903 time_t getExpire(redisDb *db, robj *key);
904 void setExpire(redisDb *db, robj *key, time_t when);
905 robj *lookupKey(redisDb *db, robj *key);
906 robj *lookupKeyRead(redisDb *db, robj *key);
907 robj *lookupKeyWrite(redisDb *db, robj *key);
908 robj *lookupKeyReadOrReply(redisClient *c, robj *key, robj *reply);
909 robj *lookupKeyWriteOrReply(redisClient *c, robj *key, robj *reply);
910 int dbAdd(redisDb *db, robj *key, robj *val);
911 int dbReplace(redisDb *db, robj *key, robj *val);
912 int dbExists(redisDb *db, robj *key);
913 robj *dbRandomKey(redisDb *db);
914 int dbDelete(redisDb *db, robj *key);
915 long long emptyDb();
916 int selectDb(redisClient *c, int id);
917 void signalModifiedKey(redisDb *db, robj *key);
918 void signalFlushedDb(int dbid);
919
920 /* Git SHA1 */
921 char *redisGitSHA1(void);
922 char *redisGitDirty(void);
923
924 /* Commands prototypes */
925 void authCommand(redisClient *c);
926 void pingCommand(redisClient *c);
927 void echoCommand(redisClient *c);
928 void setCommand(redisClient *c);
929 void setnxCommand(redisClient *c);
930 void setexCommand(redisClient *c);
931 void getCommand(redisClient *c);
932 void delCommand(redisClient *c);
933 void existsCommand(redisClient *c);
934 void setbitCommand(redisClient *c);
935 void getbitCommand(redisClient *c);
936 void setrangeCommand(redisClient *c);
937 void getrangeCommand(redisClient *c);
938 void incrCommand(redisClient *c);
939 void decrCommand(redisClient *c);
940 void incrbyCommand(redisClient *c);
941 void decrbyCommand(redisClient *c);
942 void selectCommand(redisClient *c);
943 void randomkeyCommand(redisClient *c);
944 void keysCommand(redisClient *c);
945 void dbsizeCommand(redisClient *c);
946 void lastsaveCommand(redisClient *c);
947 void saveCommand(redisClient *c);
948 void bgsaveCommand(redisClient *c);
949 void bgrewriteaofCommand(redisClient *c);
950 void shutdownCommand(redisClient *c);
951 void moveCommand(redisClient *c);
952 void renameCommand(redisClient *c);
953 void renamenxCommand(redisClient *c);
954 void lpushCommand(redisClient *c);
955 void rpushCommand(redisClient *c);
956 void lpushxCommand(redisClient *c);
957 void rpushxCommand(redisClient *c);
958 void linsertCommand(redisClient *c);
959 void lpopCommand(redisClient *c);
960 void rpopCommand(redisClient *c);
961 void llenCommand(redisClient *c);
962 void lindexCommand(redisClient *c);
963 void lrangeCommand(redisClient *c);
964 void ltrimCommand(redisClient *c);
965 void typeCommand(redisClient *c);
966 void lsetCommand(redisClient *c);
967 void saddCommand(redisClient *c);
968 void sremCommand(redisClient *c);
969 void smoveCommand(redisClient *c);
970 void sismemberCommand(redisClient *c);
971 void scardCommand(redisClient *c);
972 void spopCommand(redisClient *c);
973 void srandmemberCommand(redisClient *c);
974 void sinterCommand(redisClient *c);
975 void sinterstoreCommand(redisClient *c);
976 void sunionCommand(redisClient *c);
977 void sunionstoreCommand(redisClient *c);
978 void sdiffCommand(redisClient *c);
979 void sdiffstoreCommand(redisClient *c);
980 void syncCommand(redisClient *c);
981 void flushdbCommand(redisClient *c);
982 void flushallCommand(redisClient *c);
983 void sortCommand(redisClient *c);
984 void lremCommand(redisClient *c);
985 void rpoplpushCommand(redisClient *c);
986 void infoCommand(redisClient *c);
987 void mgetCommand(redisClient *c);
988 void monitorCommand(redisClient *c);
989 void expireCommand(redisClient *c);
990 void expireatCommand(redisClient *c);
991 void getsetCommand(redisClient *c);
992 void ttlCommand(redisClient *c);
993 void persistCommand(redisClient *c);
994 void slaveofCommand(redisClient *c);
995 void debugCommand(redisClient *c);
996 void msetCommand(redisClient *c);
997 void msetnxCommand(redisClient *c);
998 void zaddCommand(redisClient *c);
999 void zincrbyCommand(redisClient *c);
1000 void zrangeCommand(redisClient *c);
1001 void zrangebyscoreCommand(redisClient *c);
1002 void zrevrangebyscoreCommand(redisClient *c);
1003 void zcountCommand(redisClient *c);
1004 void zrevrangeCommand(redisClient *c);
1005 void zcardCommand(redisClient *c);
1006 void zremCommand(redisClient *c);
1007 void zscoreCommand(redisClient *c);
1008 void zremrangebyscoreCommand(redisClient *c);
1009 void multiCommand(redisClient *c);
1010 void execCommand(redisClient *c);
1011 void discardCommand(redisClient *c);
1012 void blpopCommand(redisClient *c);
1013 void brpopCommand(redisClient *c);
1014 void brpoplpushCommand(redisClient *c);
1015 void appendCommand(redisClient *c);
1016 void strlenCommand(redisClient *c);
1017 void zrankCommand(redisClient *c);
1018 void zrevrankCommand(redisClient *c);
1019 void hsetCommand(redisClient *c);
1020 void hsetnxCommand(redisClient *c);
1021 void hgetCommand(redisClient *c);
1022 void hmsetCommand(redisClient *c);
1023 void hmgetCommand(redisClient *c);
1024 void hdelCommand(redisClient *c);
1025 void hlenCommand(redisClient *c);
1026 void zremrangebyrankCommand(redisClient *c);
1027 void zunionstoreCommand(redisClient *c);
1028 void zinterstoreCommand(redisClient *c);
1029 void hkeysCommand(redisClient *c);
1030 void hvalsCommand(redisClient *c);
1031 void hgetallCommand(redisClient *c);
1032 void hexistsCommand(redisClient *c);
1033 void configCommand(redisClient *c);
1034 void hincrbyCommand(redisClient *c);
1035 void subscribeCommand(redisClient *c);
1036 void unsubscribeCommand(redisClient *c);
1037 void psubscribeCommand(redisClient *c);
1038 void punsubscribeCommand(redisClient *c);
1039 void publishCommand(redisClient *c);
1040 void watchCommand(redisClient *c);
1041 void unwatchCommand(redisClient *c);
1042
1043 #if defined(__GNUC__)
1044 void *calloc(size_t count, size_t size) __attribute__ ((deprecated));
1045 void free(void *ptr) __attribute__ ((deprecated));
1046 void *malloc(size_t size) __attribute__ ((deprecated));
1047 void *realloc(void *ptr, size_t size) __attribute__ ((deprecated));
1048 #endif
1049
1050 #endif