]> git.saurik.com Git - redis.git/blob - src/redis.h
8495a87e56305f64b55a9821fcc73b3a9149d762
[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 /* If more then REDIS_WRITEV_THRESHOLD write packets are pending use writev */
54 #define REDIS_WRITEV_THRESHOLD 3
55 /* Max number of iovecs used for each writev call */
56 #define REDIS_WRITEV_IOVEC_COUNT 256
57
58 /* Hash table parameters */
59 #define REDIS_HT_MINFILL 10 /* Minimal hash table fill 10% */
60
61 /* Command flags:
62 * REDIS_CMD_DENYOOM:
63 * Commands marked with this flag will return an error when 'maxmemory' is
64 * set and the server is using more than 'maxmemory' bytes of memory.
65 * In short: commands with this flag are denied on low memory conditions.
66 * REDIS_CMD_FORCE_REPLICATION:
67 * Force replication even if dirty is 0. */
68 #define REDIS_CMD_DENYOOM 4
69 #define REDIS_CMD_FORCE_REPLICATION 8
70
71 /* Object types */
72 #define REDIS_STRING 0
73 #define REDIS_LIST 1
74 #define REDIS_SET 2
75 #define REDIS_ZSET 3
76 #define REDIS_HASH 4
77 #define REDIS_VMPOINTER 8
78
79 /* Objects encoding. Some kind of objects like Strings and Hashes can be
80 * internally represented in multiple ways. The 'encoding' field of the object
81 * is set to one of this fields for this object. */
82 #define REDIS_ENCODING_RAW 0 /* Raw representation */
83 #define REDIS_ENCODING_INT 1 /* Encoded as integer */
84 #define REDIS_ENCODING_HT 2 /* Encoded as hash table */
85 #define REDIS_ENCODING_ZIPMAP 3 /* Encoded as zipmap */
86 #define REDIS_ENCODING_LINKEDLIST 4 /* Encoded as regular linked list */
87 #define REDIS_ENCODING_ZIPLIST 5 /* Encoded as ziplist */
88 #define REDIS_ENCODING_INTSET 6 /* Encoded as intset */
89
90 /* Object types only used for dumping to disk */
91 #define REDIS_EXPIRETIME 253
92 #define REDIS_SELECTDB 254
93 #define REDIS_EOF 255
94
95 /* Defines related to the dump file format. To store 32 bits lengths for short
96 * keys requires a lot of space, so we check the most significant 2 bits of
97 * the first byte to interpreter the length:
98 *
99 * 00|000000 => if the two MSB are 00 the len is the 6 bits of this byte
100 * 01|000000 00000000 => 01, the len is 14 byes, 6 bits + 8 bits of next byte
101 * 10|000000 [32 bit integer] => if it's 01, a full 32 bit len will follow
102 * 11|000000 this means: specially encoded object will follow. The six bits
103 * number specify the kind of object that follows.
104 * See the REDIS_RDB_ENC_* defines.
105 *
106 * Lenghts up to 63 are stored using a single byte, most DB keys, and may
107 * values, will fit inside. */
108 #define REDIS_RDB_6BITLEN 0
109 #define REDIS_RDB_14BITLEN 1
110 #define REDIS_RDB_32BITLEN 2
111 #define REDIS_RDB_ENCVAL 3
112 #define REDIS_RDB_LENERR UINT_MAX
113
114 /* When a length of a string object stored on disk has the first two bits
115 * set, the remaining two bits specify a special encoding for the object
116 * accordingly to the following defines: */
117 #define REDIS_RDB_ENC_INT8 0 /* 8 bit signed integer */
118 #define REDIS_RDB_ENC_INT16 1 /* 16 bit signed integer */
119 #define REDIS_RDB_ENC_INT32 2 /* 32 bit signed integer */
120 #define REDIS_RDB_ENC_LZF 3 /* string compressed with FASTLZ */
121
122 /* Virtual memory object->where field. */
123 #define REDIS_VM_MEMORY 0 /* The object is on memory */
124 #define REDIS_VM_SWAPPED 1 /* The object is on disk */
125 #define REDIS_VM_SWAPPING 2 /* Redis is swapping this object on disk */
126 #define REDIS_VM_LOADING 3 /* Redis is loading this object from disk */
127
128 /* Virtual memory static configuration stuff.
129 * Check vmFindContiguousPages() to know more about this magic numbers. */
130 #define REDIS_VM_MAX_NEAR_PAGES 65536
131 #define REDIS_VM_MAX_RANDOM_JUMP 4096
132 #define REDIS_VM_MAX_THREADS 32
133 #define REDIS_THREAD_STACK_SIZE (1024*1024*4)
134 /* The following is the *percentage* of completed I/O jobs to process when the
135 * handelr is called. While Virtual Memory I/O operations are performed by
136 * threads, this operations must be processed by the main thread when completed
137 * in order to take effect. */
138 #define REDIS_MAX_COMPLETED_JOBS_PROCESSED 1
139
140 /* Client flags */
141 #define REDIS_SLAVE 1 /* This client is a slave server */
142 #define REDIS_MASTER 2 /* This client is a master server */
143 #define REDIS_MONITOR 4 /* This client is a slave monitor, see MONITOR */
144 #define REDIS_MULTI 8 /* This client is in a MULTI context */
145 #define REDIS_BLOCKED 16 /* The client is waiting in a blocking operation */
146 #define REDIS_IO_WAIT 32 /* The client is waiting for Virtual Memory I/O */
147 #define REDIS_DIRTY_CAS 64 /* Watched keys modified. EXEC will fail. */
148 #define REDIS_CLOSE_AFTER_REPLY 128 /* Close after writing entire reply. */
149
150 /* Client request types */
151 #define REDIS_REQ_INLINE 1
152 #define REDIS_REQ_MULTIBULK 2
153
154 /* Slave replication state - slave side */
155 #define REDIS_REPL_NONE 0 /* No active replication */
156 #define REDIS_REPL_CONNECT 1 /* Must connect to master */
157 #define REDIS_REPL_TRANSFER 2 /* Receiving .rdb from master */
158 #define REDIS_REPL_CONNECTED 3 /* Connected to master */
159
160 /* Slave replication state - from the point of view of master
161 * Note that in SEND_BULK and ONLINE state the slave receives new updates
162 * in its output queue. In the WAIT_BGSAVE state instead the server is waiting
163 * to start the next background saving in order to send updates to it. */
164 #define REDIS_REPL_WAIT_BGSAVE_START 3 /* master waits bgsave to start feeding it */
165 #define REDIS_REPL_WAIT_BGSAVE_END 4 /* master waits bgsave to start bulk DB transmission */
166 #define REDIS_REPL_SEND_BULK 5 /* master is sending the bulk DB */
167 #define REDIS_REPL_ONLINE 6 /* bulk DB already transmitted, receive updates */
168
169 /* List related stuff */
170 #define REDIS_HEAD 0
171 #define REDIS_TAIL 1
172
173 /* Sort operations */
174 #define REDIS_SORT_GET 0
175 #define REDIS_SORT_ASC 1
176 #define REDIS_SORT_DESC 2
177 #define REDIS_SORTKEY_MAX 1024
178
179 /* Log levels */
180 #define REDIS_DEBUG 0
181 #define REDIS_VERBOSE 1
182 #define REDIS_NOTICE 2
183 #define REDIS_WARNING 3
184
185 /* Anti-warning macro... */
186 #define REDIS_NOTUSED(V) ((void) V)
187
188 #define ZSKIPLIST_MAXLEVEL 32 /* Should be enough for 2^32 elements */
189 #define ZSKIPLIST_P 0.25 /* Skiplist P = 1/4 */
190
191 /* Append only defines */
192 #define APPENDFSYNC_NO 0
193 #define APPENDFSYNC_ALWAYS 1
194 #define APPENDFSYNC_EVERYSEC 2
195
196 /* Zip structure related defaults */
197 #define REDIS_HASH_MAX_ZIPMAP_ENTRIES 64
198 #define REDIS_HASH_MAX_ZIPMAP_VALUE 512
199 #define REDIS_LIST_MAX_ZIPLIST_ENTRIES 512
200 #define REDIS_LIST_MAX_ZIPLIST_VALUE 64
201 #define REDIS_SET_MAX_INTSET_ENTRIES 512
202
203 /* Sets operations codes */
204 #define REDIS_OP_UNION 0
205 #define REDIS_OP_DIFF 1
206 #define REDIS_OP_INTER 2
207
208 /* Redis maxmemory strategies */
209 #define REDIS_MAXMEMORY_VOLATILE_LRU 0
210 #define REDIS_MAXMEMORY_VOLATILE_TTL 1
211 #define REDIS_MAXMEMORY_VOLATILE_RANDOM 2
212 #define REDIS_MAXMEMORY_ALLKEYS_LRU 3
213 #define REDIS_MAXMEMORY_ALLKEYS_RANDOM 4
214 #define REDIS_MAXMEMORY_NO_EVICTION 5
215
216 /* We can print the stacktrace, so our assert is defined this way: */
217 #define redisAssert(_e) ((_e)?(void)0 : (_redisAssert(#_e,__FILE__,__LINE__),_exit(1)))
218 #define redisPanic(_e) _redisPanic(#_e,__FILE__,__LINE__),_exit(1)
219 void _redisAssert(char *estr, char *file, int line);
220 void _redisPanic(char *msg, char *file, int line);
221
222 /*-----------------------------------------------------------------------------
223 * Data types
224 *----------------------------------------------------------------------------*/
225
226 /* A redis object, that is a type able to hold a string / list / set */
227
228 /* The actual Redis Object */
229 #define REDIS_LRU_CLOCK_MAX ((1<<21)-1) /* Max value of obj->lru */
230 #define REDIS_LRU_CLOCK_RESOLUTION 10 /* LRU clock resolution in seconds */
231 typedef struct redisObject {
232 unsigned type:4;
233 unsigned storage:2; /* REDIS_VM_MEMORY or REDIS_VM_SWAPPING */
234 unsigned encoding:4;
235 unsigned lru:22; /* lru time (relative to server.lruclock) */
236 int refcount;
237 void *ptr;
238 /* VM fields are only allocated if VM is active, otherwise the
239 * object allocation function will just allocate
240 * sizeof(redisObjct) minus sizeof(redisObjectVM), so using
241 * Redis without VM active will not have any overhead. */
242 } robj;
243
244 /* The VM pointer structure - identifies an object in the swap file.
245 *
246 * This object is stored in place of the value
247 * object in the main key->value hash table representing a database.
248 * Note that the first fields (type, storage) are the same as the redisObject
249 * structure so that vmPointer strucuters can be accessed even when casted
250 * as redisObject structures.
251 *
252 * This is useful as we don't know if a value object is or not on disk, but we
253 * are always able to read obj->storage to check this. For vmPointer
254 * structures "type" is set to REDIS_VMPOINTER (even if without this field
255 * is still possible to check the kind of object from the value of 'storage').*/
256 typedef struct vmPointer {
257 unsigned type:4;
258 unsigned storage:2; /* REDIS_VM_SWAPPED or REDIS_VM_LOADING */
259 unsigned notused:26;
260 unsigned int vtype; /* type of the object stored in the swap file */
261 off_t page; /* the page at witch the object is stored on disk */
262 off_t usedpages; /* number of pages used on disk */
263 } vmpointer;
264
265 /* Macro used to initalize a Redis object allocated on the stack.
266 * Note that this macro is taken near the structure definition to make sure
267 * we'll update it when the structure is changed, to avoid bugs like
268 * bug #85 introduced exactly in this way. */
269 #define initStaticStringObject(_var,_ptr) do { \
270 _var.refcount = 1; \
271 _var.type = REDIS_STRING; \
272 _var.encoding = REDIS_ENCODING_RAW; \
273 _var.ptr = _ptr; \
274 _var.storage = REDIS_VM_MEMORY; \
275 } while(0);
276
277 typedef struct redisDb {
278 dict *dict; /* The keyspace for this DB */
279 dict *expires; /* Timeout of keys with a timeout set */
280 dict *blocking_keys; /* Keys with clients waiting for data (BLPOP) */
281 dict *io_keys; /* Keys with clients waiting for VM I/O */
282 dict *watched_keys; /* WATCHED keys for MULTI/EXEC CAS */
283 int id;
284 } redisDb;
285
286 /* Client MULTI/EXEC state */
287 typedef struct multiCmd {
288 robj **argv;
289 int argc;
290 struct redisCommand *cmd;
291 } multiCmd;
292
293 typedef struct multiState {
294 multiCmd *commands; /* Array of MULTI commands */
295 int count; /* Total number of MULTI commands */
296 } multiState;
297
298 typedef struct blockingState {
299 robj **keys; /* The key we are waiting to terminate a blocking
300 * operation such as BLPOP. Otherwise NULL. */
301 int count; /* Number of blocking keys */
302 time_t timeout; /* Blocking operation timeout. If UNIX current time
303 * is >= timeout then the operation timed out. */
304 robj *target; /* The key that should receive the element,
305 * for BRPOPLPUSH. */
306 } blockingState;
307
308 /* With multiplexing we need to take per-clinet state.
309 * Clients are taken in a liked list. */
310 typedef struct redisClient {
311 int fd;
312 redisDb *db;
313 int dictid;
314 sds querybuf;
315 int argc;
316 robj **argv;
317 int reqtype;
318 int multibulklen; /* number of multi bulk arguments left to read */
319 long bulklen; /* length of bulk argument in multi bulk request */
320 list *reply;
321 int sentlen;
322 time_t lastinteraction; /* time of the last interaction, used for timeout */
323 int flags; /* REDIS_SLAVE | REDIS_MONITOR | REDIS_MULTI ... */
324 int slaveseldb; /* slave selected db, if this client is a slave */
325 int authenticated; /* when requirepass is non-NULL */
326 int replstate; /* replication state if this is a slave */
327 int repldbfd; /* replication DB file descriptor */
328 long repldboff; /* replication DB file offset */
329 off_t repldbsize; /* replication DB file size */
330 multiState mstate; /* MULTI/EXEC state */
331 blockingState bpop; /* blocking state */
332 list *io_keys; /* Keys this client is waiting to be loaded from the
333 * swap file in order to continue. */
334 list *watched_keys; /* Keys WATCHED for MULTI/EXEC CAS */
335 dict *pubsub_channels; /* channels a client is interested in (SUBSCRIBE) */
336 list *pubsub_patterns; /* patterns a client is interested in (SUBSCRIBE) */
337
338 /* Response buffer */
339 int bufpos;
340 char buf[REDIS_REPLY_CHUNK_BYTES];
341 } redisClient;
342
343 struct saveparam {
344 time_t seconds;
345 int changes;
346 };
347
348 struct sharedObjectsStruct {
349 robj *crlf, *ok, *err, *emptybulk, *czero, *cone, *cnegone, *pong, *space,
350 *colon, *nullbulk, *nullmultibulk, *queued,
351 *emptymultibulk, *wrongtypeerr, *nokeyerr, *syntaxerr, *sameobjecterr,
352 *outofrangeerr, *loadingerr, *plus,
353 *select0, *select1, *select2, *select3, *select4,
354 *select5, *select6, *select7, *select8, *select9,
355 *messagebulk, *pmessagebulk, *subscribebulk, *unsubscribebulk, *mbulk3,
356 *mbulk4, *psubscribebulk, *punsubscribebulk,
357 *integers[REDIS_SHARED_INTEGERS];
358 };
359
360 /* Global server state structure */
361 struct redisServer {
362 pthread_t mainthread;
363 int port;
364 char *bindaddr;
365 char *unixsocket;
366 int ipfd;
367 int sofd;
368 redisDb *db;
369 long long dirty; /* changes to DB from the last save */
370 long long dirty_before_bgsave; /* used to restore dirty on failed BGSAVE */
371 list *clients;
372 dict *commands; /* Command table hahs table */
373 /* RDB / AOF loading information */
374 int loading;
375 off_t loading_total_bytes;
376 off_t loading_loaded_bytes;
377 time_t loading_start_time;
378 /* Fast pointers to often looked up command */
379 struct redisCommand *delCommand, *multiCommand;
380 list *slaves, *monitors;
381 char neterr[ANET_ERR_LEN];
382 aeEventLoop *el;
383 int cronloops; /* number of times the cron function run */
384 time_t lastsave; /* Unix time of last save succeeede */
385 /* Fields used only for stats */
386 time_t stat_starttime; /* server start time */
387 long long stat_numcommands; /* number of processed commands */
388 long long stat_numconnections; /* number of connections received */
389 long long stat_expiredkeys; /* number of expired keys */
390 long long stat_keyspace_hits; /* number of successful lookups of keys */
391 long long stat_keyspace_misses; /* number of failed lookups of keys */
392 /* Configuration */
393 int verbosity;
394 int glueoutputbuf;
395 int maxidletime;
396 int dbnum;
397 int daemonize;
398 int appendonly;
399 int appendfsync;
400 int no_appendfsync_on_rewrite;
401 int shutdown_asap;
402 time_t lastfsync;
403 int appendfd;
404 int appendseldb;
405 char *pidfile;
406 pid_t bgsavechildpid;
407 pid_t bgrewritechildpid;
408 sds bgrewritebuf; /* buffer taken by parent during oppend only rewrite */
409 sds aofbuf; /* AOF buffer, written before entering the event loop */
410 struct saveparam *saveparams;
411 int saveparamslen;
412 char *logfile;
413 int syslog_enabled;
414 char *syslog_ident;
415 int syslog_facility;
416 char *dbfilename;
417 char *appendfilename;
418 char *requirepass;
419 int rdbcompression;
420 int activerehashing;
421 /* Replication related */
422 int isslave;
423 /* Slave specific fields */
424 char *masterauth;
425 char *masterhost;
426 int masterport;
427 redisClient *master; /* client that is master for this slave */
428 int replstate; /* replication status if the instance is a slave */
429 off_t repl_transfer_left; /* bytes left reading .rdb */
430 int repl_transfer_s; /* slave -> master SYNC socket */
431 int repl_transfer_fd; /* slave -> master SYNC temp file descriptor */
432 char *repl_transfer_tmpfile; /* slave-> master SYNC temp file name */
433 time_t repl_transfer_lastio; /* unix time of the latest read, for timeout */
434 int repl_serve_stale_data; /* Serve stale data when link is down? */
435 /* Limits */
436 unsigned int maxclients;
437 unsigned long long maxmemory;
438 int maxmemory_policy;
439 int maxmemory_samples;
440 /* Blocked clients */
441 unsigned int bpop_blocked_clients;
442 unsigned int vm_blocked_clients;
443 list *unblocked_clients;
444 /* Sort parameters - qsort_r() is only available under BSD so we
445 * have to take this state global, in order to pass it to sortCompare() */
446 int sort_desc;
447 int sort_alpha;
448 int sort_bypattern;
449 /* Virtual memory configuration */
450 int vm_enabled;
451 char *vm_swap_file;
452 off_t vm_page_size;
453 off_t vm_pages;
454 unsigned long long vm_max_memory;
455 /* Zip structure config */
456 size_t hash_max_zipmap_entries;
457 size_t hash_max_zipmap_value;
458 size_t list_max_ziplist_entries;
459 size_t list_max_ziplist_value;
460 size_t set_max_intset_entries;
461 /* Virtual memory state */
462 FILE *vm_fp;
463 int vm_fd;
464 off_t vm_next_page; /* Next probably empty page */
465 off_t vm_near_pages; /* Number of pages allocated sequentially */
466 unsigned char *vm_bitmap; /* Bitmap of free/used pages */
467 time_t unixtime; /* Unix time sampled every second. */
468 /* Virtual memory I/O threads stuff */
469 /* An I/O thread process an element taken from the io_jobs queue and
470 * put the result of the operation in the io_done list. While the
471 * job is being processed, it's put on io_processing queue. */
472 list *io_newjobs; /* List of VM I/O jobs yet to be processed */
473 list *io_processing; /* List of VM I/O jobs being processed */
474 list *io_processed; /* List of VM I/O jobs already processed */
475 list *io_ready_clients; /* Clients ready to be unblocked. All keys loaded */
476 pthread_mutex_t io_mutex; /* lock to access io_jobs/io_done/io_thread_job */
477 pthread_mutex_t io_swapfile_mutex; /* So we can lseek + write */
478 pthread_attr_t io_threads_attr; /* attributes for threads creation */
479 int io_active_threads; /* Number of running I/O threads */
480 int vm_max_threads; /* Max number of I/O threads running at the same time */
481 /* Our main thread is blocked on the event loop, locking for sockets ready
482 * to be read or written, so when a threaded I/O operation is ready to be
483 * processed by the main thread, the I/O thread will use a unix pipe to
484 * awake the main thread. The followings are the two pipe FDs. */
485 int io_ready_pipe_read;
486 int io_ready_pipe_write;
487 /* Virtual memory stats */
488 unsigned long long vm_stats_used_pages;
489 unsigned long long vm_stats_swapped_objects;
490 unsigned long long vm_stats_swapouts;
491 unsigned long long vm_stats_swapins;
492 /* Pubsub */
493 dict *pubsub_channels; /* Map channels to list of subscribed clients */
494 list *pubsub_patterns; /* A list of pubsub_patterns */
495 /* Misc */
496 unsigned lruclock:22; /* clock incrementing every minute, for LRU */
497 unsigned lruclock_padding:10;
498 };
499
500 typedef struct pubsubPattern {
501 redisClient *client;
502 robj *pattern;
503 } pubsubPattern;
504
505 typedef void redisCommandProc(redisClient *c);
506 typedef void redisVmPreloadProc(redisClient *c, struct redisCommand *cmd, int argc, robj **argv);
507 struct redisCommand {
508 char *name;
509 redisCommandProc *proc;
510 int arity;
511 int flags;
512 /* Use a function to determine which keys need to be loaded
513 * in the background prior to executing this command. Takes precedence
514 * over vm_firstkey and others, ignored when NULL */
515 redisVmPreloadProc *vm_preload_proc;
516 /* What keys should be loaded in background when calling this command? */
517 int vm_firstkey; /* The first argument that's a key (0 = no keys) */
518 int vm_lastkey; /* THe last argument that's a key */
519 int vm_keystep; /* The step between first and last key */
520 };
521
522 struct redisFunctionSym {
523 char *name;
524 unsigned long pointer;
525 };
526
527 typedef struct _redisSortObject {
528 robj *obj;
529 union {
530 double score;
531 robj *cmpobj;
532 } u;
533 } redisSortObject;
534
535 typedef struct _redisSortOperation {
536 int type;
537 robj *pattern;
538 } redisSortOperation;
539
540 /* ZSETs use a specialized version of Skiplists */
541 typedef struct zskiplistNode {
542 robj *obj;
543 double score;
544 struct zskiplistNode *backward;
545 struct zskiplistLevel {
546 struct zskiplistNode *forward;
547 unsigned int span;
548 } level[];
549 } zskiplistNode;
550
551 typedef struct zskiplist {
552 struct zskiplistNode *header, *tail;
553 unsigned long length;
554 int level;
555 } zskiplist;
556
557 typedef struct zset {
558 dict *dict;
559 zskiplist *zsl;
560 } zset;
561
562 /* VM threaded I/O request message */
563 #define REDIS_IOJOB_LOAD 0 /* Load from disk to memory */
564 #define REDIS_IOJOB_PREPARE_SWAP 1 /* Compute needed pages */
565 #define REDIS_IOJOB_DO_SWAP 2 /* Swap from memory to disk */
566 typedef struct iojob {
567 int type; /* Request type, REDIS_IOJOB_* */
568 redisDb *db;/* Redis database */
569 robj *key; /* This I/O request is about swapping this key */
570 robj *id; /* Unique identifier of this job:
571 this is the object to swap for REDIS_IOREQ_*_SWAP, or the
572 vmpointer objct for REDIS_IOREQ_LOAD. */
573 robj *val; /* the value to swap for REDIS_IOREQ_*_SWAP, otherwise this
574 * field is populated by the I/O thread for REDIS_IOREQ_LOAD. */
575 off_t page; /* Swap page where to read/write the object */
576 off_t pages; /* Swap pages needed to save object. PREPARE_SWAP return val */
577 int canceled; /* True if this command was canceled by blocking side of VM */
578 pthread_t thread; /* ID of the thread processing this entry */
579 } iojob;
580
581 /* Structure to hold list iteration abstraction. */
582 typedef struct {
583 robj *subject;
584 unsigned char encoding;
585 unsigned char direction; /* Iteration direction */
586 unsigned char *zi;
587 listNode *ln;
588 } listTypeIterator;
589
590 /* Structure for an entry while iterating over a list. */
591 typedef struct {
592 listTypeIterator *li;
593 unsigned char *zi; /* Entry in ziplist */
594 listNode *ln; /* Entry in linked list */
595 } listTypeEntry;
596
597 /* Structure to hold set iteration abstraction. */
598 typedef struct {
599 robj *subject;
600 int encoding;
601 int ii; /* intset iterator */
602 dictIterator *di;
603 } setTypeIterator;
604
605 /* Structure to hold hash iteration abstration. Note that iteration over
606 * hashes involves both fields and values. Because it is possible that
607 * not both are required, store pointers in the iterator to avoid
608 * unnecessary memory allocation for fields/values. */
609 typedef struct {
610 int encoding;
611 unsigned char *zi;
612 unsigned char *zk, *zv;
613 unsigned int zklen, zvlen;
614
615 dictIterator *di;
616 dictEntry *de;
617 } hashTypeIterator;
618
619 #define REDIS_HASH_KEY 1
620 #define REDIS_HASH_VALUE 2
621
622 /*-----------------------------------------------------------------------------
623 * Extern declarations
624 *----------------------------------------------------------------------------*/
625
626 extern struct redisServer server;
627 extern struct sharedObjectsStruct shared;
628 extern dictType setDictType;
629 extern dictType zsetDictType;
630 extern double R_Zero, R_PosInf, R_NegInf, R_Nan;
631 dictType hashDictType;
632
633 /*-----------------------------------------------------------------------------
634 * Functions prototypes
635 *----------------------------------------------------------------------------*/
636
637 /* networking.c -- Networking and Client related operations */
638 redisClient *createClient(int fd);
639 void closeTimedoutClients(void);
640 void freeClient(redisClient *c);
641 void resetClient(redisClient *c);
642 void sendReplyToClient(aeEventLoop *el, int fd, void *privdata, int mask);
643 void sendReplyToClientWritev(aeEventLoop *el, int fd, void *privdata, int mask);
644 void addReply(redisClient *c, robj *obj);
645 void *addDeferredMultiBulkLength(redisClient *c);
646 void setDeferredMultiBulkLength(redisClient *c, void *node, long length);
647 void addReplySds(redisClient *c, sds s);
648 void processInputBuffer(redisClient *c);
649 void acceptTcpHandler(aeEventLoop *el, int fd, void *privdata, int mask);
650 void acceptUnixHandler(aeEventLoop *el, int fd, void *privdata, int mask);
651 void readQueryFromClient(aeEventLoop *el, int fd, void *privdata, int mask);
652 void addReplyBulk(redisClient *c, robj *obj);
653 void addReplyBulkCString(redisClient *c, char *s);
654 void addReplyBulkCBuffer(redisClient *c, void *p, size_t len);
655 void addReplyBulkLongLong(redisClient *c, long long ll);
656 void acceptHandler(aeEventLoop *el, int fd, void *privdata, int mask);
657 void addReply(redisClient *c, robj *obj);
658 void addReplySds(redisClient *c, sds s);
659 void addReplyError(redisClient *c, char *err);
660 void addReplyStatus(redisClient *c, char *status);
661 void addReplyDouble(redisClient *c, double d);
662 void addReplyLongLong(redisClient *c, long long ll);
663 void addReplyMultiBulkLen(redisClient *c, long length);
664 void *dupClientReplyValue(void *o);
665
666 #ifdef __GNUC__
667 void addReplyErrorFormat(redisClient *c, const char *fmt, ...)
668 __attribute__((format(printf, 2, 3)));
669 void addReplyStatusFormat(redisClient *c, const char *fmt, ...)
670 __attribute__((format(printf, 2, 3)));
671 #else
672 void addReplyErrorFormat(redisClient *c, const char *fmt, ...);
673 void addReplyStatusFormat(redisClient *c, const char *fmt, ...);
674 #endif
675
676 /* List data type */
677 void listTypeTryConversion(robj *subject, robj *value);
678 void listTypePush(robj *subject, robj *value, int where);
679 robj *listTypePop(robj *subject, int where);
680 unsigned long listTypeLength(robj *subject);
681 listTypeIterator *listTypeInitIterator(robj *subject, int index, unsigned char direction);
682 void listTypeReleaseIterator(listTypeIterator *li);
683 int listTypeNext(listTypeIterator *li, listTypeEntry *entry);
684 robj *listTypeGet(listTypeEntry *entry);
685 void listTypeInsert(listTypeEntry *entry, robj *value, int where);
686 int listTypeEqual(listTypeEntry *entry, robj *o);
687 void listTypeDelete(listTypeEntry *entry);
688 void listTypeConvert(robj *subject, int enc);
689 void unblockClientWaitingData(redisClient *c);
690 int handleClientsWaitingListPush(redisClient *c, robj *key, robj *ele);
691 void popGenericCommand(redisClient *c, int where);
692
693 /* MULTI/EXEC/WATCH... */
694 void unwatchAllKeys(redisClient *c);
695 void initClientMultiState(redisClient *c);
696 void freeClientMultiState(redisClient *c);
697 void queueMultiCommand(redisClient *c, struct redisCommand *cmd);
698 void touchWatchedKey(redisDb *db, robj *key);
699 void touchWatchedKeysOnFlush(int dbid);
700
701 /* Redis object implementation */
702 void decrRefCount(void *o);
703 void incrRefCount(robj *o);
704 void freeStringObject(robj *o);
705 void freeListObject(robj *o);
706 void freeSetObject(robj *o);
707 void freeZsetObject(robj *o);
708 void freeHashObject(robj *o);
709 robj *createObject(int type, void *ptr);
710 robj *createStringObject(char *ptr, size_t len);
711 robj *dupStringObject(robj *o);
712 robj *tryObjectEncoding(robj *o);
713 robj *getDecodedObject(robj *o);
714 size_t stringObjectLen(robj *o);
715 robj *createStringObjectFromLongLong(long long value);
716 robj *createListObject(void);
717 robj *createZiplistObject(void);
718 robj *createSetObject(void);
719 robj *createIntsetObject(void);
720 robj *createHashObject(void);
721 robj *createZsetObject(void);
722 int getLongFromObjectOrReply(redisClient *c, robj *o, long *target, const char *msg);
723 int checkType(redisClient *c, robj *o, int type);
724 int getLongLongFromObjectOrReply(redisClient *c, robj *o, long long *target, const char *msg);
725 int getDoubleFromObjectOrReply(redisClient *c, robj *o, double *target, const char *msg);
726 int getLongLongFromObject(robj *o, long long *target);
727 char *strEncoding(int encoding);
728 int compareStringObjects(robj *a, robj *b);
729 int equalStringObjects(robj *a, robj *b);
730 unsigned long estimateObjectIdleTime(robj *o);
731
732 /* Synchronous I/O with timeout */
733 int syncWrite(int fd, char *ptr, ssize_t size, int timeout);
734 int syncRead(int fd, char *ptr, ssize_t size, int timeout);
735 int syncReadLine(int fd, char *ptr, ssize_t size, int timeout);
736 int fwriteBulkString(FILE *fp, char *s, unsigned long len);
737 int fwriteBulkDouble(FILE *fp, double d);
738 int fwriteBulkLongLong(FILE *fp, long long l);
739 int fwriteBulkObject(FILE *fp, robj *obj);
740
741 /* Replication */
742 void replicationFeedSlaves(list *slaves, int dictid, robj **argv, int argc);
743 void replicationFeedMonitors(list *monitors, int dictid, robj **argv, int argc);
744 int syncWithMaster(void);
745 void updateSlavesWaitingBgsave(int bgsaveerr);
746 void replicationCron(void);
747
748 /* Generic persistence functions */
749 void startLoading(FILE *fp);
750 void loadingProgress(off_t pos);
751 void stopLoading(void);
752
753 /* RDB persistence */
754 int rdbLoad(char *filename);
755 int rdbSaveBackground(char *filename);
756 void rdbRemoveTempFile(pid_t childpid);
757 int rdbSave(char *filename);
758 int rdbSaveObject(FILE *fp, robj *o);
759 off_t rdbSavedObjectLen(robj *o);
760 off_t rdbSavedObjectPages(robj *o);
761 robj *rdbLoadObject(int type, FILE *fp);
762 void backgroundSaveDoneHandler(int statloc);
763
764 /* AOF persistence */
765 void flushAppendOnlyFile(void);
766 void feedAppendOnlyFile(struct redisCommand *cmd, int dictid, robj **argv, int argc);
767 void aofRemoveTempFile(pid_t childpid);
768 int rewriteAppendOnlyFileBackground(void);
769 int loadAppendOnlyFile(char *filename);
770 void stopAppendOnly(void);
771 int startAppendOnly(void);
772 void backgroundRewriteDoneHandler(int statloc);
773
774 /* Sorted sets data type */
775 zskiplist *zslCreate(void);
776 void zslFree(zskiplist *zsl);
777 zskiplistNode *zslInsert(zskiplist *zsl, double score, robj *obj);
778
779 /* Core functions */
780 void freeMemoryIfNeeded(void);
781 int processCommand(redisClient *c);
782 void setupSigSegvAction(void);
783 struct redisCommand *lookupCommand(sds name);
784 struct redisCommand *lookupCommandByCString(char *s);
785 void call(redisClient *c, struct redisCommand *cmd);
786 int prepareForShutdown();
787 void redisLog(int level, const char *fmt, ...);
788 void usage();
789 void updateDictResizePolicy(void);
790 int htNeedsResize(dict *dict);
791 void oom(const char *msg);
792 void populateCommandTable(void);
793
794 /* Virtual Memory */
795 void vmInit(void);
796 void vmMarkPagesFree(off_t page, off_t count);
797 robj *vmLoadObject(robj *o);
798 robj *vmPreviewObject(robj *o);
799 int vmSwapOneObjectBlocking(void);
800 int vmSwapOneObjectThreaded(void);
801 int vmCanSwapOut(void);
802 void vmThreadedIOCompletedJob(aeEventLoop *el, int fd, void *privdata, int mask);
803 void vmCancelThreadedIOJob(robj *o);
804 void lockThreadedIO(void);
805 void unlockThreadedIO(void);
806 int vmSwapObjectThreaded(robj *key, robj *val, redisDb *db);
807 void freeIOJob(iojob *j);
808 void queueIOJob(iojob *j);
809 int vmWriteObjectOnSwap(robj *o, off_t page);
810 robj *vmReadObjectFromSwap(off_t page, int type);
811 void waitEmptyIOJobsQueue(void);
812 void vmReopenSwapFile(void);
813 int vmFreePage(off_t page);
814 void zunionInterBlockClientOnSwappedKeys(redisClient *c, struct redisCommand *cmd, int argc, robj **argv);
815 void execBlockClientOnSwappedKeys(redisClient *c, struct redisCommand *cmd, int argc, robj **argv);
816 int blockClientOnSwappedKeys(redisClient *c, struct redisCommand *cmd);
817 int dontWaitForSwappedKey(redisClient *c, robj *key);
818 void handleClientsBlockedOnSwappedKey(redisDb *db, robj *key);
819 vmpointer *vmSwapObjectBlocking(robj *val);
820
821 /* Set data type */
822 robj *setTypeCreate(robj *value);
823 int setTypeAdd(robj *subject, robj *value);
824 int setTypeRemove(robj *subject, robj *value);
825 int setTypeIsMember(robj *subject, robj *value);
826 setTypeIterator *setTypeInitIterator(robj *subject);
827 void setTypeReleaseIterator(setTypeIterator *si);
828 int setTypeNext(setTypeIterator *si, robj **objele, int64_t *llele);
829 robj *setTypeNextObject(setTypeIterator *si);
830 int setTypeRandomElement(robj *setobj, robj **objele, int64_t *llele);
831 unsigned long setTypeSize(robj *subject);
832 void setTypeConvert(robj *subject, int enc);
833
834 /* Hash data type */
835 void convertToRealHash(robj *o);
836 void hashTypeTryConversion(robj *subject, robj **argv, int start, int end);
837 void hashTypeTryObjectEncoding(robj *subject, robj **o1, robj **o2);
838 int hashTypeGet(robj *o, robj *key, robj **objval, unsigned char **v, unsigned int *vlen);
839 robj *hashTypeGetObject(robj *o, robj *key);
840 int hashTypeExists(robj *o, robj *key);
841 int hashTypeSet(robj *o, robj *key, robj *value);
842 int hashTypeDelete(robj *o, robj *key);
843 unsigned long hashTypeLength(robj *o);
844 hashTypeIterator *hashTypeInitIterator(robj *subject);
845 void hashTypeReleaseIterator(hashTypeIterator *hi);
846 int hashTypeNext(hashTypeIterator *hi);
847 int hashTypeCurrent(hashTypeIterator *hi, int what, robj **objval, unsigned char **v, unsigned int *vlen);
848 robj *hashTypeCurrentObject(hashTypeIterator *hi, int what);
849 robj *hashTypeLookupWriteOrCreate(redisClient *c, robj *key);
850
851 /* Pub / Sub */
852 int pubsubUnsubscribeAllChannels(redisClient *c, int notify);
853 int pubsubUnsubscribeAllPatterns(redisClient *c, int notify);
854 void freePubsubPattern(void *p);
855 int listMatchPubsubPattern(void *a, void *b);
856
857 /* Utility functions */
858 int stringmatchlen(const char *pattern, int patternLen,
859 const char *string, int stringLen, int nocase);
860 int stringmatch(const char *pattern, const char *string, int nocase);
861 long long memtoll(const char *p, int *err);
862 int ll2string(char *s, size_t len, long long value);
863 int isStringRepresentableAsLong(sds s, long *longval);
864 int isStringRepresentableAsLongLong(sds s, long long *longval);
865 int isObjectRepresentableAsLongLong(robj *o, long long *llongval);
866
867 /* Configuration */
868 void loadServerConfig(char *filename);
869 void appendServerSaveParams(time_t seconds, int changes);
870 void resetServerSaveParams();
871
872 /* db.c -- Keyspace access API */
873 int removeExpire(redisDb *db, robj *key);
874 void propagateExpire(redisDb *db, robj *key);
875 int expireIfNeeded(redisDb *db, robj *key);
876 time_t getExpire(redisDb *db, robj *key);
877 void setExpire(redisDb *db, robj *key, time_t when);
878 robj *lookupKey(redisDb *db, robj *key);
879 robj *lookupKeyRead(redisDb *db, robj *key);
880 robj *lookupKeyWrite(redisDb *db, robj *key);
881 robj *lookupKeyReadOrReply(redisClient *c, robj *key, robj *reply);
882 robj *lookupKeyWriteOrReply(redisClient *c, robj *key, robj *reply);
883 int dbAdd(redisDb *db, robj *key, robj *val);
884 int dbReplace(redisDb *db, robj *key, robj *val);
885 int dbExists(redisDb *db, robj *key);
886 robj *dbRandomKey(redisDb *db);
887 int dbDelete(redisDb *db, robj *key);
888 long long emptyDb();
889 int selectDb(redisClient *c, int id);
890
891 /* Git SHA1 */
892 char *redisGitSHA1(void);
893 char *redisGitDirty(void);
894
895 /* Commands prototypes */
896 void authCommand(redisClient *c);
897 void pingCommand(redisClient *c);
898 void echoCommand(redisClient *c);
899 void setCommand(redisClient *c);
900 void setnxCommand(redisClient *c);
901 void setexCommand(redisClient *c);
902 void getCommand(redisClient *c);
903 void delCommand(redisClient *c);
904 void existsCommand(redisClient *c);
905 void setbitCommand(redisClient *c);
906 void getbitCommand(redisClient *c);
907 void setrangeCommand(redisClient *c);
908 void getrangeCommand(redisClient *c);
909 void incrCommand(redisClient *c);
910 void decrCommand(redisClient *c);
911 void incrbyCommand(redisClient *c);
912 void decrbyCommand(redisClient *c);
913 void selectCommand(redisClient *c);
914 void randomkeyCommand(redisClient *c);
915 void keysCommand(redisClient *c);
916 void dbsizeCommand(redisClient *c);
917 void lastsaveCommand(redisClient *c);
918 void saveCommand(redisClient *c);
919 void bgsaveCommand(redisClient *c);
920 void bgrewriteaofCommand(redisClient *c);
921 void shutdownCommand(redisClient *c);
922 void moveCommand(redisClient *c);
923 void renameCommand(redisClient *c);
924 void renamenxCommand(redisClient *c);
925 void lpushCommand(redisClient *c);
926 void rpushCommand(redisClient *c);
927 void lpushxCommand(redisClient *c);
928 void rpushxCommand(redisClient *c);
929 void linsertCommand(redisClient *c);
930 void lpopCommand(redisClient *c);
931 void rpopCommand(redisClient *c);
932 void llenCommand(redisClient *c);
933 void lindexCommand(redisClient *c);
934 void lrangeCommand(redisClient *c);
935 void ltrimCommand(redisClient *c);
936 void typeCommand(redisClient *c);
937 void lsetCommand(redisClient *c);
938 void saddCommand(redisClient *c);
939 void sremCommand(redisClient *c);
940 void smoveCommand(redisClient *c);
941 void sismemberCommand(redisClient *c);
942 void scardCommand(redisClient *c);
943 void spopCommand(redisClient *c);
944 void srandmemberCommand(redisClient *c);
945 void sinterCommand(redisClient *c);
946 void sinterstoreCommand(redisClient *c);
947 void sunionCommand(redisClient *c);
948 void sunionstoreCommand(redisClient *c);
949 void sdiffCommand(redisClient *c);
950 void sdiffstoreCommand(redisClient *c);
951 void syncCommand(redisClient *c);
952 void flushdbCommand(redisClient *c);
953 void flushallCommand(redisClient *c);
954 void sortCommand(redisClient *c);
955 void lremCommand(redisClient *c);
956 void rpoplpushCommand(redisClient *c);
957 void infoCommand(redisClient *c);
958 void mgetCommand(redisClient *c);
959 void monitorCommand(redisClient *c);
960 void expireCommand(redisClient *c);
961 void expireatCommand(redisClient *c);
962 void getsetCommand(redisClient *c);
963 void ttlCommand(redisClient *c);
964 void persistCommand(redisClient *c);
965 void slaveofCommand(redisClient *c);
966 void debugCommand(redisClient *c);
967 void msetCommand(redisClient *c);
968 void msetnxCommand(redisClient *c);
969 void zaddCommand(redisClient *c);
970 void zincrbyCommand(redisClient *c);
971 void zrangeCommand(redisClient *c);
972 void zrangebyscoreCommand(redisClient *c);
973 void zrevrangebyscoreCommand(redisClient *c);
974 void zcountCommand(redisClient *c);
975 void zrevrangeCommand(redisClient *c);
976 void zcardCommand(redisClient *c);
977 void zremCommand(redisClient *c);
978 void zscoreCommand(redisClient *c);
979 void zremrangebyscoreCommand(redisClient *c);
980 void multiCommand(redisClient *c);
981 void execCommand(redisClient *c);
982 void discardCommand(redisClient *c);
983 void blpopCommand(redisClient *c);
984 void brpopCommand(redisClient *c);
985 void brpoplpushCommand(redisClient *c);
986 void appendCommand(redisClient *c);
987 void strlenCommand(redisClient *c);
988 void zrankCommand(redisClient *c);
989 void zrevrankCommand(redisClient *c);
990 void hsetCommand(redisClient *c);
991 void hsetnxCommand(redisClient *c);
992 void hgetCommand(redisClient *c);
993 void hmsetCommand(redisClient *c);
994 void hmgetCommand(redisClient *c);
995 void hdelCommand(redisClient *c);
996 void hlenCommand(redisClient *c);
997 void zremrangebyrankCommand(redisClient *c);
998 void zunionstoreCommand(redisClient *c);
999 void zinterstoreCommand(redisClient *c);
1000 void hkeysCommand(redisClient *c);
1001 void hvalsCommand(redisClient *c);
1002 void hgetallCommand(redisClient *c);
1003 void hexistsCommand(redisClient *c);
1004 void configCommand(redisClient *c);
1005 void hincrbyCommand(redisClient *c);
1006 void subscribeCommand(redisClient *c);
1007 void unsubscribeCommand(redisClient *c);
1008 void psubscribeCommand(redisClient *c);
1009 void punsubscribeCommand(redisClient *c);
1010 void publishCommand(redisClient *c);
1011 void watchCommand(redisClient *c);
1012 void unwatchCommand(redisClient *c);
1013
1014 #if defined(__GNUC__)
1015 void *calloc(size_t count, size_t size) __attribute__ ((deprecated));
1016 void free(void *ptr) __attribute__ ((deprecated));
1017 void *malloc(size_t size) __attribute__ ((deprecated));
1018 void *realloc(void *ptr, size_t size) __attribute__ ((deprecated));
1019 #endif
1020
1021 #endif