Redis/redis
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Hash Field Expiration (#13303)

Browse Source 
## Background

This PR introduces support for field-level expiration in Redis hashes. Previously, Redis supported expiration only at the key level, but this enhancement allows setting expiration times for individual fields within a hash.

## New commands
* HEXPIRE
* HEXPIREAT
* HEXPIRETIME
* HPERSIST
* HPEXPIRE
* HPEXPIREAT
* HPEXPIRETIME
* HPTTL
* HTTL

## Short example
from @moticless
```sh
127.0.0.1:6379>  hset myhash f1 v1 f2 v2 f3 v3                                                   
(integer) 3
127.0.0.1:6379>  hpexpire myhash 10000 NX fields 2 f2 f3                                         
1) (integer) 1
2) (integer) 1
127.0.0.1:6379>  hpttl myhash fields 3 f1 f2 f3                                                                                                                                                                         
1) (integer) -1
2) (integer) 9997
3) (integer) 9997
127.0.0.1:6379>  hgetall myhash  
1) "f3"
2) "v3"
3) "f2"
4) "v2"
5) "f1"
6) "v1"

... after 10 seconds ...

127.0.0.1:6379>  hgetall myhash  
1) "f1"
2) "v1"
127.0.0.1:6379>
```

## Expiration strategy
1. Integrate active
    Redis periodically performs active expiration and deletion of hash keys that contain expired fields, with a maximum attempt limit.
3. Lazy expiration
    When a client touches fields within a hash, Redis checks if the fields are expired. If a field is expired, it will be deleted. However, we do not delete expired fields during a traversal, we implicitly skip over them.

## RDB changes
Add two new rdb type s`RDB_TYPE_HASH_METADATA` and `RDB_TYPE_HASH_LISTPACK_EX`.

## Notification
1. Add `hpersist` notification for `HPERSIST` command.
5. Add `hexpire` notification for `HEXPIRE`, `HEXPIREAT`, `HPEXPIRE` and `HPEXPIREAT` commands.

## Internal
1. Add new data structure `ebuckets`, which is used to store TTL and keys, enabling quick retrieval of keys based on TTL.
2. Add new data structure `mstr` like sds, which is used to store a string with TTL.

This work was done by @moticless, @tezc, @ronen-kalish, @sundb, I just release it.
This commit is contained in:
debing.sun
2024-05-30 15:26:19 +08:00
committed by GitHub
parent 2d1bb42cba f0389f2823
commit 7b9e960690
53 changed files with 10243 additions and 573 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/Makefile
+1 -1
View File
@@ -354,7 +354,7 @@ endif
REDIS_SERVER_NAME=redis-server$(PROG_SUFFIX)
REDIS_SENTINEL_NAME=redis-sentinel$(PROG_SUFFIX)
REDIS_SERVER_OBJ=threads_mngr.o adlist.o quicklist.o ae.o anet.o dict.o kvstore.o server.o sds.o zmalloc.o lzf_c.o lzf_d.o pqsort.o zipmap.o sha1.o ziplist.o release.o networking.o util.o object.o db.o replication.o rdb.o t_string.o t_list.o t_set.o t_zset.o t_hash.o config.o aof.o pubsub.o multi.o debug.o sort.o intset.o syncio.o cluster.o cluster_legacy.o crc16.o endianconv.o slowlog.o eval.o bio.o rio.o rand.o memtest.o syscheck.o crcspeed.o crc64.o bitops.o sentinel.o notify.o setproctitle.o blocked.o hyperloglog.o latency.o sparkline.o redis-check-rdb.o redis-check-aof.o geo.o lazyfree.o module.o evict.o expire.o geohash.o geohash_helper.o childinfo.o defrag.o siphash.o rax.o t_stream.o listpack.o localtime.o lolwut.o lolwut5.o lolwut6.o acl.o tracking.o socket.o tls.o sha256.o timeout.o setcpuaffinity.o monotonic.o mt19937-64.o resp_parser.o call_reply.o script_lua.o script.o functions.o function_lua.o commands.o strl.o connection.o unix.o logreqres.o
REDIS_SERVER_OBJ=threads_mngr.o adlist.o quicklist.o ae.o anet.o dict.o ebuckets.o mstr.o kvstore.o server.o sds.o zmalloc.o lzf_c.o lzf_d.o pqsort.o zipmap.o sha1.o ziplist.o release.o networking.o util.o object.o db.o replication.o rdb.o t_string.o t_list.o t_set.o t_zset.o t_hash.o config.o aof.o pubsub.o multi.o debug.o sort.o intset.o syncio.o cluster.o cluster_legacy.o crc16.o endianconv.o slowlog.o eval.o bio.o rio.o rand.o memtest.o syscheck.o crcspeed.o crc64.o bitops.o sentinel.o notify.o setproctitle.o blocked.o hyperloglog.o latency.o sparkline.o redis-check-rdb.o redis-check-aof.o geo.o lazyfree.o module.o evict.o expire.o geohash.o geohash_helper.o childinfo.o defrag.o siphash.o rax.o t_stream.o listpack.o localtime.o lolwut.o lolwut5.o lolwut6.o acl.o tracking.o socket.o tls.o sha256.o timeout.o setcpuaffinity.o monotonic.o mt19937-64.o resp_parser.o call_reply.o script_lua.o script.o functions.o function_lua.o commands.o strl.o connection.o unix.o logreqres.o
REDIS_CLI_NAME=redis-cli$(PROG_SUFFIX)
REDIS_CLI_OBJ=anet.o adlist.o dict.o redis-cli.o zmalloc.o release.o ae.o redisassert.o crcspeed.o crc64.o siphash.o crc16.o monotonic.o cli_common.o mt19937-64.o strl.o cli_commands.o
REDIS_BENCHMARK_NAME=redis-benchmark$(PROG_SUFFIX)
src/aof.c
+51 -26
View File
@@ -1939,19 +1939,21 @@ int rewriteSortedSetObject(rio *r, robj *key, robj *o) {
*
* The function returns 0 on error, non-zero on success. */
static int rioWriteHashIteratorCursor(rio *r, hashTypeIterator *hi, int what) {
if (hi->encoding == OBJ_ENCODING_LISTPACK) {
if ((hi->encoding == OBJ_ENCODING_LISTPACK) || (hi->encoding == OBJ_ENCODING_LISTPACK_EX)) {
unsigned char *vstr = NULL;
unsigned int vlen = UINT_MAX;
long long vll = LLONG_MAX;
hashTypeCurrentFromListpack(hi, what, &vstr, &vlen, &vll);
hashTypeCurrentFromListpack(hi, what, &vstr, &vlen, &vll, NULL);
if (vstr)
return rioWriteBulkString(r, (char*)vstr, vlen);
else
return rioWriteBulkLongLong(r, vll);
} else if (hi->encoding == OBJ_ENCODING_HT) {
sds value = hashTypeCurrentFromHashTable(hi, what);
return rioWriteBulkString(r, value, sdslen(value));
char *str;
size_t len;
hashTypeCurrentFromHashTable(hi, what, &str, &len, NULL);
return rioWriteBulkString(r, str, len);
}
serverPanic("Unknown hash encoding");
@@ -1961,37 +1963,60 @@ static int rioWriteHashIteratorCursor(rio *r, hashTypeIterator *hi, int what) {
/* Emit the commands needed to rebuild a hash object.
* The function returns 0 on error, 1 on success. */
int rewriteHashObject(rio *r, robj *key, robj *o) {
int res = 0; /*fail*/
hashTypeIterator *hi;
long long count = 0, items = hashTypeLength(o);
long long count = 0, items = hashTypeLength(o, 0);
int isHFE = hashTypeGetMinExpire(o) != EB_EXPIRE_TIME_INVALID;
hi = hashTypeInitIterator(o);
while (hashTypeNext(hi) != C_ERR) {
if (count == 0) {
int cmd_items = (items > AOF_REWRITE_ITEMS_PER_CMD) ?
AOF_REWRITE_ITEMS_PER_CMD : items;
if (!rioWriteBulkCount(r,'*',2+cmd_items*2) ||
!rioWriteBulkString(r,"HMSET",5) ||
!rioWriteBulkObject(r,key))
{
hashTypeReleaseIterator(hi);
return 0;
if (!isHFE) {
while (hashTypeNext(hi, 0) != C_ERR) {
if (count == 0) {
int cmd_items = (items > AOF_REWRITE_ITEMS_PER_CMD) ?
AOF_REWRITE_ITEMS_PER_CMD : items;
if (!rioWriteBulkCount(r, '*', 2 + cmd_items * 2) ||
!rioWriteBulkString(r, "HMSET", 5) ||
!rioWriteBulkObject(r, key))
goto reHashEnd;
}
if (!rioWriteHashIteratorCursor(r, hi, OBJ_HASH_KEY) ||
!rioWriteHashIteratorCursor(r, hi, OBJ_HASH_VALUE))
goto reHashEnd;
if (++count == AOF_REWRITE_ITEMS_PER_CMD) count = 0;
items--;
}
} else {
while (hashTypeNext(hi, 0) != C_ERR) {
char hmsetCmd[] = "*4\r\n$5\r\nHMSET\r\n";
if ( (!rioWrite(r, hmsetCmd, sizeof(hmsetCmd) - 1)) ||
(!rioWriteBulkObject(r, key)) ||
(!rioWriteHashIteratorCursor(r, hi, OBJ_HASH_KEY)) ||
(!rioWriteHashIteratorCursor(r, hi, OBJ_HASH_VALUE)) )
goto reHashEnd;
if (hi->expire_time != EB_EXPIRE_TIME_INVALID) {
char cmd[] = "*6\r\n$10\r\nHPEXPIREAT\r\n";
if ( (!rioWrite(r, cmd, sizeof(cmd) - 1)) ||
(!rioWriteBulkObject(r, key)) ||
(!rioWriteBulkLongLong(r, hi->expire_time)) ||
(!rioWriteBulkString(r, "FIELDS", 6)) ||
(!rioWriteBulkString(r, "1", 1)) ||
(!rioWriteHashIteratorCursor(r, hi, OBJ_HASH_KEY)) )
goto reHashEnd;
}
}
if (!rioWriteHashIteratorCursor(r, hi, OBJ_HASH_KEY) ||
!rioWriteHashIteratorCursor(r, hi, OBJ_HASH_VALUE))
{
hashTypeReleaseIterator(hi);
return 0;
}
if (++count == AOF_REWRITE_ITEMS_PER_CMD) count = 0;
items--;
}
hashTypeReleaseIterator(hi);
res = 1; /* success */
return 1;
reHashEnd:
hashTypeReleaseIterator(hi);
return res;
}
/* Helper for rewriteStreamObject() that generates a bulk string into the
src/cluster.c
+9 -2
View File
@@ -176,6 +176,7 @@ void dumpCommand(client *c) {
/* RESTORE key ttl serialized-value [REPLACE] [ABSTTL] [IDLETIME seconds] [FREQ frequency] */
void restoreCommand(client *c) {
uint64_t minExpiredField = EB_EXPIRE_TIME_INVALID;
long long ttl, lfu_freq = -1, lru_idle = -1, lru_clock = -1;
rio payload;
int j, type, replace = 0, absttl = 0;
@@ -239,7 +240,7 @@ void restoreCommand(client *c) {
rioInitWithBuffer(&payload,c->argv[3]->ptr);
if (((type = rdbLoadObjectType(&payload)) == -1) ||
((obj = rdbLoadObject(type,&payload,key->ptr,c->db->id,NULL)) == NULL))
((obj = rdbLoadObject(type,&payload,key->ptr,c->db,NULL, &minExpiredField)) == NULL))
{
addReplyError(c,"Bad data format");
return;
@@ -265,7 +266,13 @@ void restoreCommand(client *c) {
}
/* Create the key and set the TTL if any */
dbAdd(c->db,key,obj);
dictEntry *de = dbAdd(c->db,key,obj);
/* If minExpiredField was set, then the object is hash with expiration
* on fields and need to register it in global HFE DS */
if (minExpiredField != EB_EXPIRE_TIME_INVALID)
hashTypeAddToExpires(c->db, dictGetKey(de), obj, minExpiredField);
if (ttl) {
setExpire(c,c->db,key,ttl);
if (!absttl) {
src/commands.def
+292
View File
@@ -3303,6 +3303,107 @@ struct COMMAND_ARG HEXISTS_Args[] = {
{MAKE_ARG("field",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
};
/********** HEXPIRE ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
/* HEXPIRE history */
#define HEXPIRE_History NULL
#endif
#ifndef SKIP_CMD_TIPS_TABLE
/* HEXPIRE tips */
#define HEXPIRE_Tips NULL
#endif
#ifndef SKIP_CMD_KEY_SPECS_TABLE
/* HEXPIRE key specs */
keySpec HEXPIRE_Keyspecs[1] = {
{NULL,CMD_KEY_RW|CMD_KEY_UPDATE,KSPEC_BS_INDEX,.bs.index={1},KSPEC_FK_RANGE,.fk.range={0,1,0}}
};
#endif
/* HEXPIRE condition argument table */
struct COMMAND_ARG HEXPIRE_condition_Subargs[] = {
{MAKE_ARG("nx",ARG_TYPE_PURE_TOKEN,-1,"NX",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("xx",ARG_TYPE_PURE_TOKEN,-1,"XX",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("gt",ARG_TYPE_PURE_TOKEN,-1,"GT",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("lt",ARG_TYPE_PURE_TOKEN,-1,"LT",NULL,NULL,CMD_ARG_NONE,0,NULL)},
};
/* HEXPIRE argument table */
struct COMMAND_ARG HEXPIRE_Args[] = {
{MAKE_ARG("key",ARG_TYPE_KEY,0,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("seconds",ARG_TYPE_INTEGER,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("condition",ARG_TYPE_ONEOF,-1,NULL,NULL,NULL,CMD_ARG_OPTIONAL,4,NULL),.subargs=HEXPIRE_condition_Subargs},
{MAKE_ARG("fields",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("numfields",ARG_TYPE_INTEGER,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("field",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_MULTIPLE,0,NULL)},
};
/********** HEXPIREAT ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
/* HEXPIREAT history */
#define HEXPIREAT_History NULL
#endif
#ifndef SKIP_CMD_TIPS_TABLE
/* HEXPIREAT tips */
#define HEXPIREAT_Tips NULL
#endif
#ifndef SKIP_CMD_KEY_SPECS_TABLE
/* HEXPIREAT key specs */
keySpec HEXPIREAT_Keyspecs[1] = {
{NULL,CMD_KEY_RW|CMD_KEY_UPDATE,KSPEC_BS_INDEX,.bs.index={1},KSPEC_FK_RANGE,.fk.range={0,1,0}}
};
#endif
/* HEXPIREAT condition argument table */
struct COMMAND_ARG HEXPIREAT_condition_Subargs[] = {
{MAKE_ARG("nx",ARG_TYPE_PURE_TOKEN,-1,"NX",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("xx",ARG_TYPE_PURE_TOKEN,-1,"XX",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("gt",ARG_TYPE_PURE_TOKEN,-1,"GT",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("lt",ARG_TYPE_PURE_TOKEN,-1,"LT",NULL,NULL,CMD_ARG_NONE,0,NULL)},
};
/* HEXPIREAT argument table */
struct COMMAND_ARG HEXPIREAT_Args[] = {
{MAKE_ARG("key",ARG_TYPE_KEY,0,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("unix-time-seconds",ARG_TYPE_UNIX_TIME,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("condition",ARG_TYPE_ONEOF,-1,NULL,NULL,NULL,CMD_ARG_OPTIONAL,4,NULL),.subargs=HEXPIREAT_condition_Subargs},
{MAKE_ARG("fields",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("numfields",ARG_TYPE_INTEGER,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("field",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_MULTIPLE,0,NULL)},
};
/********** HEXPIRETIME ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
/* HEXPIRETIME history */
#define HEXPIRETIME_History NULL
#endif
#ifndef SKIP_CMD_TIPS_TABLE
/* HEXPIRETIME tips */
#define HEXPIRETIME_Tips NULL
#endif
#ifndef SKIP_CMD_KEY_SPECS_TABLE
/* HEXPIRETIME key specs */
keySpec HEXPIRETIME_Keyspecs[1] = {
{NULL,CMD_KEY_RO|CMD_KEY_ACCESS,KSPEC_BS_INDEX,.bs.index={1},KSPEC_FK_RANGE,.fk.range={0,1,0}}
};
#endif
/* HEXPIRETIME argument table */
struct COMMAND_ARG HEXPIRETIME_Args[] = {
{MAKE_ARG("key",ARG_TYPE_KEY,0,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("fields",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("numfields",ARG_TYPE_INTEGER,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("field",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_MULTIPLE,0,NULL)},
};
/********** HGET ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
@@ -3512,6 +3613,161 @@ struct COMMAND_ARG HMSET_Args[] = {
{MAKE_ARG("data",ARG_TYPE_BLOCK,-1,NULL,NULL,NULL,CMD_ARG_MULTIPLE,2,NULL),.subargs=HMSET_data_Subargs},
};
/********** HPERSIST ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
/* HPERSIST history */
#define HPERSIST_History NULL
#endif
#ifndef SKIP_CMD_TIPS_TABLE
/* HPERSIST tips */
#define HPERSIST_Tips NULL
#endif
#ifndef SKIP_CMD_KEY_SPECS_TABLE
/* HPERSIST key specs */
keySpec HPERSIST_Keyspecs[1] = {
{NULL,CMD_KEY_RW|CMD_KEY_UPDATE,KSPEC_BS_INDEX,.bs.index={1},KSPEC_FK_RANGE,.fk.range={0,1,0}}
};
#endif
/* HPERSIST argument table */
struct COMMAND_ARG HPERSIST_Args[] = {
{MAKE_ARG("key",ARG_TYPE_KEY,0,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("fields",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("numfields",ARG_TYPE_INTEGER,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("field",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_MULTIPLE,0,NULL)},
};
/********** HPEXPIRE ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
/* HPEXPIRE history */
#define HPEXPIRE_History NULL
#endif
#ifndef SKIP_CMD_TIPS_TABLE
/* HPEXPIRE tips */
#define HPEXPIRE_Tips NULL
#endif
#ifndef SKIP_CMD_KEY_SPECS_TABLE
/* HPEXPIRE key specs */
keySpec HPEXPIRE_Keyspecs[1] = {
{NULL,CMD_KEY_RW|CMD_KEY_UPDATE,KSPEC_BS_INDEX,.bs.index={1},KSPEC_FK_RANGE,.fk.range={0,1,0}}
};
#endif
/* HPEXPIRE condition argument table */
struct COMMAND_ARG HPEXPIRE_condition_Subargs[] = {
{MAKE_ARG("nx",ARG_TYPE_PURE_TOKEN,-1,"NX",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("xx",ARG_TYPE_PURE_TOKEN,-1,"XX",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("gt",ARG_TYPE_PURE_TOKEN,-1,"GT",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("lt",ARG_TYPE_PURE_TOKEN,-1,"LT",NULL,NULL,CMD_ARG_NONE,0,NULL)},
};
/* HPEXPIRE argument table */
struct COMMAND_ARG HPEXPIRE_Args[] = {
{MAKE_ARG("key",ARG_TYPE_KEY,0,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("milliseconds",ARG_TYPE_INTEGER,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("condition",ARG_TYPE_ONEOF,-1,NULL,NULL,NULL,CMD_ARG_OPTIONAL,4,NULL),.subargs=HPEXPIRE_condition_Subargs},
{MAKE_ARG("fields",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("numfields",ARG_TYPE_INTEGER,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("field",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_MULTIPLE,0,NULL)},
};
/********** HPEXPIREAT ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
/* HPEXPIREAT history */
#define HPEXPIREAT_History NULL
#endif
#ifndef SKIP_CMD_TIPS_TABLE
/* HPEXPIREAT tips */
#define HPEXPIREAT_Tips NULL
#endif
#ifndef SKIP_CMD_KEY_SPECS_TABLE
/* HPEXPIREAT key specs */
keySpec HPEXPIREAT_Keyspecs[1] = {
{NULL,CMD_KEY_RW|CMD_KEY_UPDATE,KSPEC_BS_INDEX,.bs.index={1},KSPEC_FK_RANGE,.fk.range={0,1,0}}
};
#endif
/* HPEXPIREAT condition argument table */
struct COMMAND_ARG HPEXPIREAT_condition_Subargs[] = {
{MAKE_ARG("nx",ARG_TYPE_PURE_TOKEN,-1,"NX",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("xx",ARG_TYPE_PURE_TOKEN,-1,"XX",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("gt",ARG_TYPE_PURE_TOKEN,-1,"GT",NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("lt",ARG_TYPE_PURE_TOKEN,-1,"LT",NULL,NULL,CMD_ARG_NONE,0,NULL)},
};
/* HPEXPIREAT argument table */
struct COMMAND_ARG HPEXPIREAT_Args[] = {
{MAKE_ARG("key",ARG_TYPE_KEY,0,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("unix-time-milliseconds",ARG_TYPE_UNIX_TIME,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("condition",ARG_TYPE_ONEOF,-1,NULL,NULL,NULL,CMD_ARG_OPTIONAL,4,NULL),.subargs=HPEXPIREAT_condition_Subargs},
{MAKE_ARG("fields",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("numfields",ARG_TYPE_INTEGER,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("field",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_MULTIPLE,0,NULL)},
};
/********** HPEXPIRETIME ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
/* HPEXPIRETIME history */
#define HPEXPIRETIME_History NULL
#endif
#ifndef SKIP_CMD_TIPS_TABLE
/* HPEXPIRETIME tips */
#define HPEXPIRETIME_Tips NULL
#endif
#ifndef SKIP_CMD_KEY_SPECS_TABLE
/* HPEXPIRETIME key specs */
keySpec HPEXPIRETIME_Keyspecs[1] = {
{NULL,CMD_KEY_RO|CMD_KEY_ACCESS,KSPEC_BS_INDEX,.bs.index={1},KSPEC_FK_RANGE,.fk.range={0,1,0}}
};
#endif
/* HPEXPIRETIME argument table */
struct COMMAND_ARG HPEXPIRETIME_Args[] = {
{MAKE_ARG("key",ARG_TYPE_KEY,0,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("fields",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("numfields",ARG_TYPE_INTEGER,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("field",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_MULTIPLE,0,NULL)},
};
/********** HPTTL ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
/* HPTTL history */
#define HPTTL_History NULL
#endif
#ifndef SKIP_CMD_TIPS_TABLE
/* HPTTL tips */
#define HPTTL_Tips NULL
#endif
#ifndef SKIP_CMD_KEY_SPECS_TABLE
/* HPTTL key specs */
keySpec HPTTL_Keyspecs[1] = {
{NULL,CMD_KEY_RO|CMD_KEY_ACCESS,KSPEC_BS_INDEX,.bs.index={1},KSPEC_FK_RANGE,.fk.range={0,1,0}}
};
#endif
/* HPTTL argument table */
struct COMMAND_ARG HPTTL_Args[] = {
{MAKE_ARG("key",ARG_TYPE_KEY,0,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("fields",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("numfields",ARG_TYPE_INTEGER,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("field",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_MULTIPLE,0,NULL)},
};
/********** HRANDFIELD ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
@@ -3659,6 +3915,33 @@ struct COMMAND_ARG HSTRLEN_Args[] = {
{MAKE_ARG("field",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
};
/********** HTTL ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
/* HTTL history */
#define HTTL_History NULL
#endif
#ifndef SKIP_CMD_TIPS_TABLE
/* HTTL tips */
#define HTTL_Tips NULL
#endif
#ifndef SKIP_CMD_KEY_SPECS_TABLE
/* HTTL key specs */
keySpec HTTL_Keyspecs[1] = {
{NULL,CMD_KEY_RO|CMD_KEY_ACCESS,KSPEC_BS_INDEX,.bs.index={1},KSPEC_FK_RANGE,.fk.range={0,1,0}}
};
#endif
/* HTTL argument table */
struct COMMAND_ARG HTTL_Args[] = {
{MAKE_ARG("key",ARG_TYPE_KEY,0,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("fields",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("numfields",ARG_TYPE_INTEGER,-1,NULL,NULL,NULL,CMD_ARG_NONE,0,NULL)},
{MAKE_ARG("field",ARG_TYPE_STRING,-1,NULL,NULL,NULL,CMD_ARG_MULTIPLE,0,NULL)},
};
/********** HVALS ********************/
#ifndef SKIP_CMD_HISTORY_TABLE
@@ -10710,6 +10993,9 @@ struct COMMAND_STRUCT redisCommandTable[] = {
/* hash */
{MAKE_CMD("hdel","Deletes one or more fields and their values from a hash. Deletes the hash if no fields remain.","O(N) where N is the number of fields to be removed.","2.0.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HDEL_History,1,HDEL_Tips,0,hdelCommand,-3,CMD_WRITE|CMD_FAST,ACL_CATEGORY_HASH,HDEL_Keyspecs,1,NULL,2),.args=HDEL_Args},
{MAKE_CMD("hexists","Determines whether a field exists in a hash.","O(1)","2.0.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HEXISTS_History,0,HEXISTS_Tips,0,hexistsCommand,3,CMD_READONLY|CMD_FAST,ACL_CATEGORY_HASH,HEXISTS_Keyspecs,1,NULL,2),.args=HEXISTS_Args},
{MAKE_CMD("hexpire","Set expiry for hash field using relative time to expire (seconds)","O(N) where N is the number of specified fields","7.4.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HEXPIRE_History,0,HEXPIRE_Tips,0,hexpireCommand,-6,CMD_WRITE|CMD_DENYOOM|CMD_FAST,ACL_CATEGORY_HASH,HEXPIRE_Keyspecs,1,NULL,6),.args=HEXPIRE_Args},
{MAKE_CMD("hexpireat","Set expiry for hash field using an absolute Unix timestamp (seconds)","O(N) where N is the number of specified fields","7.4.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HEXPIREAT_History,0,HEXPIREAT_Tips,0,hexpireatCommand,-6,CMD_WRITE|CMD_DENYOOM|CMD_FAST,ACL_CATEGORY_HASH,HEXPIREAT_Keyspecs,1,NULL,6),.args=HEXPIREAT_Args},
{MAKE_CMD("hexpiretime","Returns the expiration time of a hash field as a Unix timestamp, in seconds.","O(N) where N is the number of specified fields","7.4.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HEXPIRETIME_History,0,HEXPIRETIME_Tips,0,hexpiretimeCommand,-5,CMD_READONLY|CMD_FAST,ACL_CATEGORY_HASH,HEXPIRETIME_Keyspecs,1,NULL,4),.args=HEXPIRETIME_Args},
{MAKE_CMD("hget","Returns the value of a field in a hash.","O(1)","2.0.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HGET_History,0,HGET_Tips,0,hgetCommand,3,CMD_READONLY|CMD_FAST,ACL_CATEGORY_HASH,HGET_Keyspecs,1,NULL,2),.args=HGET_Args},
{MAKE_CMD("hgetall","Returns all fields and values in a hash.","O(N) where N is the size of the hash.","2.0.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HGETALL_History,0,HGETALL_Tips,1,hgetallCommand,2,CMD_READONLY,ACL_CATEGORY_HASH,HGETALL_Keyspecs,1,NULL,1),.args=HGETALL_Args},
{MAKE_CMD("hincrby","Increments the integer value of a field in a hash by a number. Uses 0 as initial value if the field doesn't exist.","O(1)","2.0.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HINCRBY_History,0,HINCRBY_Tips,0,hincrbyCommand,4,CMD_WRITE|CMD_DENYOOM|CMD_FAST,ACL_CATEGORY_HASH,HINCRBY_Keyspecs,1,NULL,3),.args=HINCRBY_Args},
@@ -10718,11 +11004,17 @@ struct COMMAND_STRUCT redisCommandTable[] = {
{MAKE_CMD("hlen","Returns the number of fields in a hash.","O(1)","2.0.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HLEN_History,0,HLEN_Tips,0,hlenCommand,2,CMD_READONLY|CMD_FAST,ACL_CATEGORY_HASH,HLEN_Keyspecs,1,NULL,1),.args=HLEN_Args},
{MAKE_CMD("hmget","Returns the values of all fields in a hash.","O(N) where N is the number of fields being requested.","2.0.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HMGET_History,0,HMGET_Tips,0,hmgetCommand,-3,CMD_READONLY|CMD_FAST,ACL_CATEGORY_HASH,HMGET_Keyspecs,1,NULL,2),.args=HMGET_Args},
{MAKE_CMD("hmset","Sets the values of multiple fields.","O(N) where N is the number of fields being set.","2.0.0",CMD_DOC_DEPRECATED,"`HSET` with multiple field-value pairs","4.0.0","hash",COMMAND_GROUP_HASH,HMSET_History,0,HMSET_Tips,0,hsetCommand,-4,CMD_WRITE|CMD_DENYOOM|CMD_FAST,ACL_CATEGORY_HASH,HMSET_Keyspecs,1,NULL,2),.args=HMSET_Args},
{MAKE_CMD("hpersist","Removes the expiration time for each specified field","O(N) where N is the number of specified fields","7.4.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HPERSIST_History,0,HPERSIST_Tips,0,hpersistCommand,-5,CMD_WRITE|CMD_FAST,ACL_CATEGORY_HASH,HPERSIST_Keyspecs,1,NULL,4),.args=HPERSIST_Args},
{MAKE_CMD("hpexpire","Set expiry for hash field using relative time to expire (milliseconds)","O(N) where N is the number of specified fields","7.4.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HPEXPIRE_History,0,HPEXPIRE_Tips,0,hpexpireCommand,-6,CMD_WRITE|CMD_DENYOOM|CMD_FAST,ACL_CATEGORY_HASH,HPEXPIRE_Keyspecs,1,NULL,6),.args=HPEXPIRE_Args},
{MAKE_CMD("hpexpireat","Set expiry for hash field using an absolute Unix timestamp (milliseconds)","O(N) where N is the number of specified fields","7.4.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HPEXPIREAT_History,0,HPEXPIREAT_Tips,0,hpexpireatCommand,-6,CMD_WRITE|CMD_DENYOOM|CMD_FAST,ACL_CATEGORY_HASH,HPEXPIREAT_Keyspecs,1,NULL,6),.args=HPEXPIREAT_Args},
{MAKE_CMD("hpexpiretime","Returns the expiration time of a hash field as a Unix timestamp, in msec.","O(N) where N is the number of specified fields","7.4.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HPEXPIRETIME_History,0,HPEXPIRETIME_Tips,0,hpexpiretimeCommand,-5,CMD_READONLY|CMD_FAST,ACL_CATEGORY_HASH,HPEXPIRETIME_Keyspecs,1,NULL,4),.args=HPEXPIRETIME_Args},
{MAKE_CMD("hpttl","Returns the TTL in milliseconds of a hash field.","O(N) where N is the number of specified fields","7.4.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HPTTL_History,0,HPTTL_Tips,0,hpttlCommand,-5,CMD_READONLY|CMD_FAST,ACL_CATEGORY_HASH,HPTTL_Keyspecs,1,NULL,4),.args=HPTTL_Args},
{MAKE_CMD("hrandfield","Returns one or more random fields from a hash.","O(N) where N is the number of fields returned","6.2.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HRANDFIELD_History,0,HRANDFIELD_Tips,1,hrandfieldCommand,-2,CMD_READONLY,ACL_CATEGORY_HASH,HRANDFIELD_Keyspecs,1,NULL,2),.args=HRANDFIELD_Args},
{MAKE_CMD("hscan","Iterates over fields and values of a hash.","O(1) for every call. O(N) for a complete iteration, including enough command calls for the cursor to return back to 0. N is the number of elements inside the collection.","2.8.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HSCAN_History,0,HSCAN_Tips,1,hscanCommand,-3,CMD_READONLY,ACL_CATEGORY_HASH,HSCAN_Keyspecs,1,NULL,5),.args=HSCAN_Args},
{MAKE_CMD("hset","Creates or modifies the value of a field in a hash.","O(1) for each field/value pair added, so O(N) to add N field/value pairs when the command is called with multiple field/value pairs.","2.0.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HSET_History,1,HSET_Tips,0,hsetCommand,-4,CMD_WRITE|CMD_DENYOOM|CMD_FAST,ACL_CATEGORY_HASH,HSET_Keyspecs,1,NULL,2),.args=HSET_Args},
{MAKE_CMD("hsetnx","Sets the value of a field in a hash only when the field doesn't exist.","O(1)","2.0.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HSETNX_History,0,HSETNX_Tips,0,hsetnxCommand,4,CMD_WRITE|CMD_DENYOOM|CMD_FAST,ACL_CATEGORY_HASH,HSETNX_Keyspecs,1,NULL,3),.args=HSETNX_Args},
{MAKE_CMD("hstrlen","Returns the length of the value of a field.","O(1)","3.2.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HSTRLEN_History,0,HSTRLEN_Tips,0,hstrlenCommand,3,CMD_READONLY|CMD_FAST,ACL_CATEGORY_HASH,HSTRLEN_Keyspecs,1,NULL,2),.args=HSTRLEN_Args},
{MAKE_CMD("httl","Returns the TTL in seconds of a hash field.","O(N) where N is the number of specified fields","7.4.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HTTL_History,0,HTTL_Tips,0,httlCommand,-5,CMD_READONLY|CMD_FAST,ACL_CATEGORY_HASH,HTTL_Keyspecs,1,NULL,4),.args=HTTL_Args},
{MAKE_CMD("hvals","Returns all values in a hash.","O(N) where N is the size of the hash.","2.0.0",CMD_DOC_NONE,NULL,NULL,"hash",COMMAND_GROUP_HASH,HVALS_History,0,HVALS_Tips,1,hvalsCommand,2,CMD_READONLY,ACL_CATEGORY_HASH,HVALS_Keyspecs,1,NULL,1),.args=HVALS_Args},
/* hyperloglog */
{MAKE_CMD("pfadd","Adds elements to a HyperLogLog key. Creates the key if it doesn't exist.","O(1) to add every element.","2.8.9",CMD_DOC_NONE,NULL,NULL,"hyperloglog",COMMAND_GROUP_HYPERLOGLOG,PFADD_History,0,PFADD_Tips,0,pfaddCommand,-2,CMD_WRITE|CMD_DENYOOM|CMD_FAST,ACL_CATEGORY_HYPERLOGLOG,PFADD_Keyspecs,1,NULL,2),.args=PFADD_Args},
src/commands/hexpire.json
+116
View File
@@ -0,0 +1,116 @@
{
"HEXPIRE": {
"summary": "Set expiry for hash field using relative time to expire (seconds)",
"complexity": "O(N) where N is the number of specified fields",
"group": "hash",
"since": "7.4.0",
"arity": -6,
"function": "hexpireCommand",
"history": [],
"command_flags": [
"WRITE",
"DENYOOM",
"FAST"
],
"acl_categories": [
"HASH"
],
"key_specs": [
{
"flags": [
"RW",
"UPDATE"
],
"begin_search": {
"index": {
"pos": 1
}
},
"find_keys": {
"range": {
"lastkey": 0,
"step": 1,
"limit": 0
}
}
}
],
"reply_schema": {
"description": "Array of results. Returns empty array if the key does not exist.",
"type": "array",
"minItems": 0,
"maxItems": 4294967295,
"items": {
"oneOf": [
{
"description": "The field does not exist.",
"const": -2
},
{
"description": "Specified NX | XX | GT | LT condition not met",
"const": 0
},
{
"description": "Expiration time was set or updated.",
"const": 1
},
{
"description": "Field deleted because the specified expiration time is in the past.",
"const": 2
}
]
}
},
"arguments": [
{
"name": "key",
"type": "key",
"key_spec_index": 0
},
{
"name": "seconds",
"type": "integer"
},
{
"name": "condition",
"type": "oneof",
"optional": true,
"arguments": [
{
"name": "nx",
"type": "pure-token",
"token": "NX"
},
{
"name": "xx",
"type": "pure-token",
"token": "XX"
},
{
"name": "gt",
"type": "pure-token",
"token": "GT"
},
{
"name": "lt",
"type": "pure-token",
"token": "LT"
}
]
},
{
"name": "FIELDS",
"type": "string"
},
{
"name": "numfields",
"type": "integer"
},
{
"name": "field",
"type": "string",
"multiple": true
}
]
}
}
src/commands/hexpireat.json
+116
View File
@@ -0,0 +1,116 @@
{
"HEXPIREAT": {
"summary": "Set expiry for hash field using an absolute Unix timestamp (seconds)",
"complexity": "O(N) where N is the number of specified fields",
"group": "hash",
"since": "7.4.0",
"arity": -6,
"function": "hexpireatCommand",
"history": [],
"command_flags": [
"WRITE",
"DENYOOM",
"FAST"
],
"acl_categories": [
"HASH"
],
"key_specs": [
{
"flags": [
"RW",
"UPDATE"
],
"begin_search": {
"index": {
"pos": 1
}
},
"find_keys": {
"range": {
"lastkey": 0,
"step": 1,
"limit": 0
}
}
}
],
"reply_schema": {
"description": "Array of results. Returns empty array if the key does not exist.",
"type": "array",
"minItems": 0,
"maxItems": 4294967295,
"items": {
"oneOf": [
{
"description": "The field does not exist.",
"const": -2
},
{
"description": "Specified NX | XX | GT | LT condition not met",
"const": 0
},
{
"description": "Expiration time was set or updated.",
"const": 1
},
{
"description": "Field deleted because the specified expiration time is in the past.",
"const": 2
}
]
}
},
"arguments": [
{
"name": "key",
"type": "key",
"key_spec_index": 0
},
{
"name": "unix-time-seconds",
"type": "unix-time"
},
{
"name": "condition",
"type": "oneof",
"optional": true,
"arguments": [
{
"name": "nx",
"type": "pure-token",
"token": "NX"
},
{
"name": "xx",
"type": "pure-token",
"token": "XX"
},
{
"name": "gt",
"type": "pure-token",
"token": "GT"
},
{
"name": "lt",
"type": "pure-token",
"token": "LT"
}
]
},
{
"name": "FIELDS",
"type": "string"
},
{
"name": "numfields",
"type": "integer"
},
{
"name": "field",
"type": "string",
"multiple": true
}
]
}
}
src/commands/hexpiretime.json
+81
View File
@@ -0,0 +1,81 @@
{
"HEXPIRETIME": {
"summary": "Returns the expiration time of a hash field as a Unix timestamp, in seconds.",
"complexity": "O(N) where N is the number of specified fields",
"group": "hash",
"since": "7.4.0",
"arity": -5,
"function": "hexpiretimeCommand",
"history": [],
"command_flags": [
"READONLY",
"FAST"
],
"acl_categories": [
"HASH"
],
"key_specs": [
{
"flags": [
"RO",
"ACCESS"
],
"begin_search": {
"index": {
"pos": 1
}
},
"find_keys": {
"range": {
"lastkey": 0,
"step": 1,
"limit": 0
}
}
}
],
"reply_schema": {
"description": "Array of results. Returns empty array if the key does not exist.",
"type": "array",
"minItems": 0,
"maxItems": 4294967295,
"items": {
"oneOf": [
{
"description": "The field does not exist.",
"const": -2
},
{
"description": "The field exists but has no associated expire.",
"const": -1
},
{
"description": "Expiration Unix timestamp in seconds.",
"type": "integer",
"minimum": 1
}
]
}
},
"arguments": [
{
"name": "key",
"type": "key",
"key_spec_index": 0
},
{
"name": "FIELDS",
"type": "string"
},
{
"name": "numfields",
"type": "integer"
},
{
"name": "field",
"type": "string",
"multiple": true
}
]
}
}
src/commands/hpersist.json
+80
View File
@@ -0,0 +1,80 @@
{
"HPERSIST": {
"summary": "Removes the expiration time for each specified field",
"complexity": "O(N) where N is the number of specified fields",
"group": "hash",
"since": "7.4.0",
"arity": -5,
"function": "hpersistCommand",
"history": [],
"command_flags": [
"WRITE",
"FAST"
],
"acl_categories": [
"HASH"
],
"key_specs": [
{
"flags": [
"RW",
"UPDATE"
],
"begin_search": {
"index": {
"pos": 1
}
},
"find_keys": {
"range": {
"lastkey": 0,
"step": 1,
"limit": 0
}
}
}
],
"reply_schema": {
"description": "Array of results. Returns empty array if the key does not exist.",
"type": "array",
"minItems": 0,
"maxItems": 4294967295,
"items": {
"oneOf": [
{
"description": "The field does not exist.",
"const": -2
},
{
"description": "The field exists but has no associated expire.",
"const": -1
},
{
"description": "Expiration time was removed",
"const": 1
}
]
}
},
"arguments": [
{
"name": "key",
"type": "key",
"key_spec_index": 0
},
{
"name": "FIELDS",
"type": "string"
},
{
"name": "numfields",
"type": "integer"
},
{
"name": "field",
"type": "string",
"multiple": true
}
]
}
}
src/commands/hpexpire.json
+116
View File
@@ -0,0 +1,116 @@
{
"HPEXPIRE": {
"summary": "Set expiry for hash field using relative time to expire (milliseconds)",
"complexity": "O(N) where N is the number of specified fields",
"group": "hash",
"since": "7.4.0",
"arity": -6,
"function": "hpexpireCommand",
"history": [],
"command_flags": [
"WRITE",
"DENYOOM",
"FAST"
],
"acl_categories": [
"HASH"
],
"key_specs": [
{
"flags": [
"RW",
"UPDATE"
],
"begin_search": {
"index": {
"pos": 1
}
},
"find_keys": {
"range": {
"lastkey": 0,
"step": 1,
"limit": 0
}
}
}
],
"reply_schema": {
"description": "Array of results. Returns empty array if the key does not exist.",
"type": "array",
"minItems": 0,
"maxItems": 4294967295,
"items": {
"oneOf": [
{
"description": "The field does not exist.",
"const": -2
},
{
"description": "Specified NX | XX | GT | LT condition not met",
"const": 0
},
{
"description": "Expiration time was set or updated.",
"const": 1
},
{
"description": "Field deleted because the specified expiration time is in the past.",
"const": 2
}
]
}
},
"arguments": [
{
"name": "key",
"type": "key",
"key_spec_index": 0
},
{
"name": "milliseconds",
"type": "integer"
},
{
"name": "condition",
"type": "oneof",
"optional": true,
"arguments": [
{
"name": "nx",
"type": "pure-token",
"token": "NX"
},
{
"name": "xx",
"type": "pure-token",
"token": "XX"
},
{
"name": "gt",
"type": "pure-token",
"token": "GT"
},
{
"name": "lt",
"type": "pure-token",
"token": "LT"
}
]
},
{
"name": "FIELDS",
"type": "string"
},
{
"name": "numfields",
"type": "integer"
},
{
"name": "field",
"type": "string",
"multiple": true
}
]
}
}
src/commands/hpexpireat.json
+116
View File
@@ -0,0 +1,116 @@
{
"HPEXPIREAT": {
"summary": "Set expiry for hash field using an absolute Unix timestamp (milliseconds)",
"complexity": "O(N) where N is the number of specified fields",
"group": "hash",
"since": "7.4.0",
"arity": -6,
"function": "hpexpireatCommand",
"history": [],
"command_flags": [
"WRITE",
"DENYOOM",
"FAST"
],
"acl_categories": [
"HASH"
],
"key_specs": [
{
"flags": [
"RW",
"UPDATE"
],
"begin_search": {
"index": {
"pos": 1
}
},
"find_keys": {
"range": {
"lastkey": 0,
"step": 1,
"limit": 0
}
}
}
],
"reply_schema": {
"description": "Array of results. Returns empty array if the key does not exist.",
"type": "array",
"minItems": 0,
"maxItems": 4294967295,
"items": {
"oneOf": [
{
"description": "The field does not exist.",
"const": -2
},
{
"description": "Specified NX | XX | GT | LT condition not met",
"const": 0
},
{
"description": "Expiration time was set or updated.",
"const": 1
},
{
"description": "Field deleted because the specified expiration time is in the past.",
"const": 2
}
]
}
},
"arguments": [
{
"name": "key",
"type": "key",
"key_spec_index": 0
},
{
"name": "unix-time-milliseconds",
"type": "unix-time"
},
{
"name": "condition",
"type": "oneof",
"optional": true,
"arguments": [
{
"name": "nx",
"type": "pure-token",
"token": "NX"
},
{
"name": "xx",
"type": "pure-token",
"token": "XX"
},
{
"name": "gt",
"type": "pure-token",
"token": "GT"
},
{
"name": "lt",
"type": "pure-token",
"token": "LT"
}
]
},
{
"name": "FIELDS",
"type": "string"
},
{
"name": "numfields",
"type": "integer"
},
{
"name": "field",
"type": "string",
"multiple": true
}
]
}
}
src/commands/hpexpiretime.json
+81
View File
@@ -0,0 +1,81 @@
{
"HPEXPIRETIME": {
"summary": "Returns the expiration time of a hash field as a Unix timestamp, in msec.",
"complexity": "O(N) where N is the number of specified fields",
"group": "hash",
"since": "7.4.0",
"arity": -5,
"function": "hpexpiretimeCommand",
"history": [],
"command_flags": [
"READONLY",
"FAST"
],
"acl_categories": [
"HASH"
],
"key_specs": [
{
"flags": [
"RO",
"ACCESS"
],
"begin_search": {
"index": {
"pos": 1
}
},
"find_keys": {
"range": {
"lastkey": 0,
"step": 1,
"limit": 0
}
}
}
],
"reply_schema": {
"description": "Array of results. Returns empty array if the key does not exist.",
"type": "array",
"minItems": 0,
"maxItems": 4294967295,
"items": {
"oneOf": [
{
"description": "The field does not exist.",
"const": -2
},
{
"description": "The field exists but has no associated expire.",
"const": -1
},
{
"description": "Expiration Unix timestamp in milliseconds.",
"type": "integer",
"minimum": 1
}
]
}
},
"arguments": [
{
"name": "key",
"type": "key",
"key_spec_index": 0
},
{
"name": "FIELDS",
"type": "string"
},
{
"name": "numfields",
"type": "integer"
},
{
"name": "field",
"type": "string",
"multiple": true
}
]
}
}
src/commands/hpttl.json
+81
View File
@@ -0,0 +1,81 @@
{
"HPTTL": {
"summary": "Returns the TTL in milliseconds of a hash field.",
"complexity": "O(N) where N is the number of specified fields",
"group": "hash",
"since": "7.4.0",
"arity": -5,
"function": "hpttlCommand",
"history": [],
"command_flags": [
"READONLY",
"FAST"
],
"acl_categories": [
"HASH"
],
"key_specs": [
{
"flags": [
"RO",
"ACCESS"
],
"begin_search": {
"index": {
"pos": 1
}
},
"find_keys": {
"range": {
"lastkey": 0,
"step": 1,
"limit": 0
}
}
}
],
"reply_schema": {
"description": "Array of results. Returns empty array if the key does not exist.",
"type": "array",
"minItems": 0,
"maxItems": 4294967295,
"items": {
"oneOf": [
{
"description": "The field does not exist.",
"const": -2
},
{
"description": "The field exists but has no associated expire.",
"const": -1
},
{
"description": "TTL in milliseconds.",
"type": "integer",
"minimum": 1
}
]
}
},
"arguments": [
{
"name": "key",
"type": "key",
"key_spec_index": 0
},
{
"name": "FIELDS",
"type": "string"
},
{
"name": "numfields",
"type": "integer"
},
{
"name": "field",
"type": "string",
"multiple": true
}
]
}
}
src/commands/httl.json
+81
View File
@@ -0,0 +1,81 @@
{
"HTTL": {
"summary": "Returns the TTL in seconds of a hash field.",
"complexity": "O(N) where N is the number of specified fields",
"group": "hash",
"since": "7.4.0",
"arity": -5,
"function": "httlCommand",
"history": [],
"command_flags": [
"READONLY",
"FAST"
],
"acl_categories": [
"HASH"
],
"key_specs": [
{
"flags": [
"RO",
"ACCESS"
],
"begin_search": {
"index": {
"pos": 1
}
},
"find_keys": {
"range": {
"lastkey": 0,
"step": 1,
"limit": 0
}
}
}
],
"reply_schema": {
"description": "Array of results. Returns empty array if the key does not exist.",
"type": "array",
"minItems": 0,
"maxItems": 4294967295,
"items": {
"oneOf": [
{
"description": "The field does not exist.",
"const": -2
},
{
"description": "The field exists but has no associated expire.",
"const": -1
},
{
"description": "TTL in seconds.",
"type": "integer",
"minimum": 1
}
]
}
},
"arguments": [
{
"name": "key",
"type": "key",
"key_spec_index": 0
},
{
"name": "FIELDS",
"type": "string"
},
{
"name": "numfields",
"type": "integer"
},
{
"name": "field",
"type": "string",
"multiple": true
}
]
}
}
src/db.c
+126 -21
View File
@@ -177,13 +177,13 @@ robj *lookupKeyWriteOrReply(client *c, robj *key, robj *reply) {
*
* If the update_if_existing argument is false, the program is aborted
* if the key already exists, otherwise, it can fall back to dbOverwrite. */
static void dbAddInternal(redisDb *db, robj *key, robj *val, int update_if_existing) {
static dictEntry *dbAddInternal(redisDb *db, robj *key, robj *val, int update_if_existing) {
dictEntry *existing;
int slot = getKeySlot(key->ptr);
dictEntry *de = kvstoreDictAddRaw(db->keys, slot, key->ptr, &existing);
if (update_if_existing && existing) {
dbSetValue(db, key, val, 1, existing);
return;
return existing;
}
serverAssertWithInfo(NULL, key, de != NULL);
kvstoreDictSetKey(db->keys, slot, de, sdsdup(key->ptr));
@@ -191,10 +191,11 @@ static void dbAddInternal(redisDb *db, robj *key, robj *val, int update_if_exist
kvstoreDictSetVal(db->keys, slot, de, val);
signalKeyAsReady(db, key, val->type);
notifyKeyspaceEvent(NOTIFY_NEW,"new",key,db->id);
return de;
}
void dbAdd(redisDb *db, robj *key, robj *val) {
dbAddInternal(db, key, val, 0);
dictEntry *dbAdd(redisDb *db, robj *key, robj *val) {
return dbAddInternal(db, key, val, 0);
}
/* Returns key's hash slot when cluster mode is enabled, or 0 when disabled.
@@ -275,6 +276,11 @@ static void dbSetValue(redisDb *db, robj *key, robj *val, int overwrite, dictEnt
old = dictGetVal(de);
}
kvstoreDictSetVal(db->keys, slot, de, val);
/* if hash with HFEs, take care to remove from global HFE DS */
if (old->type == OBJ_HASH)
hashTypeRemoveFromExpires(&db->hexpires, old);
if (server.lazyfree_lazy_server_del) {
freeObjAsync(key,old,db->id);
} else {
@@ -370,6 +376,11 @@ int dbGenericDelete(redisDb *db, robj *key, int async, int flags) {
dictEntry *de = kvstoreDictTwoPhaseUnlinkFind(db->keys, slot, key->ptr, &plink, &table);
if (de) {
robj *val = dictGetVal(de);
/* If hash object with expiry on fields, remove it from HFE DS of DB */
if (val->type == OBJ_HASH)
hashTypeRemoveFromExpires(&db->hexpires, val);
/* RM_StringDMA may call dbUnshareStringValue which may free val, so we
* need to incr to retain val */
incrRefCount(val);
@@ -475,6 +486,9 @@ long long emptyDbStructure(redisDb *dbarray, int dbnum, int async,
if (async) {
emptyDbAsync(&dbarray[j]);
} else {
/* Destroy global HFE DS before deleting the hashes since ebuckets
* DS is embedded in the stored objects. */
ebDestroy(&dbarray[j].hexpires, &hashExpireBucketsType, NULL);
kvstoreEmpty(dbarray[j].keys, callback);
kvstoreEmpty(dbarray[j].expires, callback);
}
@@ -554,6 +568,7 @@ redisDb *initTempDb(void) {
tempDb[i].id = i;
tempDb[i].keys = kvstoreCreate(&dbDictType, slot_count_bits, flags);
tempDb[i].expires = kvstoreCreate(&dbExpiresDictType, slot_count_bits, flags);
tempDb[i].hexpires = ebCreate();
}
return tempDb;
@@ -566,6 +581,9 @@ void discardTempDb(redisDb *tempDb, void(callback)(dict*)) {
/* Release temp DBs. */
emptyDbStructure(tempDb, -1, async, callback);
for (int i=0; i<server.dbnum; i++) {
/* Destroy global HFE DS before deleting the hashes since ebuckets DS is
* embedded in the stored objects. */
ebDestroy(&tempDb[i].hexpires, &hashExpireBucketsType, NULL);
kvstoreRelease(tempDb[i].keys);
kvstoreRelease(tempDb[i].expires);
}
@@ -894,6 +912,7 @@ typedef struct {
sds pattern; /* pattern string, NULL means no pattern */
long sampled; /* cumulative number of keys sampled */
int no_values; /* set to 1 means to return keys only */
size_t (*strlen)(char *s); /* (o->type == OBJ_HASH) ? hfieldlen : sdslen */
} scanData;
/* Helper function to compare key type in scan commands */
@@ -918,7 +937,7 @@ void scanCallback(void *privdata, const dictEntry *de) {
list *keys = data->keys;
robj *o = data->o;
sds val = NULL;
sds key = NULL;
void *key = NULL; /* if OBJ_HASH then key is of type `hfield`. Otherwise, `sds` */
data->sampled++;
/* o and typename can not have values at the same time. */
@@ -932,24 +951,29 @@ void scanCallback(void *privdata, const dictEntry *de) {
}*/
/* Filter element if it does not match the pattern. */
sds keysds = dictGetKey(de);
void *keyStr = dictGetKey(de);
if (data->pattern) {
if (!stringmatchlen(data->pattern, sdslen(data->pattern), keysds, sdslen(keysds), 0)) {
if (!stringmatchlen(data->pattern, sdslen(data->pattern), keyStr, data->strlen(keyStr), 0)) {
return;
}
}
if (o == NULL) {
key = keysds;
key = keyStr;
} else if (o->type == OBJ_SET) {
key = keysds;
key = keyStr;
} else if (o->type == OBJ_HASH) {
key = keysds;
key = keyStr;
val = dictGetVal(de);
/* If field is expired, then ignore */
if (hfieldIsExpired(key))
return;
} else if (o->type == OBJ_ZSET) {
char buf[MAX_LONG_DOUBLE_CHARS];
int len = ld2string(buf, sizeof(buf), *(double *)dictGetVal(de), LD_STR_AUTO);
key = sdsdup(keysds);
key = sdsdup(keyStr);
val = sdsnewlen(buf, len);
} else {
serverPanic("Type not handled in SCAN callback.");
@@ -1023,6 +1047,7 @@ char *getObjectTypeName(robj *o) {
* In the case of a Hash object the function returns both the field and value
* of every element on the Hash. */
void scanGenericCommand(client *c, robj *o, unsigned long long cursor) {
int isKeysHfield = 0;
int i, j;
listNode *node;
long count = 10;
@@ -1103,6 +1128,7 @@ void scanGenericCommand(client *c, robj *o, unsigned long long cursor) {
} else if (o->type == OBJ_SET && o->encoding == OBJ_ENCODING_HT) {
ht = o->ptr;
} else if (o->type == OBJ_HASH && o->encoding == OBJ_ENCODING_HT) {
isKeysHfield = 1;
ht = o->ptr;
} else if (o->type == OBJ_ZSET && o->encoding == OBJ_ENCODING_SKIPLIST) {
zset *zs = o->ptr;
@@ -1141,7 +1167,7 @@ void scanGenericCommand(client *c, robj *o, unsigned long long cursor) {
* working on an empty dict, one with a lot of empty buckets, and
* for the buckets are not empty, we need to limit the spampled number
* to prevent a long hang time caused by filtering too many keys;
* 6. data.no_values: to control whether values will be returned or
* 6. data.no_values: to control whether values will be returned or
* only keys are returned. */
scanData data = {
.keys = keys,
@@ -1150,6 +1176,7 @@ void scanGenericCommand(client *c, robj *o, unsigned long long cursor) {
.pattern = use_pattern ? pat : NULL,
.sampled = 0,
.no_values = no_values,
.strlen = (isKeysHfield) ? hfieldlen : sdslen,
};
/* A pattern may restrict all matching keys to one cluster slot. */
@@ -1211,6 +1238,40 @@ void scanGenericCommand(client *c, robj *o, unsigned long long cursor) {
p = lpNext(o->ptr, p);
}
cursor = 0;
} else if (o->type == OBJ_HASH && o->encoding == OBJ_ENCODING_LISTPACK_EX) {
int64_t len;
long long expire_at;
unsigned char *lp = hashTypeListpackGetLp(o);
unsigned char *p = lpFirst(lp);
unsigned char *str, *val;
unsigned char intbuf[LP_INTBUF_SIZE];
while (p) {
str = lpGet(p, &len, intbuf);
p = lpNext(lp, p);
val = p; /* Keep pointer to value */
p = lpNext(lp, p);
serverAssert(p && lpGetIntegerValue(p, &expire_at));
if (hashTypeIsExpired(o, expire_at) ||
(use_pattern && !stringmatchlen(pat, sdslen(pat), (char *)str, len, 0)))
{
/* jump to the next key/val pair */
p = lpNext(lp, p);
continue;
}
/* add key object */
listAddNodeTail(keys, sdsnewlen(str, len));
/* add value object */
if (!no_values) {
str = lpGet(val, &len, intbuf);
listAddNodeTail(keys, sdsnewlen(str, len));
}
p = lpNext(lp, p);
}
cursor = 0;
} else {
serverPanic("Not handled encoding in SCAN.");
}
@@ -1243,10 +1304,14 @@ void scanGenericCommand(client *c, robj *o, unsigned long long cursor) {
addReplyArrayLen(c, 2);
addReplyBulkLongLong(c,cursor);
unsigned long long idx = 0;
addReplyArrayLen(c, listLength(keys));
while ((node = listFirst(keys)) != NULL) {
sds key = listNodeValue(node);
addReplyBulkCBuffer(c, key, sdslen(key));
void *key = listNodeValue(node);
/* For HSCAN, list will contain keys value pairs unless no_values arg
* was given. We should call mstrlen for the keys only. */
int hfieldkey = isKeysHfield && (no_values || (idx++ % 2 == 0));
addReplyBulkCBuffer(c, key, hfieldkey ? mstrlen(key) : sdslen(key));
listDelNode(keys, node);
}
@@ -1339,6 +1404,7 @@ void renameGenericCommand(client *c, int nx) {
robj *o;
long long expire;
int samekey = 0;
uint64_t minHashExpireTime = EB_EXPIRE_TIME_INVALID;
/* When source and dest key is the same, no operation is performed,
* if the key exists, however we still return an error on unexisting key. */
@@ -1364,9 +1430,21 @@ void renameGenericCommand(client *c, int nx) {
* with the same name. */
dbDelete(c->db,c->argv[2]);
}
dbAdd(c->db,c->argv[2],o);
dictEntry *de = dbAdd(c->db, c->argv[2], o);
if (expire != -1) setExpire(c,c->db,c->argv[2],expire);
/* If hash with expiration on fields then remove it from global HFE DS and
* keep next expiration time. Otherwise, dbDelete() will remove it from the
* global HFE DS and we will lose the expiration time. */
if (o->type == OBJ_HASH)
minHashExpireTime = hashTypeRemoveFromExpires(&c->db->hexpires, o);
dbDelete(c->db,c->argv[1]);
/* If hash with HFEs, register in db->hexpires */
if (minHashExpireTime != EB_EXPIRE_TIME_INVALID)
hashTypeAddToExpires(c->db, dictGetKey(de), o, minHashExpireTime);
signalModifiedKey(c,c->db,c->argv[1]);
signalModifiedKey(c,c->db,c->argv[2]);
notifyKeyspaceEvent(NOTIFY_GENERIC,"rename_from",
@@ -1390,6 +1468,7 @@ void moveCommand(client *c) {
redisDb *src, *dst;
int srcid, dbid;
long long expire;
uint64_t hashExpireTime = EB_EXPIRE_TIME_INVALID;
if (server.cluster_enabled) {
addReplyError(c,"MOVE is not allowed in cluster mode");
@@ -1430,12 +1509,25 @@ void moveCommand(client *c) {
addReply(c,shared.czero);
return;
}
dbAdd(dst,c->argv[1],o);
dictEntry *dstDictEntry = dbAdd(dst,c->argv[1],o);
if (expire != -1) setExpire(c,dst,c->argv[1],expire);
/* If hash with expiration on fields, remove it from global HFE DS and keep
* aside registered expiration time. Must be before deletion of the object.
* hexpires (ebuckets) embed in stored items its structure. */
if (o->type == OBJ_HASH)
hashExpireTime = hashTypeRemoveFromExpires(&src->hexpires, o);
incrRefCount(o);
/* OK! key moved, free the entry in the source DB */
dbDelete(src,c->argv[1]);
/* If object of type hash with expiration on fields. Taken care to add the
* hash to hexpires of `dst` only after dbDelete(). */
if (hashExpireTime != EB_EXPIRE_TIME_INVALID)
hashTypeAddToExpires(dst, dictGetKey(dstDictEntry), o, hashExpireTime);
signalModifiedKey(c,src,c->argv[1]);
signalModifiedKey(c,dst,c->argv[1]);
notifyKeyspaceEvent(NOTIFY_GENERIC,
@@ -1518,12 +1610,13 @@ void copyCommand(client *c) {
/* Duplicate object according to object's type. */
robj *newobj;
uint64_t minHashExpire = EB_EXPIRE_TIME_INVALID; /* HFE feature */
switch(o->type) {
case OBJ_STRING: newobj = dupStringObject(o); break;
case OBJ_LIST: newobj = listTypeDup(o); break;
case OBJ_SET: newobj = setTypeDup(o); break;
case OBJ_ZSET: newobj = zsetDup(o); break;
case OBJ_HASH: newobj = hashTypeDup(o); break;
case OBJ_HASH: newobj = hashTypeDup(o, newkey->ptr, &minHashExpire); break;
case OBJ_STREAM: newobj = streamDup(o); break;
case OBJ_MODULE:
newobj = moduleTypeDupOrReply(c, key, newkey, dst->id, o);
@@ -1538,8 +1631,16 @@ void copyCommand(client *c) {
dbDelete(dst,newkey);
}
dbAdd(dst,newkey,newobj);
if (expire != -1) setExpire(c, dst, newkey, expire);
dictEntry *deCopy = dbAdd(dst,newkey,newobj);
/* if key with expiration then set it */
if (expire != -1)
setExpire(c, dst, newkey, expire);
/* If minExpiredField was set, then the object is hash with expiration
* on fields and need to register it in global HFE DS */
if (minHashExpire != EB_EXPIRE_TIME_INVALID)
hashTypeAddToExpires(dst, dictGetKey(deCopy), newobj, minHashExpire);
/* OK! key copied */
signalModifiedKey(c,dst,c->argv[2]);
@@ -1629,11 +1730,13 @@ int dbSwapDatabases(int id1, int id2) {
* remain in the same DB they were. */
db1->keys = db2->keys;
db1->expires = db2->expires;
db1->hexpires = db2->hexpires;
db1->avg_ttl = db2->avg_ttl;
db1->expires_cursor = db2->expires_cursor;
db2->keys = aux.keys;
db2->expires = aux.expires;
db2->hexpires = aux.hexpires;
db2->avg_ttl = aux.avg_ttl;
db2->expires_cursor = aux.expires_cursor;
@@ -1671,11 +1774,13 @@ void swapMainDbWithTempDb(redisDb *tempDb) {
* remain in the same DB they were. */
activedb->keys = newdb->keys;
activedb->expires = newdb->expires;
activedb->hexpires = newdb->hexpires;
activedb->avg_ttl = newdb->avg_ttl;
activedb->expires_cursor = newdb->expires_cursor;
newdb->keys = aux.keys;
newdb->expires = aux.expires;
newdb->hexpires = aux.hexpires;
newdb->avg_ttl = aux.avg_ttl;
newdb->expires_cursor = aux.expires_cursor;
@@ -1864,7 +1969,7 @@ int keyIsExpired(redisDb *db, robj *key) {
* EXPIRE_AVOID_DELETE_EXPIRED flag.
*
* The return value of the function is KEY_VALID if the key is still valid.
* The function returns KEY_EXPIRED if the key is expired BUT not deleted,
* The function returns KEY_EXPIRED if the key is expired BUT not deleted,
* or returns KEY_DELETED if the key is expired and deleted. */
keyStatus expireIfNeeded(redisDb *db, robj *key, int flags) {
if (server.lazy_expire_disabled) return KEY_VALID;
@@ -1878,7 +1983,7 @@ keyStatus expireIfNeeded(redisDb *db, robj *key, int flags) {
* replicas.
*
* Still we try to return the right information to the caller,
* that is, KEY_VALID if we think the key should still be valid,
* that is, KEY_VALID if we think the key should still be valid,
* KEY_EXPIRED if we think the key is expired but don't want to delete it at this time.
*
* When replicating commands from the master, keys are never considered
src/debug.c
+16 -7
View File
@@ -200,17 +200,22 @@ void xorObjectDigest(redisDb *db, robj *keyobj, unsigned char *digest, robj *o)
}
} else if (o->type == OBJ_HASH) {
hashTypeIterator *hi = hashTypeInitIterator(o);
while (hashTypeNext(hi) != C_ERR) {
while (hashTypeNext(hi, 0) != C_ERR) {
unsigned char eledigest[20];
sds sdsele;
/* field */
memset(eledigest,0,20);
sdsele = hashTypeCurrentObjectNewSds(hi,OBJ_HASH_KEY);
mixDigest(eledigest,sdsele,sdslen(sdsele));
sdsfree(sdsele);
/* val */
sdsele = hashTypeCurrentObjectNewSds(hi,OBJ_HASH_VALUE);
mixDigest(eledigest,sdsele,sdslen(sdsele));
sdsfree(sdsele);
/* hash-field expiration (HFE) */
if (hi->expire_time != EB_EXPIRE_TIME_INVALID)
xorDigest(eledigest,"!!hexpire!!",11);
xorDigest(digest,eledigest,20);
}
hashTypeReleaseIterator(hi);
@@ -445,9 +450,9 @@ void debugCommand(client *c) {
"SEGFAULT",
" Crash the server with sigsegv.",
"SET-ACTIVE-EXPIRE <0|1>",
" Setting it to 0 disables expiring keys in background when they are not",
" accessed (otherwise the Redis behavior). Setting it to 1 reenables back the",
" default.",
" Setting it to 0 disables expiring keys (and hash-fields) in background ",
" when they are not accessed (otherwise the Redis behavior). Setting it",
" to 1 reenables back the default.",
"QUICKLIST-PACKED-THRESHOLD <size>",
" Sets the threshold for elements to be inserted as plain vs packed nodes",
" Default value is 1GB, allows values up to 4GB. Setting to 0 restores to default.",
@@ -664,10 +669,14 @@ NULL
if ((o = objectCommandLookupOrReply(c,c->argv[2],shared.nokeyerr))
== NULL) return;
if (o->encoding != OBJ_ENCODING_LISTPACK) {
if (o->encoding != OBJ_ENCODING_LISTPACK && o->encoding != OBJ_ENCODING_LISTPACK_EX) {
addReplyError(c,"Not a listpack encoded object.");
} else {
lpRepr(o->ptr);
if (o->encoding == OBJ_ENCODING_LISTPACK)
lpRepr(o->ptr);
else if (o->encoding == OBJ_ENCODING_LISTPACK_EX)
lpRepr(((listpackEx*)o->ptr)->lp);
addReplyStatus(c,"Listpack structure printed on stdout");
}
} else if (!strcasecmp(c->argv[1]->ptr,"quicklist") && (c->argc == 3 || c->argc == 4)) {
@@ -1081,7 +1090,7 @@ void serverLogObjectDebugInfo(const robj *o) {
} else if (o->type == OBJ_SET) {
serverLog(LL_WARNING,"Set size: %d", (int) setTypeSize(o));
} else if (o->type == OBJ_HASH) {
serverLog(LL_WARNING,"Hash size: %d", (int) hashTypeLength(o));
serverLog(LL_WARNING,"Hash size: %d", (int) hashTypeLength(o, 0));
} else if (o->type == OBJ_ZSET) {
serverLog(LL_WARNING,"Sorted set size: %d", (int) zsetLength(o));
if (o->encoding == OBJ_ENCODING_SKIPLIST)
src/defrag.c
+78 -6
View File
@@ -70,6 +70,22 @@ sds activeDefragSds(sds sdsptr) {
return NULL;
}
/* Defrag helper for hfield strings
*
* returns NULL in case the allocation wasn't moved.
* when it returns a non-null value, the old pointer was already released
* and should NOT be accessed. */
hfield activeDefragHfield(hfield hf) {
void *ptr = hfieldGetAllocPtr(hf);
void *newptr = activeDefragAlloc(ptr);
if (newptr) {
size_t offset = hf - (char*)ptr;
hf = (char*)newptr + offset;
return hf;
}
return NULL;
}
/* Defrag helper for robj and/or string objects with expected refcount.
*
* Like activeDefragStringOb, but it requires the caller to pass in the expected
@@ -250,6 +266,31 @@ void activeDefragSdsDictCallback(void *privdata, const dictEntry *de) {
UNUSED(de);
}
void activeDefragHfieldDictCallback(void *privdata, const dictEntry *de) {
dict *d = privdata;
hfield newhf, hf = dictGetKey(de);
if (hfieldGetExpireTime(hf) == EB_EXPIRE_TIME_INVALID) {
/* If the hfield does not have TTL, we directly defrag it. */
newhf = activeDefragHfield(hf);
} else {
/* Update its reference in the ebucket while defragging it. */
ebuckets *eb = hashTypeGetDictMetaHFE(d);
newhf = ebDefragItem(eb, &hashFieldExpireBucketsType, hf, (ebDefragFunction *)activeDefragHfield);
}
if (newhf) {
/* We can't search in dict for that key after we've released
* the pointer it holds, since it won't be able to do the string
* compare, but we can find the entry using key hash and pointer. */
dictUseStoredKeyApi(d, 1);
uint64_t hash = dictGetHash(d, newhf);
dictUseStoredKeyApi(d, 0);
dictEntry *de = dictFindEntryByPtrAndHash(d, hf, hash);
serverAssert(de);
dictSetKey(d, de, newhf);
}
}
/* Defrag a dict with sds key and optional value (either ptr, sds or robj string) */
void activeDefragSdsDict(dict* d, int val_type) {
unsigned long cursor = 0;
@@ -268,6 +309,20 @@ void activeDefragSdsDict(dict* d, int val_type) {
} while (cursor != 0);
}
/* Defrag a dict with hfield key and sds value. */
void activeDefragHfieldDict(dict *d) {
unsigned long cursor = 0;
dictDefragFunctions defragfns = {
.defragAlloc = activeDefragAlloc,
.defragKey = NULL, /* Will be defragmented in activeDefragHfieldDictCallback. */
.defragVal = (dictDefragAllocFunction *)activeDefragSds
};
do {
cursor = dictScanDefrag(d, cursor, activeDefragHfieldDictCallback,
&defragfns, d);
} while (cursor != 0);
}
/* Defrag a list of ptr, sds or robj string values */
void activeDefragList(list *l, int val_type) {
listNode *ln, *newln;
@@ -422,10 +477,10 @@ void scanLaterHash(robj *ob, unsigned long *cursor) {
dict *d = ob->ptr;
dictDefragFunctions defragfns = {
.defragAlloc = activeDefragAlloc,
.defragKey = (dictDefragAllocFunction *)activeDefragSds,
.defragKey = NULL, /* Will be defragmented in activeDefragHfieldDictCallback. */
.defragVal = (dictDefragAllocFunction *)activeDefragSds
};
*cursor = dictScanDefrag(d, *cursor, scanCallbackCountScanned, &defragfns, NULL);
*cursor = dictScanDefrag(d, *cursor, activeDefragHfieldDictCallback, &defragfns, d);
}
void defragQuicklist(redisDb *db, dictEntry *kde) {
@@ -477,7 +532,7 @@ void defragHash(redisDb *db, dictEntry *kde) {
if (dictSize(d) > server.active_defrag_max_scan_fields)
defragLater(db, kde);
else
activeDefragSdsDict(d, DEFRAG_SDS_DICT_VAL_IS_SDS);
activeDefragHfieldDict(d);
/* defrag the dict struct and tables */
if ((newd = dictDefragTables(ob->ptr)))
ob->ptr = newd;
@@ -672,7 +727,7 @@ void defragModule(redisDb *db, dictEntry *kde) {
* all the various pointers it has. */
void defragKey(defragCtx *ctx, dictEntry *de) {
sds keysds = dictGetKey(de);
robj *newob, *ob;
robj *newob, *ob = dictGetVal(de);
unsigned char *newzl;
sds newsds;
redisDb *db = ctx->privdata;
@@ -689,11 +744,22 @@ void defragKey(defragCtx *ctx, dictEntry *de) {
dictEntry *expire_de = kvstoreDictFindEntryByPtrAndHash(db->expires, slot, keysds, hash);
if (expire_de) kvstoreDictSetKey(db->expires, slot, expire_de, newsds);
}
/* Update the key's reference in the dict's metadata or the listpackEx. */
if (unlikely(ob->type == OBJ_HASH))
hashTypeUpdateKeyRef(ob, newsds);
}
/* Try to defrag robj and / or string value. */
ob = dictGetVal(de);
if ((newob = activeDefragStringOb(ob))) {
if (unlikely(ob->type == OBJ_HASH && hashTypeGetMinExpire(ob) != EB_EXPIRE_TIME_INVALID)) {
/* Update its reference in the ebucket while defragging it. */
newob = ebDefragItem(&db->hexpires, &hashExpireBucketsType, ob,
(ebDefragFunction *)activeDefragStringOb);
} else {
/* If the dict doesn't have metadata, we directly defrag it. */
newob = activeDefragStringOb(ob);
}
if (newob) {
kvstoreDictSetVal(db->keys, slot, de, newob);
ob = newob;
}
@@ -734,6 +800,12 @@ void defragKey(defragCtx *ctx, dictEntry *de) {
if (ob->encoding == OBJ_ENCODING_LISTPACK) {
if ((newzl = activeDefragAlloc(ob->ptr)))
ob->ptr = newzl;
} else if (ob->encoding == OBJ_ENCODING_LISTPACK_EX) {
listpackEx *newlpt, *lpt = (listpackEx*)ob->ptr;
if ((newlpt = activeDefragAlloc(lpt)))
ob->ptr = lpt = newlpt;
if ((newzl = activeDefragAlloc(lpt->lp)))
lpt->lp = newzl;
} else if (ob->encoding == OBJ_ENCODING_HT) {
defragHash(db, de);
} else {
src/dict.c
+59 -10
View File
@@ -67,6 +67,25 @@ static int _dictInit(dict *d, dictType *type);
static dictEntry *dictGetNext(const dictEntry *de);
static dictEntry **dictGetNextRef(dictEntry *de);
static void dictSetNext(dictEntry *de, dictEntry *next);
static int dictDefaultCompare(dict *d, const void *key1, const void *key2);
/* -------------------------- misc inline functions -------------------------------- */
typedef int (*keyCmpFunc)(dict *d, const void *key1, const void *key2);
static inline keyCmpFunc dictGetKeyCmpFunc(dict *d) {
if (d->useStoredKeyApi && d->type->storedKeyCompare)
return d->type->storedKeyCompare;
if (d->type->keyCompare)
return d->type->keyCompare;
return dictDefaultCompare;
}
static inline uint64_t dictHashKey(dict *d, const void *key, int isStoredKey) {
if (isStoredKey && d->type->storedHashFunction)
return d->type->storedHashFunction(key);
else
return d->type->hashFunction(key);
}
/* -------------------------- hash functions -------------------------------- */
@@ -173,6 +192,19 @@ dict *dictCreate(dictType *type)
return d;
}
/* Change dictType of dict to another one with metadata support
* Rest of dictType's values must stay the same */
void dictTypeAddMeta(dict **d, dictType *typeWithMeta) {
/* Verify new dictType is compatible with the old one */
dictType toCmp = *typeWithMeta;
toCmp.dictMetadataBytes = NULL; /* Expected old one not to have metadata */
toCmp.onDictRelease = (*d)->type->onDictRelease; /* Ignore 'onDictRelease' in comparison */
assert(memcmp((*d)->type, &toCmp, sizeof(dictType)) == 0); /* The rest of the dictType fields must be the same */
*d = zrealloc(*d, sizeof(dict) + typeWithMeta->dictMetadataBytes(*d));
(*d)->type = typeWithMeta;
}
/* Initialize the hash table */
int _dictInit(dict *d, dictType *type)
{
@@ -182,6 +214,7 @@ int _dictInit(dict *d, dictType *type)
d->rehashidx = -1;
d->pauserehash = 0;
d->pauseAutoResize = 0;
d->useStoredKeyApi = 0;
return DICT_OK;
}
@@ -285,7 +318,7 @@ static void rehashEntriesInBucketAtIndex(dict *d, uint64_t idx) {
void *key = dictGetKey(de);
/* Get the index in the new hash table */
if (d->ht_size_exp[1] > d->ht_size_exp[0]) {
h = dictHashKey(d, key) & DICTHT_SIZE_MASK(d->ht_size_exp[1]);
h = dictHashKey(d, key, 1) & DICTHT_SIZE_MASK(d->ht_size_exp[1]);
} else {
/* We're shrinking the table. The tables sizes are powers of
* two, so we simply mask the bucket index in the larger table
@@ -572,7 +605,7 @@ static dictEntry *dictGenericDelete(dict *d, const void *key, int nofree) {
/* dict is empty */
if (dictSize(d) == 0) return NULL;
h = dictHashKey(d, key);
h = dictHashKey(d, key, d->useStoredKeyApi);
idx = h & DICTHT_SIZE_MASK(d->ht_size_exp[0]);
if (dictIsRehashing(d)) {
@@ -587,6 +620,8 @@ static dictEntry *dictGenericDelete(dict *d, const void *key, int nofree) {
}
}
keyCmpFunc cmpFunc = dictGetKeyCmpFunc(d);
for (table = 0; table <= 1; table++) {
if (table == 0 && (long)idx < d->rehashidx) continue;
idx = h & DICTHT_SIZE_MASK(d->ht_size_exp[table]);
@@ -594,7 +629,7 @@ static dictEntry *dictGenericDelete(dict *d, const void *key, int nofree) {
prevHe = NULL;
while(he) {
void *he_key = dictGetKey(he);
if (key == he_key || dictCompareKeys(d, key, he_key)) {
if (key == he_key || cmpFunc(d, key, he_key)) {
/* Unlink the element from the list */
if (prevHe)
dictSetNext(prevHe, dictGetNext(he));
@@ -689,6 +724,10 @@ void dictRelease(dict *d)
* destroying the dict fake completion. */
if (dictIsRehashing(d) && d->type->rehashingCompleted)
d->type->rehashingCompleted(d);
if (d->type->onDictRelease)
d->type->onDictRelease(d);
_dictClear(d,0,NULL);
_dictClear(d,1,NULL);
zfree(d);
@@ -701,8 +740,9 @@ dictEntry *dictFind(dict *d, const void *key)
if (dictSize(d) == 0) return NULL; /* dict is empty */
h = dictHashKey(d, key);
h = dictHashKey(d, key, d->useStoredKeyApi);
idx = h & DICTHT_SIZE_MASK(d->ht_size_exp[0]);
keyCmpFunc cmpFunc = dictGetKeyCmpFunc(d);
if (dictIsRehashing(d)) {
if ((long)idx >= d->rehashidx && d->ht_table[0][idx]) {
@@ -722,7 +762,7 @@ dictEntry *dictFind(dict *d, const void *key)
he = d->ht_table[table][idx];
while(he) {
void *he_key = dictGetKey(he);
if (key == he_key || dictCompareKeys(d, key, he_key))
if (key == he_key || cmpFunc(d, key, he_key))
return he;
he = dictGetNext(he);
}
@@ -759,7 +799,9 @@ dictEntry *dictTwoPhaseUnlinkFind(dict *d, const void *key, dictEntry ***plink,
if (dictSize(d) == 0) return NULL; /* dict is empty */
if (dictIsRehashing(d)) _dictRehashStep(d);
h = dictHashKey(d, key);
h = dictHashKey(d, key, d->useStoredKeyApi);
keyCmpFunc cmpFunc = dictGetKeyCmpFunc(d);
for (table = 0; table <= 1; table++) {
idx = h & DICTHT_SIZE_MASK(d->ht_size_exp[table]);
@@ -767,7 +809,7 @@ dictEntry *dictTwoPhaseUnlinkFind(dict *d, const void *key, dictEntry ***plink,
dictEntry **ref = &d->ht_table[table][idx];
while (ref && *ref) {
void *de_key = dictGetKey(*ref);
if (key == de_key || dictCompareKeys(d, key, de_key)) {
if (key == de_key || cmpFunc(d, key, de_key)) {
*table_index = table;
*plink = ref;
dictPauseRehashing(d);
@@ -1530,8 +1572,8 @@ static signed char _dictNextExp(unsigned long size)
void *dictFindPositionForInsert(dict *d, const void *key, dictEntry **existing) {
unsigned long idx, table;
dictEntry *he;
uint64_t hash = dictHashKey(d, key, d->useStoredKeyApi);
if (existing) *existing = NULL;
uint64_t hash = dictHashKey(d, key);
idx = hash & DICTHT_SIZE_MASK(d->ht_size_exp[0]);
if (dictIsRehashing(d)) {
@@ -1548,6 +1590,8 @@ void *dictFindPositionForInsert(dict *d, const void *key, dictEntry **existing)
/* Expand the hash table if needed */
_dictExpandIfNeeded(d);
keyCmpFunc cmpFunc = dictGetKeyCmpFunc(d);
for (table = 0; table <= 1; table++) {
if (table == 0 && (long)idx < d->rehashidx) continue;
idx = hash & DICTHT_SIZE_MASK(d->ht_size_exp[table]);
@@ -1555,7 +1599,7 @@ void *dictFindPositionForInsert(dict *d, const void *key, dictEntry **existing)
he = d->ht_table[table][idx];
while(he) {
void *he_key = dictGetKey(he);
if (key == he_key || dictCompareKeys(d, key, he_key)) {
if (key == he_key || cmpFunc(d, key, he_key)) {
if (existing) *existing = he;
return NULL;
}
@@ -1587,7 +1631,7 @@ void dictSetResizeEnabled(dictResizeEnable enable) {
}
uint64_t dictGetHash(dict *d, const void *key) {
return dictHashKey(d, key);
return dictHashKey(d, key, d->useStoredKeyApi);
}
/* Finds the dictEntry using pointer and pre-calculated hash.
@@ -1732,6 +1776,11 @@ void dictGetStats(char *buf, size_t bufsize, dict *d, int full) {
orig_buf[orig_bufsize-1] = '\0';
}
static int dictDefaultCompare(dict *d, const void *key1, const void *key2) {
(void)(d); /*unused*/
return key1 == key2;
}
/* ------------------------------- Benchmark ---------------------------------*/
#ifdef REDIS_TEST
src/dict.h
+31 -2
View File
@@ -62,6 +62,32 @@ typedef struct dictType {
unsigned int keys_are_odd:1;
/* TODO: Add a 'keys_are_even' flag and use a similar optimization if that
* flag is set. */
/* Sometimes we want the ability to store a key in a given way inside the hash
* function, and lookup it in some other way without resorting to any kind of
* conversion. For instance the key may be stored as a structure also
* representing other things, but the lookup happens via just a pointer to a
* null terminated string. Optionally providing additional hash/cmp functions,
* dict supports such usage. In that case we'll have a hashFunction() that will
* expect a null terminated C string, and a storedHashFunction() that will
* instead expect the structure. Similarly, the two comparison functions will
* work differently. The keyCompare() will treat the first argument as a pointer
* to a C string and the other as a structure (this way we can directly lookup
* the structure key using the C string). While the storedKeyCompare() will
* check if two pointers to the key in structure form are the same.
*
* However, functions of dict that gets key as argument (void *key) don't get
* any indication whether it is a lookup or stored key. To indicate that
* you intend to use key of type stored-key, and, consequently, use
* dedicated compare and hash functions of stored-key, is by calling
* dictUseStoredKeyApi(1) before using any of the dict functions that gets
* key as a parameter and then call again dictUseStoredKeyApi(0) once done.
*
* Set to NULL both functions, if you don't want to support this feature. */
uint64_t (*storedHashFunction)(const void *key);
int (*storedKeyCompare)(dict *d, const void *key1, const void *key2);
/* Optional callback called when the dict is destroyed. */
void (*onDictRelease)(dict *d);
} dictType;
#define DICTHT_SIZE(exp) ((exp) == -1 ? 0 : (unsigned long)1<<(exp))
@@ -76,7 +102,9 @@ struct dict {
long rehashidx; /* rehashing not in progress if rehashidx == -1 */
/* Keep small vars at end for optimal (minimal) struct padding */
int16_t pauserehash; /* If >0 rehashing is paused (<0 indicates coding error) */
unsigned pauserehash : 15; /* If >0 rehashing is paused */
unsigned useStoredKeyApi : 1; /* See comment of storedHashFunction above */
signed char ht_size_exp[2]; /* exponent of size. (size = 1<<exp) */
int16_t pauseAutoResize; /* If >0 automatic resizing is disallowed (<0 indicates coding error) */
void *metadata[];
@@ -136,7 +164,6 @@ typedef struct {
#define dictMetadataSize(d) ((d)->type->dictMetadataBytes \
? (d)->type->dictMetadataBytes(d) : 0)
#define dictHashKey(d, key) ((d)->type->hashFunction(key))
#define dictBuckets(d) (DICTHT_SIZE((d)->ht_size_exp[0])+DICTHT_SIZE((d)->ht_size_exp[1]))
#define dictSize(d) ((d)->ht_used[0]+(d)->ht_used[1])
#define dictIsEmpty(d) ((d)->ht_used[0] == 0 && (d)->ht_used[1] == 0)
@@ -146,6 +173,7 @@ typedef struct {
#define dictIsRehashingPaused(d) ((d)->pauserehash > 0)
#define dictPauseAutoResize(d) ((d)->pauseAutoResize++)
#define dictResumeAutoResize(d) ((d)->pauseAutoResize--)
#define dictUseStoredKeyApi(d, flag) ((d)->useStoredKeyApi = (flag))
/* If our unsigned long type can store a 64 bit number, use a 64 bit PRNG. */
#if ULONG_MAX >= 0xffffffffffffffff
@@ -162,6 +190,7 @@ typedef enum {
/* API */
dict *dictCreate(dictType *type);
void dictTypeAddMeta(dict **d, dictType *typeWithMeta);
int dictExpand(dict *d, unsigned long size);
int dictTryExpand(dict *d, unsigned long size);
int dictShrink(dict *d, unsigned long size);
src/ebuckets.c
+2422
View File
File diff suppressed because it is too large Load Diff
src/ebuckets.h
+308
View File
@@ -0,0 +1,308 @@
/*
* Copyright Redis Ltd. 2024 - present
*
* Licensed under your choice of the Redis Source Available License 2.0 (RSALv2)
* or the Server Side Public License v1 (SSPLv1).
*
*
* WHAT IS EBUCKETS?
* -----------------
* ebuckets is being used to store items that are set with expiration-time. It
* supports the basic API of add, remove and active expiration. The implementation
* of it is based on rax-tree, or plain linked-list when small. The expiration time
* of the items are used as the key to traverse rax-tree.
*
* Instead of holding a distinct item in each leaf of the rax-tree we can aggregate
* items into small segments and hold it in each leaf. This way we can avoid
* frequent modification of the rax-tree, since many of the modifications
* will be done only at the segment level. It will also save memory because
* rax-tree can be costly, around 40 bytes per leaf (with rax-key limited to 6
* bytes). Whereas each additional item in the segment will cost the size of the
* 'next' pointer in a list (8 bytes) and few more bytes for maintenance of the
* segment.
*
* EBUCKETS STRUCTURE
* ------------------
* The ebuckets data structure is organized in a hierarchical manner as follows:
*
* 1. ebuckets: This is the top-level data structure. It can be either a rax tree
* or a plain linked list. It contains one or more buckets, each representing
* an interval in time.
*
* 2. bucket: Each bucket represents an interval in time and contains one or more
* segments. The key in the rax-tree for each bucket represents low
* bound expiration-time for the items within this bucket. The key of the
* following bucket represents the upper bound expiration-time.
*
* 3. segment: Each segment within a bucket can hold up to `EB_SEG_MAX_ITEMS`
* items as a linked list. If there are more, the segment will try to
* split the bucket. To avoid wasting memory, it is a singly linked list (only
* next-item pointer). It is a cyclic linked-list to allow efficient removal of
* items from the middle of the segment without traversing the rax tree.
*
* 4. item: Each item that is stored in ebuckets should embed the ExpireMeta
* struct and supply getter function (see EbucketsType.getExpireMeta). This
* struct holds the expire-time of the item and few more fields that are used
* to maintain the segments data-structure.
*
* SPLITTING BUCKET
* ----------------
* Each segment can hold up-to `EB_SEG_MAX_ITEMS` items. On insertion of new
* item, it will try to split the segment. Here is an example For adding item
* with expiration of 42 to a segment that already reached its maximum capacity
* which will cause to split of the segment and in turn split of the bucket as
* well to a finer grained ranges:
*
* BUCKETS BUCKETS
* [ 00-10 ] -> size(Seg0) = 11 ==> [ 00-10 ] -> size(Seg0) = 11
* [ 11-76 ] -> size(Seg1) = 16 [ 11-36 ] -> size(Seg1) = 9
* [ 37-76 ] -> size(Seg2) = 7
*
* EXTENDING BUCKET
* ----------------
* In the example above, the reason it wasn't split evenly is that Seg1 must have
* been holding items with same TTL and they must reside together in the same
* bucket after the split. Which brings us to another important point. If there
* is a segment that reached its maximum capacity and all the items have same
* expiration-time key, then we cannot split the bucket but aggregate all the
* items, with same expiration time key, by allocating an extended-segment and
* chain it to the first segment in visited bucket. In that sense, extended
* segments will only hold items with same expiration-time key.
*
* BUCKETS BUCKETS
* [ 00-10 ] -> size(Seg0)=11 ==> [ 00-10 ] -> size(Seg0)=11
* [ 11-12 ] -> size(Seg1)=16 [ 11-12 ] -> size(Seg1)=1 -> size(Seg2)=16
*
* LIMITING RAX TREE DEPTH
* -----------------------
* The rax tree is basically a B-tree and its depth is bounded by the sizeof of
* the key. Holding 6 bytes for expiration-time key is more than enough to represent
* unix-time in msec, and in turn the depth of the tree is limited to 6 levels.
* At a first glance it might look sufficient but we need take into consideration
* the heavyweight maintenance and traversal of each node in the B-tree.
*
* And so, we can further prune the tree such that holding keys with msec precision
* in the tree doesn't bring with it much value. The active-expiration operation can
* live with deletion of expired items, say, older than 1 sec, which means the size
* of time-expiration keys to the rax tree become no more than ~4.5 bytes and we
* also get rid of the "noisy" bits which most probably will cause to yet another
* branching and modification of the rax tree in case of items with time-expiration
* difference of less than 1 second. The lazy expiration will still be precise and
* without compromise on accuracy because the exact expiration-time is kept
* attached as well to each item, in `ExpireMeta`, and each traversal of item with
* expiration will behave as expected down to the msec. Take care to configure
* `EB_BUCKET_KEY_PRECISION` according to your needs.
*
* EBUCKET KEY
* -----------
* Taking into account configured value of `EB_BUCKET_KEY_PRECISION`, two items
* with expiration-time t1 and t2 will be considered to have the same key in the
* rax-tree/buckets if and only if:
*
* EB_BUCKET_KEY(t1) == EB_BUCKET_KEY(t2)
*
* EBUCKETS CREATION
* -----------------
* To avoid the cost of allocating rax data-structure for only few elements,
* ebuckets will start as a simple linked-list and only when it reaches some
* threshold, it will be converted to rax.
*
* TODO
* ----
* - ebRemove() optimize to merge small segments into one segment.
* - ebAdd() Fix pathological case of cascade addition of items into rax such
* that their values are smaller/bigger than visited extended-segment which ends
* up with multiple segments with a single item in each segment.
*/
#ifndef __EBUCKETS_H
#define __EBUCKETS_H
#include <stdlib.h>
#include <sys/types.h>
#include <stdarg.h>
#include <stdint.h>
#include "rax.h"
/*
* EB_BUCKET_KEY_PRECISION - Defines the number of bits to ignore from the
* expiration-time when mapping to buckets. The higher the value, the more items
* with similar expiration-time will be aggregated into the same bucket. The lower
* the value, the more "accurate" the active expiration of buckets will be.
*
* Note that the accurate time expiration of each item is preserved anyway and
* enforced by lazy expiration. It only impacts the active expiration that will
* be able to work on buckets older than (1<<EB_BUCKET_KEY_PRECISION) msec ago.
* For example if EB_BUCKET_KEY_PRECISION is 10, then active expiration
* will work only on buckets that already got expired at least 1sec ago.
*
* The idea of it is to trim the rax tree depth, avoid having too many branches,
* and reduce frequent modifications of the tree to the minimum.
*/
#define EB_BUCKET_KEY_PRECISION 0 /* TBD: modify to 10 */
/* From expiration time to bucket-key */
#define EB_BUCKET_KEY(exptime) ((exptime) >> EB_BUCKET_KEY_PRECISION)
#define EB_EXPIRE_TIME_MAX ((uint64_t)0x0000FFFFFFFFFFFF) /* Maximum expire-time. */
#define EB_EXPIRE_TIME_INVALID (EB_EXPIRE_TIME_MAX+1) /* assumed bigger than max */
/* Handler to ebuckets DS. Pointer to a list, rax or NULL (empty DS). See also ebIsList(). */
typedef void *ebuckets;
/* Users of ebuckets will store `eItem` which is just a void pointer to their
* element. In addition, eItem should embed the ExpireMeta struct and supply
* getter function (see EbucketsType.getExpireMeta).
*/
typedef void *eItem;
/* This struct Should be embedded inside `eItem` and must be aligned in memory. */
typedef struct ExpireMeta {
/* 48bits of unix-time in msec. This value is sufficient to represent, in
* unix-time, until the date of 02 August, 10889
*/
uint32_t expireTimeLo; /* Low bits of expireTime. */
uint16_t expireTimeHi; /* High bits of expireTime. */
unsigned int lastInSegment : 1; /* Last item in segment. If set, then 'next' will
point to the NextSegHdr, unless lastItemBucket=1
then it will point to segment header of the
current segment. */
unsigned int firstItemBucket : 1; /* First item in bucket. This flag assist
to manipulate segments directly without
the need to traverse from start the
rax tree */
unsigned int lastItemBucket : 1; /* Last item in bucket. This flag assist
to manipulate segments directly without
the need to traverse from start the
rax tree */
unsigned int numItems : 5; /* Only first item in segment will maintain
this value. */
unsigned int trash : 1; /* This flag indicates whether the ExpireMeta
associated with the item is leftover.
There is always a potential to reuse the
item after removal/deletion. Note that,
the user can still safely O(1) TTL lookup
a given item and verify whether attached
TTL is valid or leftover. See function
ebGetExpireTime(). */
unsigned int userData : 3; /* ebuckets can be used to store in same
instance few different types of items,
such as, listpack and hash. This field
is reserved to store such identification
associated with the item and can help
to distinct on delete or expire callback.
It is not used by ebuckets internally and
should be maintained by the user */
unsigned int reserved : 4;
void *next; /* - If not last item in segment then next
points to next eItem (lastInSegment=0).
- If last in segment but not last in
bucket (lastItemBucket=0) then it
points to next segment header.
- If last in bucket then it points to
current segment header (Can be either
of type FirstSegHdr or NextSegHdr). */
} ExpireMeta;
/* Each instance of ebuckets need to have corresponding EbucketsType that holds
* the necessary callbacks and configuration to operate correctly on the type
* of items that are stored in it. Conceptually it should have hold reference
* from ebuckets instance to this type, but to save memory we will pass it as
* an argument to each API call. */
typedef struct EbucketsType {
/* getter to extract the ExpireMeta from the item */
ExpireMeta* (*getExpireMeta)(const eItem item);
/* Called during ebDestroy(). Set to NULL if not needed. */
void (*onDeleteItem)(eItem item, void *ctx);
/* Is addresses of items are odd in memory. It is taken into consideration
* and used by ebuckets to know how to distinct between ebuckets pointer to
* rax versus a pointer to item which is head of list. */
unsigned int itemsAddrAreOdd;
} EbucketsType;
/* Returned value by `onExpireItem` callback to indicate the action to be taken by
* ebExpire(). */
typedef enum ExpireAction {
ACT_REMOVE_EXP_ITEM=0, /* Remove the item from ebuckets. */
ACT_UPDATE_EXP_ITEM, /* Re-insert the item with updated expiration-time.
Before returning this value, the cb need to
update expiration time of the item by assisting
function ebSetMetaExpTime(). The item will be
kept aside and will be added again to ebuckets
at the end of ebExpire() */
ACT_STOP_ACTIVE_EXP /* Stop active-expiration. It will assume that
provided 'item' wasn't deleted by the callback. */
} ExpireAction;
/* ExpireInfo is used to pass input and output parameters to ebExpire(). */
typedef struct ExpireInfo {
/* onExpireItem - Called during active-expiration by ebExpire() */
ExpireAction (*onExpireItem)(eItem item, void *ctx);
uint64_t maxToExpire; /* [INPUT ] Limit of number expired items to scan */
void *ctx; /* [INPUT ] context to pass to onExpireItem */
uint64_t now; /* [INPUT ] Current time in msec. */
uint64_t nextExpireTime; /* [OUTPUT] Next expiration time. Return 0, if none left. */
/* TODO: Distinct between expired & updated */
uint64_t itemsExpired; /* [OUTPUT] Returns the number of expired or updated items. */
} ExpireInfo;
/* ebuckets API */
static inline ebuckets ebCreate(void) { return NULL; } /* Empty ebuckets */
void ebDestroy(ebuckets *eb, EbucketsType *type, void *deletedItemsCbCtx);
void ebExpire(ebuckets *eb, EbucketsType *type, ExpireInfo *info);
uint64_t ebExpireDryRun(ebuckets eb, EbucketsType *type, uint64_t now);
static inline int ebIsEmpty(ebuckets eb) { return eb == NULL; }
uint64_t ebGetNextTimeToExpire(ebuckets eb, EbucketsType *type);
uint64_t ebGetMaxExpireTime(ebuckets eb, EbucketsType *type, int accurate);
uint64_t ebGetTotalItems(ebuckets eb, EbucketsType *type);
/* Item related API */
int ebRemove(ebuckets *eb, EbucketsType *type, eItem item);
int ebAdd(ebuckets *eb, EbucketsType *type, eItem item, uint64_t expireTime);
uint64_t ebGetExpireTime(EbucketsType *type, eItem item);
typedef eItem (ebDefragFunction)(const eItem item);
eItem ebDefragItem(ebuckets *eb, EbucketsType *type, eItem item, ebDefragFunction *fn);
static inline uint64_t ebGetMetaExpTime(ExpireMeta *expMeta) {
return (((uint64_t)(expMeta)->expireTimeHi << 32) | (expMeta)->expireTimeLo);
}
static inline void ebSetMetaExpTime(ExpireMeta *expMeta, uint64_t t) {
expMeta->expireTimeLo = (uint32_t)(t&0xFFFFFFFF);
expMeta->expireTimeHi = (uint16_t)((t) >> 32);
}
/* Debug API */
void ebValidate(ebuckets eb, EbucketsType *type);
void ebPrint(ebuckets eb, EbucketsType *type);
#ifdef REDIS_TEST
int ebucketsTest(int argc, char *argv[], int flags);
#endif
#endif /* __EBUCKETS_H */
src/expire.c
+53
View File
@@ -94,6 +94,7 @@ int activeExpireCycleTryExpire(redisDb *db, dictEntry *de, long long now) {
#define ACTIVE_EXPIRE_CYCLE_SLOW_TIME_PERC 25 /* Max % of CPU to use. */
#define ACTIVE_EXPIRE_CYCLE_ACCEPTABLE_STALE 10 /* % of stale keys after which
we do extra efforts. */
#define HFE_ACTIVE_EXPIRE_CYCLE_FIELDS 1000
/* Data used by the expire dict scan callback. */
typedef struct {
@@ -134,6 +135,53 @@ static inline int isExpiryDictValidForSamplingCb(dict *d) {
return C_OK;
}
/* Active expiration Cycle for hash-fields.
*
* Note that releasing fields is expected to be more predictable and rewarding
* than releasing keys because it is stored in `ebuckets` DS which optimized for
* active expiration and in addition the deletion of fields is simple to handle. */
static inline void activeExpireHashFieldCycle(int type) {
/* Remember current db across calls */
static unsigned int currentDb = 0;
/* Tracks the count of fields actively expired for the current database.
* This count continues as long as it fails to actively expire all expired
* fields of currentDb, indicating a possible need to adjust the value of
* maxToExpire. */
static uint64_t activeExpirySequence = 0;
/* Threshold for adjusting maxToExpire */
const uint32_t EXPIRED_FIELDS_TH = 1000000;
/* Maximum number of fields to actively expire in a single call */
uint32_t maxToExpire = HFE_ACTIVE_EXPIRE_CYCLE_FIELDS;
redisDb *db = server.db + currentDb;
/* If db is empty, move to next db and return */
if (ebIsEmpty(db->hexpires)) {
activeExpirySequence = 0;
currentDb = (currentDb + 1) % server.dbnum;
return;
}
/* If running for a while and didn't manage to active-expire all expired fields of
* currentDb (i.e. activeExpirySequence becomes significant) then adjust maxToExpire */
if ((activeExpirySequence > EXPIRED_FIELDS_TH) && (type == ACTIVE_EXPIRE_CYCLE_SLOW)) {
/* maxToExpire is multiplied by a factor between 1 and 32, proportional to
* the number of times activeExpirySequence exceeded EXPIRED_FIELDS_TH */
uint64_t factor = activeExpirySequence / EXPIRED_FIELDS_TH;
maxToExpire *= (factor<32) ? factor : 32;
}
if (hashTypeDbActiveExpire(db, maxToExpire) == maxToExpire) {
/* active-expire reached maxToExpire limit */
activeExpirySequence += maxToExpire;
} else {
/* Managed to active-expire all expired fields of currentDb */
activeExpirySequence = 0;
currentDb = (currentDb + 1) % server.dbnum;
}
}
void activeExpireCycle(int type) {
/* Adjust the running parameters according to the configured expire
* effort. The default effort is 1, and the maximum configurable effort
@@ -232,6 +280,11 @@ void activeExpireCycle(int type) {
* distribute the time evenly across DBs. */
current_db++;
/* Interleaving hash-field expiration with key expiration. Better
* call it before handling expired keys because HFE DS is optimized for
* active expiration */
activeExpireHashFieldCycle(type);
if (kvstoreSize(db->expires))
dbs_performed++;
src/lazyfree.c
+6 -2
View File
@@ -3,6 +3,7 @@
#include "atomicvar.h"
#include "functions.h"
#include "cluster.h"
#include "ebuckets.h"
static redisAtomic size_t lazyfree_objects = 0;
static redisAtomic size_t lazyfreed_objects = 0;
@@ -22,7 +23,8 @@ void lazyfreeFreeObject(void *args[]) {
void lazyfreeFreeDatabase(void *args[]) {
kvstore *da1 = args[0];
kvstore *da2 = args[1];
ebuckets oldHfe = args[2];
ebDestroy(&oldHfe, &hashExpireBucketsType, NULL);
size_t numkeys = kvstoreSize(da1);
kvstoreRelease(da1);
kvstoreRelease(da2);
@@ -201,10 +203,12 @@ void emptyDbAsync(redisDb *db) {
flags |= KVSTORE_FREE_EMPTY_DICTS;
}
kvstore *oldkeys = db->keys, *oldexpires = db->expires;
ebuckets oldHfe = db->hexpires;
db->keys = kvstoreCreate(&dbDictType, slot_count_bits, flags);
db->expires = kvstoreCreate(&dbExpiresDictType, slot_count_bits, flags);
db->hexpires = ebCreate();
atomicIncr(lazyfree_objects, kvstoreSize(oldkeys));
bioCreateLazyFreeJob(lazyfreeFreeDatabase, 2, oldkeys, oldexpires);
bioCreateLazyFreeJob(lazyfreeFreeDatabase, 3, oldkeys, oldexpires, oldHfe);
}
/* Free the key tracking table.
src/listpack.c
+614 -102
View File
@@ -245,51 +245,61 @@ unsigned char* lpShrinkToFit(unsigned char *lp) {
static inline void lpEncodeIntegerGetType(int64_t v, unsigned char *intenc, uint64_t *enclen) {
if (v >= 0 && v <= 127) {
/* Single byte 0-127 integer. */
intenc[0] = v;
*enclen = 1;
if (intenc != NULL) intenc[0] = v;
if (enclen != NULL) *enclen = 1;
} else if (v >= -4096 && v <= 4095) {
/* 13 bit integer. */
if (v < 0) v = ((int64_t)1<<13)+v;
intenc[0] = (v>>8)|LP_ENCODING_13BIT_INT;
intenc[1] = v&0xff;
*enclen = 2;
if (intenc != NULL) {
intenc[0] = (v>>8)|LP_ENCODING_13BIT_INT;
intenc[1] = v&0xff;
}
if (enclen != NULL) *enclen = 2;
} else if (v >= -32768 && v <= 32767) {
/* 16 bit integer. */
if (v < 0) v = ((int64_t)1<<16)+v;
intenc[0] = LP_ENCODING_16BIT_INT;
intenc[1] = v&0xff;
intenc[2] = v>>8;
*enclen = 3;
if (intenc != NULL) {
intenc[0] = LP_ENCODING_16BIT_INT;
intenc[1] = v&0xff;
intenc[2] = v>>8;
}
if (enclen != NULL) *enclen = 3;
} else if (v >= -8388608 && v <= 8388607) {
/* 24 bit integer. */
if (v < 0) v = ((int64_t)1<<24)+v;
intenc[0] = LP_ENCODING_24BIT_INT;
intenc[1] = v&0xff;
intenc[2] = (v>>8)&0xff;
intenc[3] = v>>16;
*enclen = 4;
if (intenc != NULL) {
intenc[0] = LP_ENCODING_24BIT_INT;
intenc[1] = v&0xff;
intenc[2] = (v>>8)&0xff;
intenc[3] = v>>16;
}
if (enclen != NULL) *enclen = 4;
} else if (v >= -2147483648 && v <= 2147483647) {
/* 32 bit integer. */
if (v < 0) v = ((int64_t)1<<32)+v;
intenc[0] = LP_ENCODING_32BIT_INT;
intenc[1] = v&0xff;
intenc[2] = (v>>8)&0xff;
intenc[3] = (v>>16)&0xff;
intenc[4] = v>>24;
*enclen = 5;
if (intenc != NULL) {
intenc[0] = LP_ENCODING_32BIT_INT;
intenc[1] = v&0xff;
intenc[2] = (v>>8)&0xff;
intenc[3] = (v>>16)&0xff;
intenc[4] = v>>24;
}
if (enclen != NULL) *enclen = 5;
} else {
/* 64 bit integer. */
uint64_t uv = v;
intenc[0] = LP_ENCODING_64BIT_INT;
intenc[1] = uv&0xff;
intenc[2] = (uv>>8)&0xff;
intenc[3] = (uv>>16)&0xff;
intenc[4] = (uv>>24)&0xff;
intenc[5] = (uv>>32)&0xff;
intenc[6] = (uv>>40)&0xff;
intenc[7] = (uv>>48)&0xff;
intenc[8] = uv>>56;
*enclen = 9;
if (intenc != NULL) {
intenc[0] = LP_ENCODING_64BIT_INT;
intenc[1] = uv&0xff;
intenc[2] = (uv>>8)&0xff;
intenc[3] = (uv>>16)&0xff;
intenc[4] = (uv>>24)&0xff;
intenc[5] = (uv>>32)&0xff;
intenc[6] = (uv>>40)&0xff;
intenc[7] = (uv>>48)&0xff;
intenc[8] = uv>>56;
}
if (enclen != NULL) *enclen = 9;
}
}
@@ -659,50 +669,47 @@ unsigned char *lpGetValue(unsigned char *p, unsigned int *slen, long long *lval)
return vstr;
}
/* Find pointer to the entry equal to the specified entry. Skip 'skip' entries
* between every comparison. Returns NULL when the field could not be found. */
unsigned char *lpFind(unsigned char *lp, unsigned char *p, unsigned char *s,
uint32_t slen, unsigned int skip) {
/* This is just a wrapper to lpGet() that is able to get an integer from an entry directly.
* Returns 1 and stores the integer in 'lval' if the entry is an integer.
* Returns 0 if the entry is a string. */
int lpGetIntegerValue(unsigned char *p, long long *lval) {
int64_t ele_len;
if (!lpGet(p, &ele_len, NULL)) {
*lval = ele_len;
return 1;
}
return 0;
}
/* Find pointer to the entry with a comparator callback.
*
* 'cmp' is a comparator callback. If it returns zero, current entry pointer
* will be returned. 'user' is passed to this callback.
* Skip 'skip' entries between every comparison.
* Returns NULL when the field could not be found. */
unsigned char *lpFindCb(unsigned char *lp, unsigned char *p,
void *user, lpCmp cmp, unsigned int skip)
{
int skipcnt = 0;
unsigned char vencoding = 0;
unsigned char *value;
int64_t ll, vll;
int64_t ll;
uint64_t entry_size = 123456789; /* initialized to avoid warning. */
uint32_t lp_bytes = lpBytes(lp);
assert(p);
if (!p)
p = lpFirst(lp);
while (p) {
if (skipcnt == 0) {
value = lpGetWithSize(p, &ll, NULL, &entry_size);
if (value) {
/* check the value doesn't reach outside the listpack before accessing it */
assert(p >= lp + LP_HDR_SIZE && p + entry_size < lp + lp_bytes);
if (slen == ll && memcmp(value, s, slen) == 0) {
return p;
}
} else {
/* Find out if the searched field can be encoded. Note that
* we do it only the first time, once done vencoding is set
* to non-zero and vll is set to the integer value. */
if (vencoding == 0) {
/* If the entry can be encoded as integer we set it to
* 1, else set it to UCHAR_MAX, so that we don't retry
* again the next time. */
if (slen >= 32 || slen == 0 || !lpStringToInt64((const char*)s, slen, &vll)) {
vencoding = UCHAR_MAX;
} else {
vencoding = 1;
}
}
/* Compare current entry with specified entry, do it only
* if vencoding != UCHAR_MAX because if there is no encoding
* possible for the field it can't be a valid integer. */
if (vencoding != UCHAR_MAX && ll == vll) {
return p;
}
}
if (cmp(lp, p, user, value, ll) == 0)
return p;
/* Reset skip count */
skipcnt = skip;
p += entry_size;
@@ -727,6 +734,62 @@ unsigned char *lpFind(unsigned char *lp, unsigned char *p, unsigned char *s,
return NULL;
}
struct lpFindArg {
unsigned char *s; /* Item to search */
uint32_t slen; /* Item len */
int vencoding;
int64_t vll;
};
/* Comparator function to find item */
static inline int lpFindCmp(const unsigned char *lp, unsigned char *p,
void *user, unsigned char *s, long long slen) {
(void) lp;
(void) p;
struct lpFindArg *arg = user;
if (s) {
if (slen == arg->slen && memcmp(arg->s, s, slen) == 0) {
return 0;
}
} else {
/* Find out if the searched field can be encoded. Note that
* we do it only the first time, once done vencoding is set
* to non-zero and vll is set to the integer value. */
if (arg->vencoding == 0) {
/* If the entry can be encoded as integer we set it to
* 1, else set it to UCHAR_MAX, so that we don't retry
* again the next time. */
if (arg->slen >= 32 || arg->slen == 0 || !lpStringToInt64((const char*)arg->s, arg->slen, &arg->vll)) {
arg->vencoding = UCHAR_MAX;
} else {
arg->vencoding = 1;
}
}
/* Compare current entry with specified entry, do it only
* if vencoding != UCHAR_MAX because if there is no encoding
* possible for the field it can't be a valid integer. */
if (arg->vencoding != UCHAR_MAX && slen == arg->vll) {
return 0;
}
}
return 1;
}
/* Find pointer to the entry equal to the specified entry. Skip 'skip' entries
* between every comparison. Returns NULL when the field could not be found. */
unsigned char *lpFind(unsigned char *lp, unsigned char *p, unsigned char *s,
uint32_t slen, unsigned int skip)
{
struct lpFindArg arg = {
.s = s,
.slen = slen
};
return lpFindCb(lp, p, &arg, lpFindCmp, skip);
}
/* Insert, delete or replace the specified string element 'elestr' of length
* 'size' or integer element 'eleint' at the specified position 'p', with 'p'
* being a listpack element pointer obtained with lpFirst(), lpLast(), lpNext(),
@@ -904,6 +967,140 @@ unsigned char *lpInsert(unsigned char *lp, unsigned char *elestr, unsigned char
return lp;
}
/* Insert the specified elements with 'entries' and 'len' at the specified
* position 'p', with 'p' being a listpack element pointer obtained with
* lpFirst(), lpLast(), lpNext(), lpPrev() or lpSeek().
*
* This is similar to lpInsert() but allows you to insert batch of entries in
* one call. This function is more efficient than inserting entries one by one
* as it does single realloc()/memmove() calls for all the entries.
*
* In each listpackEntry, if 'sval' is not null, it is assumed entry is string
* and 'sval' and 'slen' will be used. Otherwise, 'lval' will be used to append
* the integer entry.
*
* The elements are inserted before or after the element pointed by 'p'
* depending on the 'where' argument, that can be LP_BEFORE or LP_AFTER.
*
* If 'newp' is not NULL, at the end of a successful call '*newp' will be set
* to the address of the element just added, so that it will be possible to
* continue an interaction with lpNext() and lpPrev().
*
* Returns NULL on out of memory or when the listpack total length would exceed
* the max allowed size of 2^32-1, otherwise the new pointer to the listpack
* holding the new element is returned (and the old pointer passed is no longer
* considered valid). */
unsigned char *lpBatchInsert(unsigned char *lp, unsigned char *p, int where,
listpackEntry *entries, unsigned int len,
unsigned char **newp)
{
assert(where == LP_BEFORE || where == LP_AFTER);
assert(entries != NULL && len > 0);
struct listpackInsertEntry {
int enctype;
uint64_t enclen;
unsigned char intenc[LP_MAX_INT_ENCODING_LEN];
unsigned char backlen[LP_MAX_BACKLEN_SIZE];
unsigned long backlen_size;
};
uint64_t addedlen = 0; /* The encoded length of the added elements. */
struct listpackInsertEntry tmp[3]; /* Encoded entries */
struct listpackInsertEntry *enc = tmp;
if (len > sizeof(tmp) / sizeof(struct listpackInsertEntry)) {
/* If 'len' is larger than local buffer size, allocate on heap. */
enc = zmalloc(len * sizeof(struct listpackInsertEntry));
}
/* If we need to insert after the current element, we just jump to the
* next element (that could be the EOF one) and handle the case of
* inserting before. So the function will actually deal with just one
* case: LP_BEFORE. */
if (where == LP_AFTER) {
p = lpSkip(p);
where = LP_BEFORE;
ASSERT_INTEGRITY(lp, p);
}
for (unsigned int i = 0; i < len; i++) {
listpackEntry *e = &entries[i];
if (e->sval) {
/* Calling lpEncodeGetType() results into the encoded version of the
* element to be stored into 'intenc' in case it is representable as
* an integer: in that case, the function returns LP_ENCODING_INT.
* Otherwise, if LP_ENCODING_STR is returned, we'll have to call
* lpEncodeString() to actually write the encoded string on place
* later.
*
* Whatever the returned encoding is, 'enclen' is populated with the
* length of the encoded element. */
enc[i].enctype = lpEncodeGetType(e->sval, e->slen,
enc[i].intenc, &enc[i].enclen);
} else {
enc[i].enctype = LP_ENCODING_INT;
lpEncodeIntegerGetType(e->lval, enc[i].intenc, &enc[i].enclen);
}
addedlen += enc[i].enclen;
/* We need to also encode the backward-parsable length of the element
* and append it to the end: this allows to traverse the listpack from
* the end to the start. */
enc[i].backlen_size = lpEncodeBacklen(enc[i].backlen, enc[i].enclen);
addedlen += enc[i].backlen_size;
}
uint64_t old_listpack_bytes = lpGetTotalBytes(lp);
uint64_t new_listpack_bytes = old_listpack_bytes + addedlen;
if (new_listpack_bytes > UINT32_MAX) return NULL;
/* Store the offset of the element 'p', so that we can obtain its
* address again after a reallocation. */
unsigned long poff = p-lp;
unsigned char *dst = lp + poff; /* May be updated after reallocation. */
/* Realloc before: we need more room. */
if (new_listpack_bytes > old_listpack_bytes &&
new_listpack_bytes > lp_malloc_size(lp)) {
if ((lp = lp_realloc(lp,new_listpack_bytes)) == NULL) return NULL;
dst = lp + poff;
}
/* Setup the listpack relocating the elements to make the exact room
* we need to store the new ones. */
memmove(dst+addedlen,dst,old_listpack_bytes-poff);
for (unsigned int i = 0; i < len; i++) {
listpackEntry *ent = &entries[i];
if (newp)
*newp = dst;
if (enc[i].enctype == LP_ENCODING_INT)
memcpy(dst, enc[i].intenc, enc[i].enclen);
else
lpEncodeString(dst, ent->sval, ent->slen);
dst += enc[i].enclen;
memcpy(dst, enc[i].backlen, enc[i].backlen_size);
dst += enc[i].backlen_size;
}
/* Update header. */
uint32_t num_elements = lpGetNumElements(lp);
if (num_elements != LP_HDR_NUMELE_UNKNOWN) {
if ((int64_t) len > (int64_t) LP_HDR_NUMELE_UNKNOWN - (int64_t) num_elements)
lpSetNumElements(lp, LP_HDR_NUMELE_UNKNOWN);
else
lpSetNumElements(lp,num_elements + len);
}
lpSetTotalBytes(lp,new_listpack_bytes);
if (enc != tmp) lp_free(enc);
return lp;
}
/* This is just a wrapper for lpInsert() to directly use a string. */
unsigned char *lpInsertString(unsigned char *lp, unsigned char *s, uint32_t slen,
unsigned char *p, int where, unsigned char **newp)
@@ -951,6 +1148,20 @@ unsigned char *lpAppendInteger(unsigned char *lp, long long lval) {
return lpInsertInteger(lp, lval, eofptr, LP_BEFORE, NULL);
}
/* Append batch of entries to the listpack.
*
* This call is more efficient than multiple lpAppend() calls as it only does
* a single realloc() for all the given entries.
*
* In each listpackEntry, if 'sval' is not null, it is assumed entry is string
* and 'sval' and 'slen' will be used. Otherwise, 'lval' will be used to append
* the integer entry. */
unsigned char *lpBatchAppend(unsigned char *lp, listpackEntry *entries, unsigned long len) {
uint64_t listpack_bytes = lpGetTotalBytes(lp);
unsigned char *eofptr = lp + listpack_bytes - 1;
return lpBatchInsert(lp, eofptr, LP_BEFORE, entries, len, NULL);
}
/* This is just a wrapper for lpInsert() to directly use a string to replace
* the current element. The function returns the new listpack as return
* value, and also updates the current cursor by updating '*p'. */
@@ -1199,13 +1410,17 @@ size_t lpBytes(unsigned char *lp) {
return lpGetTotalBytes(lp);
}
/* Returns the size 'lval' will require when encoded, in bytes */
size_t lpEntrySizeInteger(long long lval) {
uint64_t enclen;
lpEncodeIntegerGetType(lval, NULL, &enclen);
unsigned long backlen = lpEncodeBacklen(NULL, enclen);
return enclen + backlen;
}
/* Returns the size of a listpack consisting of an integer repeated 'rep' times. */
size_t lpEstimateBytesRepeatedInteger(long long lval, unsigned long rep) {
uint64_t enclen;
unsigned char intenc[LP_MAX_INT_ENCODING_LEN];
lpEncodeIntegerGetType(lval, intenc, &enclen);
unsigned long backlen = lpEncodeBacklen(NULL, enclen);
return LP_HDR_SIZE + (enclen + backlen) * rep + 1;
return LP_HDR_SIZE + lpEntrySizeInteger(lval) * rep + 1;
}
/* Seek the specified element and returns the pointer to the seeked element.
@@ -1408,15 +1623,20 @@ static inline void lpSaveValue(unsigned char *val, unsigned int len, int64_t lva
/* Randomly select a pair of key and value.
* total_count is a pre-computed length/2 of the listpack (to avoid calls to lpLength)
* 'key' and 'val' are used to store the result key value pair.
* 'val' can be NULL if the value is not needed. */
void lpRandomPair(unsigned char *lp, unsigned long total_count, listpackEntry *key, listpackEntry *val) {
* 'val' can be NULL if the value is not needed.
* 'tuple_len' indicates entry count of a single logical item. It should be 2
* if listpack was saved as key-value pair or more for key-value-...(n_entries). */
void lpRandomPair(unsigned char *lp, unsigned long total_count,
listpackEntry *key, listpackEntry *val, int tuple_len)
{
unsigned char *p;
assert(tuple_len >= 2);
/* Avoid div by zero on corrupt listpack */
assert(total_count);
/* Generate even numbers, because listpack saved K-V pair */
int r = (rand() % total_count) * 2;
int r = (rand() % total_count) * tuple_len;
assert((p = lpSeek(lp, r)));
key->sval = lpGetValue(p, &(key->slen), &(key->lval));
@@ -1466,26 +1686,31 @@ void lpRandomEntries(unsigned char *lp, unsigned int count, listpackEntry *entri
/* Randomly select count of key value pairs and store into 'keys' and
* 'vals' args. The order of the picked entries is random, and the selections
* are non-unique (repetitions are possible).
* The 'vals' arg can be NULL in which case we skip these. */
void lpRandomPairs(unsigned char *lp, unsigned int count, listpackEntry *keys, listpackEntry *vals) {
* The 'vals' arg can be NULL in which case we skip these.
* 'tuple_len' indicates entry count of a single logical item. It should be 2
* if listpack was saved as key-value pair or more for key-value-...(n_entries). */
void lpRandomPairs(unsigned char *lp, unsigned int count, listpackEntry *keys, listpackEntry *vals, int tuple_len) {
unsigned char *p, *key, *value;
unsigned int klen = 0, vlen = 0;
long long klval = 0, vlval = 0;
assert(tuple_len >= 2);
/* Notice: the index member must be first due to the use in uintCompare */
typedef struct {
unsigned int index;
unsigned int order;
} rand_pick;
rand_pick *picks = lp_malloc(sizeof(rand_pick)*count);
unsigned int total_size = lpLength(lp)/2;
unsigned int total_size = lpLength(lp)/tuple_len;
/* Avoid div by zero on corrupt listpack */
assert(total_size);
/* create a pool of random indexes (some may be duplicate). */
for (unsigned int i = 0; i < count; i++) {
picks[i].index = (rand() % total_size) * 2; /* Generate even indexes */
/* Generate indexes that key exist at */
picks[i].index = (rand() % total_size) * tuple_len;
/* keep track of the order we picked them */
picks[i].order = i;
}
@@ -1507,8 +1732,11 @@ void lpRandomPairs(unsigned char *lp, unsigned int count, listpackEntry *keys, l
lpSaveValue(value, vlen, vlval, &vals[storeorder]);
pickindex++;
}
lpindex += 2;
p = lpNext(lp, p);
lpindex += tuple_len;
for (int i = 0; i < tuple_len - 1; i++) {
p = lpNext(lp, p);
}
}
lp_free(picks);
@@ -1518,13 +1746,20 @@ void lpRandomPairs(unsigned char *lp, unsigned int count, listpackEntry *keys, l
* 'vals' args. The selections are unique (no repetitions), and the order of
* the picked entries is NOT-random.
* The 'vals' arg can be NULL in which case we skip these.
* 'tuple_len' indicates entry count of a single logical item. It should be 2
* if listpack was saved as key-value pair or more for key-value-...(n_entries).
* The return value is the number of items picked which can be lower than the
* requested count if the listpack doesn't hold enough pairs. */
unsigned int lpRandomPairsUnique(unsigned char *lp, unsigned int count, listpackEntry *keys, listpackEntry *vals) {
unsigned int lpRandomPairsUnique(unsigned char *lp, unsigned int count,
listpackEntry *keys, listpackEntry *vals,
int tuple_len)
{
assert(tuple_len >= 2);
unsigned char *p, *key;
unsigned int klen = 0;
long long klval = 0;
unsigned int total_size = lpLength(lp)/2;
unsigned int total_size = lpLength(lp)/tuple_len;
unsigned int index = 0;
if (count > total_size)
count = total_size;
@@ -1532,7 +1767,7 @@ unsigned int lpRandomPairsUnique(unsigned char *lp, unsigned int count, listpack
p = lpFirst(lp);
unsigned int picked = 0, remaining = count;
while (picked < count && p) {
assert((p = lpNextRandom(lp, p, &index, remaining, 1)));
assert((p = lpNextRandom(lp, p, &index, remaining, tuple_len)));
key = lpGetValue(p, &klen, &klval);
lpSaveValue(key, klen, klval, &keys[picked]);
assert((p = lpNext(lp, p)));
@@ -1554,8 +1789,9 @@ unsigned int lpRandomPairsUnique(unsigned char *lp, unsigned int count, listpack
* the end of the list. The 'index' needs to be initialized according to the
* current zero-based index matching the position of the starting element 'p'
* and is updated to match the returned element's zero-based index. If
* 'even_only' is nonzero, an element with an even index is picked, which is
* useful if the listpack represents a key-value pair sequence.
* 'tuple_len' indicates entry count of a single logical item. e.g. This is
* useful if listpack represents key-value pairs. In this case, tuple_len should
* be two and even indexes will be picked.
*
* Note that this function can return p. In order to skip the previously
* returned element, you need to call lpNext() or lpDelete() after each call to
@@ -1565,7 +1801,7 @@ unsigned int lpRandomPairsUnique(unsigned char *lp, unsigned int count, listpack
* p = lpFirst(lp);
* i = 0;
* while (remaining > 0) {
* p = lpNextRandom(lp, p, &i, remaining--, 0);
* p = lpNextRandom(lp, p, &i, remaining--, 1);
*
* // ... Do stuff with p ...
*
@@ -1574,8 +1810,9 @@ unsigned int lpRandomPairsUnique(unsigned char *lp, unsigned int count, listpack
* }
*/
unsigned char *lpNextRandom(unsigned char *lp, unsigned char *p, unsigned int *index,
unsigned int remaining, int even_only)
unsigned int remaining, int tuple_len)
{
assert(tuple_len > 0);
/* To only iterate once, every time we try to pick a member, the probability
* we pick it is the quotient of the count left we want to pick and the
* count still we haven't visited. This way, we could make every member be
@@ -1583,15 +1820,14 @@ unsigned char *lpNextRandom(unsigned char *lp, unsigned char *p, unsigned int *i
unsigned int i = *index;
unsigned int total_size = lpLength(lp);
while (i < total_size && p != NULL) {
if (even_only && i % 2 != 0) {
if (i % tuple_len != 0) {
p = lpNext(lp, p);
i++;
continue;
}
/* Do we pick this element? */
unsigned int available = total_size - i;
if (even_only) available /= 2;
unsigned int available = (total_size - i) / tuple_len;
double randomDouble = ((double)rand()) / RAND_MAX;
double threshold = ((double)remaining) / available;
if (randomDouble <= threshold) {
@@ -1787,6 +2023,24 @@ static int lpValidation(unsigned char *p, unsigned int head_count, void *userdat
return ret;
}
static int lpFindCbCmp(const unsigned char *lp, unsigned char *p, void *user, unsigned char *s, long long slen) {
assert(lp);
assert(p);
char *n = user;
if (!s) {
int64_t sval;
if (lpStringToInt64((const char*)n, strlen(n), &sval))
return slen == sval ? 0 : 1;
} else {
if (strlen(n) == (size_t) slen && memcmp(n, s, slen) == 0)
return 0;
}
return 1;
}
int listpackTest(int argc, char *argv[], int flags) {
UNUSED(argc);
UNUSED(argv);
@@ -2031,6 +2285,111 @@ int listpackTest(int argc, char *argv[], int flags) {
zfree(lp);
}
TEST("Batch append") {
listpackEntry ent[6] = {
{.sval = (unsigned char*)mixlist[0], .slen = strlen(mixlist[0])},
{.sval = (unsigned char*)mixlist[1], .slen = strlen(mixlist[1])},
{.sval = (unsigned char*)mixlist[2], .slen = strlen(mixlist[2])},
{.lval = 4294967296},
{.sval = (unsigned char*)mixlist[3], .slen = strlen(mixlist[3])},
{.lval = -100}
};
lp = lpNew(0);
lp = lpBatchAppend(lp, ent, 2);
verifyEntry(lpSeek(lp, 0), ent[0].sval, ent[0].slen);
verifyEntry(lpSeek(lp, 1), ent[1].sval, ent[1].slen);
assert(lpLength(lp) == 2);
lp = lpBatchAppend(lp, &ent[2], 1);
verifyEntry(lpSeek(lp, 0), ent[0].sval, ent[0].slen);
verifyEntry(lpSeek(lp, 1), ent[1].sval, ent[1].slen);
verifyEntry(lpSeek(lp, 2), ent[2].sval, ent[2].slen);
assert(lpLength(lp) == 3);
lp = lpDeleteRange(lp, 1, 1);
verifyEntry(lpSeek(lp, 0), ent[0].sval, ent[0].slen);
verifyEntry(lpSeek(lp, 1), ent[2].sval, ent[2].slen);
assert(lpLength(lp) == 2);
lp = lpBatchAppend(lp, &ent[3], 3);
verifyEntry(lpSeek(lp, 0), ent[0].sval, ent[0].slen);
verifyEntry(lpSeek(lp, 1), ent[2].sval, ent[2].slen);
verifyEntry(lpSeek(lp, 2), (unsigned char*) "4294967296", 10);
verifyEntry(lpSeek(lp, 3), ent[4].sval, ent[4].slen);
verifyEntry(lpSeek(lp, 4), (unsigned char*) "-100", 4);
assert(lpLength(lp) == 5);
lp = lpDeleteRange(lp, 1, 3);
verifyEntry(lpSeek(lp, 0), ent[0].sval, ent[0].slen);
verifyEntry(lpSeek(lp, 1), (unsigned char*) "-100", 4);
assert(lpLength(lp) == 2);
lpFree(lp);
}
TEST("Batch insert") {
lp = lpNew(0);
listpackEntry ent[6] = {
{.sval = (unsigned char*)mixlist[0], .slen = strlen(mixlist[0])},
{.sval = (unsigned char*)mixlist[1], .slen = strlen(mixlist[1])},
{.sval = (unsigned char*)mixlist[2], .slen = strlen(mixlist[2])},
{.lval = 4294967296},
{.sval = (unsigned char*)mixlist[3], .slen = strlen(mixlist[3])},
{.lval = -100}
};
lp = lpBatchAppend(lp, ent, 4);
assert(lpLength(lp) == 4);
verifyEntry(lpSeek(lp, 0), ent[0].sval, ent[0].slen);
verifyEntry(lpSeek(lp, 1), ent[1].sval, ent[1].slen);
verifyEntry(lpSeek(lp, 2), ent[2].sval, ent[2].slen);
verifyEntry(lpSeek(lp, 3), (unsigned char*)"4294967296", 10);
/* Insert with LP_BEFORE */
p = lpSeek(lp, 3);
lp = lpBatchInsert(lp, p, LP_BEFORE, &ent[4], 2, &p);
verifyEntry(p, (unsigned char*)"-100", 4);
assert(lpLength(lp) == 6);
verifyEntry(lpSeek(lp, 0), ent[0].sval, ent[0].slen);
verifyEntry(lpSeek(lp, 1), ent[1].sval, ent[1].slen);
verifyEntry(lpSeek(lp, 2), ent[2].sval, ent[2].slen);
verifyEntry(lpSeek(lp, 3), ent[4].sval, ent[4].slen);
verifyEntry(lpSeek(lp, 4), (unsigned char*)"-100", 4);
verifyEntry(lpSeek(lp, 5), (unsigned char*)"4294967296", 10);
lp = lpDeleteRange(lp, 1, 2);
assert(lpLength(lp) == 4);
verifyEntry(lpSeek(lp, 0), ent[0].sval, ent[0].slen);
verifyEntry(lpSeek(lp, 1), ent[4].sval, ent[4].slen);
verifyEntry(lpSeek(lp, 2), (unsigned char*)"-100", 4);
verifyEntry(lpSeek(lp, 3), (unsigned char*)"4294967296", 10);
/* Insert with LP_AFTER */
p = lpSeek(lp, 0);
lp = lpBatchInsert(lp, p, LP_AFTER, &ent[1], 2, &p);
verifyEntry(p, ent[2].sval, ent[2].slen);
assert(lpLength(lp) == 6);
verifyEntry(lpSeek(lp, 0), ent[0].sval, ent[0].slen);
verifyEntry(lpSeek(lp, 1), ent[1].sval, ent[1].slen);
verifyEntry(lpSeek(lp, 2), ent[2].sval, ent[2].slen);
verifyEntry(lpSeek(lp, 3), ent[4].sval, ent[4].slen);
verifyEntry(lpSeek(lp, 4), (unsigned char*)"-100", 4);
verifyEntry(lpSeek(lp, 5), (unsigned char*)"4294967296", 10);
lp = lpDeleteRange(lp, 2, 4);
assert(lpLength(lp) == 2);
p = lpSeek(lp, 1);
lp = lpBatchInsert(lp, p, LP_AFTER, &ent[2], 1, &p);
verifyEntry(p, ent[2].sval, ent[2].slen);
assert(lpLength(lp) == 3);
verifyEntry(lpSeek(lp, 0), ent[0].sval, ent[0].slen);
verifyEntry(lpSeek(lp, 1), ent[1].sval, ent[1].slen);
verifyEntry(lpSeek(lp, 2), ent[2].sval, ent[2].slen);
lpFree(lp);
}
TEST("Batch delete") {
unsigned char *lp = createList(); /* char *mixlist[] = {"hello", "foo", "quux", "1024"} */
assert(lpLength(lp) == 4); /* Pre-condition */
@@ -2210,7 +2569,7 @@ int listpackTest(int argc, char *argv[], int flags) {
unsigned index = 0;
while (remaining > 0) {
assert(p != NULL);
p = lpNextRandom(lp, p, &index, remaining--, 0);
p = lpNextRandom(lp, p, &index, remaining--, 1);
assert(p != NULL);
assert(p != prev);
prev = p;
@@ -2226,7 +2585,7 @@ int listpackTest(int argc, char *argv[], int flags) {
unsigned i = 0;
/* Pick from empty listpack returns NULL. */
assert(lpNextRandom(lp, NULL, &i, 2, 0) == NULL);
assert(lpNextRandom(lp, NULL, &i, 2, 1) == NULL);
/* Add some elements and find their pointers within the listpack. */
lp = lpAppend(lp, (unsigned char *)"abc", 3);
@@ -2239,19 +2598,19 @@ int listpackTest(int argc, char *argv[], int flags) {
assert(lpNext(lp, p2) == NULL);
/* Pick zero elements returns NULL. */
i = 0; assert(lpNextRandom(lp, lpFirst(lp), &i, 0, 0) == NULL);
i = 0; assert(lpNextRandom(lp, lpFirst(lp), &i, 0, 1) == NULL);
/* Pick all returns all. */
i = 0; assert(lpNextRandom(lp, p0, &i, 3, 0) == p0 && i == 0);
i = 1; assert(lpNextRandom(lp, p1, &i, 2, 0) == p1 && i == 1);
i = 2; assert(lpNextRandom(lp, p2, &i, 1, 0) == p2 && i == 2);
i = 0; assert(lpNextRandom(lp, p0, &i, 3, 1) == p0 && i == 0);
i = 1; assert(lpNextRandom(lp, p1, &i, 2, 1) == p1 && i == 1);
i = 2; assert(lpNextRandom(lp, p2, &i, 1, 1) == p2 && i == 2);
/* Pick more than one when there's only one left returns the last one. */
i = 2; assert(lpNextRandom(lp, p2, &i, 42, 0) == p2 && i == 2);
i = 2; assert(lpNextRandom(lp, p2, &i, 42, 1) == p2 && i == 2);
/* Pick all even elements returns p0 and p2. */
i = 0; assert(lpNextRandom(lp, p0, &i, 10, 1) == p0 && i == 0);
i = 1; assert(lpNextRandom(lp, p1, &i, 10, 1) == p2 && i == 2);
i = 0; assert(lpNextRandom(lp, p0, &i, 10, 2) == p0 && i == 0);
i = 1; assert(lpNextRandom(lp, p1, &i, 10, 2) == p2 && i == 2);
/* Don't crash even for bad index. */
for (int j = 0; j < 100; j++) {
@@ -2264,7 +2623,7 @@ int listpackTest(int argc, char *argv[], int flags) {
}
i = j % 7;
unsigned int remaining = j % 5;
p = lpNextRandom(lp, p, &i, remaining, 0);
p = lpNextRandom(lp, p, &i, remaining, 1);
assert(p == p0 || p == p1 || p == p2 || p == NULL);
}
lpFree(lp);
@@ -2275,7 +2634,7 @@ int listpackTest(int argc, char *argv[], int flags) {
unsigned char *lp = lpNew(0);
lp = lpAppend(lp, (unsigned char*)"abc", 3);
lp = lpAppend(lp, (unsigned char*)"123", 3);
lpRandomPair(lp, 1, &key, &val);
lpRandomPair(lp, 1, &key, &val, 2);
assert(memcmp(key.sval, "abc", key.slen) == 0);
assert(val.lval == 123);
lpFree(lp);
@@ -2288,7 +2647,7 @@ int listpackTest(int argc, char *argv[], int flags) {
lp = lpAppend(lp, (unsigned char*)"123", 3);
lp = lpAppend(lp, (unsigned char*)"456", 3);
lp = lpAppend(lp, (unsigned char*)"def", 3);
lpRandomPair(lp, 2, &key, &val);
lpRandomPair(lp, 2, &key, &val, 2);
if (key.sval) {
assert(!memcmp(key.sval, "abc", key.slen));
assert(key.slen == 3);
@@ -2301,6 +2660,42 @@ int listpackTest(int argc, char *argv[], int flags) {
lpFree(lp);
}
TEST("Random pair with tuple_len 3") {
listpackEntry key, val;
unsigned char *lp = lpNew(0);
lp = lpAppend(lp, (unsigned char*)"abc", 3);
lp = lpAppend(lp, (unsigned char*)"123", 3);
lp = lpAppend(lp, (unsigned char*)"xxx", 3);
lp = lpAppend(lp, (unsigned char*)"456", 3);
lp = lpAppend(lp, (unsigned char*)"def", 3);
lp = lpAppend(lp, (unsigned char*)"xxx", 3);
lp = lpAppend(lp, (unsigned char*)"281474976710655", 15);
lp = lpAppend(lp, (unsigned char*)"789", 3);
lp = lpAppend(lp, (unsigned char*)"xxx", 3);
for (int i = 0; i < 5; i++) {
lpRandomPair(lp, 3, &key, &val, 3);
if (key.sval) {
if (!memcmp(key.sval, "abc", key.slen)) {
assert(key.slen == 3);
assert(val.lval == 123);
} else {
assert(0);
};
}
if (!key.sval) {
if (key.lval == 456)
assert(!memcmp(val.sval, "def", val.slen));
else if (key.lval == 281474976710655LL)
assert(val.lval == 789);
else
assert(0);
}
}
lpFree(lp);
}
TEST("Random pairs with one element") {
int count = 5;
unsigned char *lp = lpNew(0);
@@ -2309,7 +2704,7 @@ int listpackTest(int argc, char *argv[], int flags) {
lp = lpAppend(lp, (unsigned char*)"abc", 3);
lp = lpAppend(lp, (unsigned char*)"123", 3);
lpRandomPairs(lp, count, keys, vals);
lpRandomPairs(lp, count, keys, vals, 2);
assert(memcmp(keys[4].sval, "abc", keys[4].slen) == 0);
assert(vals[4].lval == 123);
zfree(keys);
@@ -2327,7 +2722,7 @@ int listpackTest(int argc, char *argv[], int flags) {
lp = lpAppend(lp, (unsigned char*)"123", 3);
lp = lpAppend(lp, (unsigned char*)"456", 3);
lp = lpAppend(lp, (unsigned char*)"def", 3);
lpRandomPairs(lp, count, keys, vals);
lpRandomPairs(lp, count, keys, vals, 2);
for (int i = 0; i < count; i++) {
if (keys[i].sval) {
assert(!memcmp(keys[i].sval, "abc", keys[i].slen));
@@ -2344,6 +2739,47 @@ int listpackTest(int argc, char *argv[], int flags) {
lpFree(lp);
}
TEST("Random pairs with many elements and tuple_len 3") {
int count = 5;
lp = lpNew(0);
listpackEntry *keys = zcalloc(sizeof(listpackEntry) * count);
listpackEntry *vals = zcalloc(sizeof(listpackEntry) * count);
lp = lpAppend(lp, (unsigned char*)"abc", 3);
lp = lpAppend(lp, (unsigned char*)"123", 3);
lp = lpAppend(lp, (unsigned char*)"xxx", 3);
lp = lpAppend(lp, (unsigned char*)"456", 3);
lp = lpAppend(lp, (unsigned char*)"def", 3);
lp = lpAppend(lp, (unsigned char*)"xxx", 3);
lp = lpAppend(lp, (unsigned char*)"281474976710655", 15);
lp = lpAppend(lp, (unsigned char*)"789", 3);
lp = lpAppend(lp, (unsigned char*)"xxx", 3);
lpRandomPairs(lp, count, keys, vals, 3);
for (int i = 0; i < count; i++) {
if (keys[i].sval) {
if (!memcmp(keys[i].sval, "abc", keys[i].slen)) {
assert(keys[i].slen == 3);
assert(vals[i].lval == 123);
} else {
assert(0);
};
}
if (!keys[i].sval) {
if (keys[i].lval == 456)
assert(!memcmp(vals[i].sval, "def", vals[i].slen));
else if (keys[i].lval == 281474976710655LL)
assert(vals[i].lval == 789);
else
assert(0);
}
}
zfree(keys);
zfree(vals);
lpFree(lp);
}
TEST("Random pairs unique with one element") {
unsigned picked;
int count = 5;
@@ -2353,7 +2789,7 @@ int listpackTest(int argc, char *argv[], int flags) {
lp = lpAppend(lp, (unsigned char*)"abc", 3);
lp = lpAppend(lp, (unsigned char*)"123", 3);
picked = lpRandomPairsUnique(lp, count, keys, vals);
picked = lpRandomPairsUnique(lp, count, keys, vals, 2);
assert(picked == 1);
assert(memcmp(keys[0].sval, "abc", keys[0].slen) == 0);
assert(vals[0].lval == 123);
@@ -2373,7 +2809,7 @@ int listpackTest(int argc, char *argv[], int flags) {
lp = lpAppend(lp, (unsigned char*)"123", 3);
lp = lpAppend(lp, (unsigned char*)"456", 3);
lp = lpAppend(lp, (unsigned char*)"def", 3);
picked = lpRandomPairsUnique(lp, count, keys, vals);
picked = lpRandomPairsUnique(lp, count, keys, vals, 2);
assert(picked == 2);
for (int i = 0; i < 2; i++) {
if (keys[i].sval) {
@@ -2391,6 +2827,47 @@ int listpackTest(int argc, char *argv[], int flags) {
lpFree(lp);
}
TEST("Random pairs unique with many elements and tuple_len 3") {
unsigned picked;
int count = 5;
lp = lpNew(0);
listpackEntry *keys = zmalloc(sizeof(listpackEntry) * count);
listpackEntry *vals = zmalloc(sizeof(listpackEntry) * count);
lp = lpAppend(lp, (unsigned char*)"abc", 3);
lp = lpAppend(lp, (unsigned char*)"123", 3);
lp = lpAppend(lp, (unsigned char*)"xxx", 3);
lp = lpAppend(lp, (unsigned char*)"456", 3);
lp = lpAppend(lp, (unsigned char*)"def", 3);
lp = lpAppend(lp, (unsigned char*)"xxx", 3);
lp = lpAppend(lp, (unsigned char*)"281474976710655", 15);
lp = lpAppend(lp, (unsigned char*)"789", 3);
lp = lpAppend(lp, (unsigned char*)"xxx", 3);
picked = lpRandomPairsUnique(lp, count, keys, vals, 3);
assert(picked == 3);
for (int i = 0; i < 3; i++) {
if (keys[i].sval) {
if (!memcmp(keys[i].sval, "abc", keys[i].slen)) {
assert(keys[i].slen == 3);
assert(vals[i].lval == 123);
} else {
assert(0);
};
}
if (!keys[i].sval) {
if (keys[i].lval == 456)
assert(!memcmp(vals[i].sval, "def", vals[i].slen));
else if (keys[i].lval == 281474976710655LL)
assert(vals[i].lval == 789);
else
assert(0);
}
}
zfree(keys);
zfree(vals);
lpFree(lp);
}
TEST("push various encodings") {
lp = lpNew(0);
@@ -2449,6 +2926,21 @@ int listpackTest(int argc, char *argv[], int flags) {
lpFree(lp);
}
TEST("Test lpFindCb") {
lp = createList(); /* "hello", "foo", "quux", "1024" */
assert(lpFindCb(lp, lpFirst(lp), "abc", lpFindCbCmp, 0) == NULL);
verifyEntry(lpFindCb(lp, NULL, "hello", lpFindCbCmp, 0), (unsigned char*)"hello", 5);
verifyEntry(lpFindCb(lp, NULL, "1024", lpFindCbCmp, 0), (unsigned char*)"1024", 4);
verifyEntry(lpFindCb(lp, NULL, "quux", lpFindCbCmp, 0), (unsigned char*)"quux", 4);
verifyEntry(lpFindCb(lp, NULL, "foo", lpFindCbCmp, 0), (unsigned char*)"foo", 3);
lpFree(lp);
lp = lpNew(0);
assert(lpFindCb(lp, lpFirst(lp), "hello", lpFindCbCmp, 0) == NULL);
assert(lpFindCb(lp, lpFirst(lp), "1024", lpFindCbCmp, 0) == NULL);
lpFree(lp);
}
TEST("Test lpValidateIntegrity") {
lp = createList();
long count = 0;
@@ -2471,6 +2963,26 @@ int listpackTest(int argc, char *argv[], int flags) {
lpFree(lp);
}
TEST("Test number of elements exceeds LP_HDR_NUMELE_UNKNOWN with batch insert") {
listpackEntry ent[2] = {
{.sval = (unsigned char*)mixlist[0], .slen = strlen(mixlist[0])},
{.sval = (unsigned char*)mixlist[1], .slen = strlen(mixlist[1])}
};
lp = lpNew(0);
for (int i = 0; i < (LP_HDR_NUMELE_UNKNOWN/2) + 1; i++)
lp = lpBatchAppend(lp, ent, 2);
assert(lpGetNumElements(lp) == LP_HDR_NUMELE_UNKNOWN);
assert(lpLength(lp) == LP_HDR_NUMELE_UNKNOWN+1);
lp = lpDeleteRange(lp, -2, 2);
assert(lpGetNumElements(lp) == LP_HDR_NUMELE_UNKNOWN);
assert(lpLength(lp) == LP_HDR_NUMELE_UNKNOWN-1);
assert(lpGetNumElements(lp) == LP_HDR_NUMELE_UNKNOWN-1); /* update length after lpLength */
lpFree(lp);
}
TEST("Stress with random payloads of different encoding") {
unsigned long long start = usec();
int i,j,len,where;
src/listpack.h
+14 -4
View File
@@ -49,18 +49,25 @@ unsigned char *lpReplaceInteger(unsigned char *lp, unsigned char **p, long long
unsigned char *lpDelete(unsigned char *lp, unsigned char *p, unsigned char **newp);
unsigned char *lpDeleteRangeWithEntry(unsigned char *lp, unsigned char **p, unsigned long num);
unsigned char *lpDeleteRange(unsigned char *lp, long index, unsigned long num);
unsigned char *lpBatchAppend(unsigned char *lp, listpackEntry *entries, unsigned long len);
unsigned char *lpBatchInsert(unsigned char *lp, unsigned char *p, int where,
listpackEntry *entries, unsigned int len, unsigned char **newp);
unsigned char *lpBatchDelete(unsigned char *lp, unsigned char **ps, unsigned long count);
unsigned char *lpMerge(unsigned char **first, unsigned char **second);
unsigned char *lpDup(unsigned char *lp);
unsigned long lpLength(unsigned char *lp);
unsigned char *lpGet(unsigned char *p, int64_t *count, unsigned char *intbuf);
unsigned char *lpGetValue(unsigned char *p, unsigned int *slen, long long *lval);
int lpGetIntegerValue(unsigned char *p, long long *lval);
unsigned char *lpFind(unsigned char *lp, unsigned char *p, unsigned char *s, uint32_t slen, unsigned int skip);
typedef int (*lpCmp)(const unsigned char *lp, unsigned char *p, void *user, unsigned char *s, long long slen);
unsigned char *lpFindCb(unsigned char *lp, unsigned char *p, void *user, lpCmp cmp, unsigned int skip);
unsigned char *lpFirst(unsigned char *lp);
unsigned char *lpLast(unsigned char *lp);
unsigned char *lpNext(unsigned char *lp, unsigned char *p);
unsigned char *lpPrev(unsigned char *lp, unsigned char *p);
size_t lpBytes(unsigned char *lp);
size_t lpEntrySizeInteger(long long lval);
size_t lpEstimateBytesRepeatedInteger(long long lval, unsigned long rep);
unsigned char *lpSeek(unsigned char *lp, long index);
typedef int (*listpackValidateEntryCB)(unsigned char *p, unsigned int head_count, void *userdata);
@@ -69,12 +76,15 @@ int lpValidateIntegrity(unsigned char *lp, size_t size, int deep,
unsigned char *lpValidateFirst(unsigned char *lp);
int lpValidateNext(unsigned char *lp, unsigned char **pp, size_t lpbytes);
unsigned int lpCompare(unsigned char *p, unsigned char *s, uint32_t slen);
void lpRandomPair(unsigned char *lp, unsigned long total_count, listpackEntry *key, listpackEntry *val);
void lpRandomPairs(unsigned char *lp, unsigned int count, listpackEntry *keys, listpackEntry *vals);
unsigned int lpRandomPairsUnique(unsigned char *lp, unsigned int count, listpackEntry *keys, listpackEntry *vals);
void lpRandomPair(unsigned char *lp, unsigned long total_count,
listpackEntry *key, listpackEntry *val, int tuple_len);
void lpRandomPairs(unsigned char *lp, unsigned int count,
listpackEntry *keys, listpackEntry *vals, int tuple_len);
unsigned int lpRandomPairsUnique(unsigned char *lp, unsigned int count,
listpackEntry *keys, listpackEntry *vals, int tuple_len);
void lpRandomEntries(unsigned char *lp, unsigned int count, listpackEntry *entries);
unsigned char *lpNextRandom(unsigned char *lp, unsigned char *p, unsigned int *index,
unsigned int remaining, int even_only);
unsigned int remaining, int tuple_len);
int lpSafeToAdd(unsigned char* lp, size_t add);
void lpRepr(unsigned char *lp);
src/module.c
+26 -11
View File
@@ -745,7 +745,7 @@ int moduleDelKeyIfEmpty(RedisModuleKey *key) {
case OBJ_LIST: isempty = listTypeLength(o) == 0; break;
case OBJ_SET: isempty = setTypeSize(o) == 0; break;
case OBJ_ZSET: isempty = zsetLength(o) == 0; break;
case OBJ_HASH: isempty = hashTypeLength(o) == 0; break;
case OBJ_HASH: isempty = hashTypeLength(o, 0) == 0; break;
case OBJ_STREAM: isempty = streamLength(o) == 0; break;
default: isempty = 0;
}
@@ -4168,7 +4168,7 @@ size_t RM_ValueLength(RedisModuleKey *key) {
case OBJ_LIST: return listTypeLength(key->value);
case OBJ_SET: return setTypeSize(key->value);
case OBJ_ZSET: return zsetLength(key->value);
case OBJ_HASH: return hashTypeLength(key->value);
case OBJ_HASH: return hashTypeLength(key->value, 0); /* OPEN: To subtract expired fields? */
case OBJ_STREAM: return streamLength(key->value);
default: return 0;
}
@@ -5271,7 +5271,10 @@ int RM_HashSet(RedisModuleKey *key, int flags, ...) {
/* Handle XX and NX */
if (flags & (REDISMODULE_HASH_XX|REDISMODULE_HASH_NX)) {
int exists = hashTypeExists(key->value, field->ptr);
int isHashDeleted;
int exists = hashTypeExists(key->db, key->value, field->ptr, &isHashDeleted);
/* hash-field-expiration is not exposed to modules */
serverAssert(isHashDeleted == 0);
if (((flags & REDISMODULE_HASH_XX) && !exists) ||
((flags & REDISMODULE_HASH_NX) && exists))
{
@@ -5282,7 +5285,7 @@ int RM_HashSet(RedisModuleKey *key, int flags, ...) {
/* Handle deletion if value is REDISMODULE_HASH_DELETE. */
if (value == REDISMODULE_HASH_DELETE) {
count += hashTypeDelete(key->value, field->ptr);
count += hashTypeDelete(key->value, field->ptr, 1);
if (flags & REDISMODULE_HASH_CFIELDS) decrRefCount(field);
continue;
}
@@ -5295,8 +5298,8 @@ int RM_HashSet(RedisModuleKey *key, int flags, ...) {
low_flags |= HASH_SET_TAKE_FIELD;
robj *argv[2] = {field,value};
hashTypeTryConversion(key->value,argv,0,1);
int updated = hashTypeSet(key->value, field->ptr, value->ptr, low_flags);
hashTypeTryConversion(key->db,key->value,argv,0,1);
int updated = hashTypeSet(key->db, key->value, field->ptr, value->ptr, low_flags);
count += (flags & REDISMODULE_HASH_COUNT_ALL) ? 1 : updated;
/* If CFIELDS is active, SDS string ownership is now of hashTypeSet(),
@@ -5374,14 +5377,22 @@ int RM_HashGet(RedisModuleKey *key, int flags, ...) {
/* Query the hash for existence or value object. */
if (flags & REDISMODULE_HASH_EXISTS) {
existsptr = va_arg(ap,int*);
if (key->value)
*existsptr = hashTypeExists(key->value,field->ptr);
else
if (key->value) {
int isHashDeleted;
*existsptr = hashTypeExists(key->db, key->value, field->ptr, &isHashDeleted);
/* hash-field-expiration is not exposed to modules */
serverAssert(isHashDeleted == 0);
} else {
*existsptr = 0;
}
} else {
int isHashDeleted;
valueptr = va_arg(ap,RedisModuleString**);
if (key->value) {
*valueptr = hashTypeGetValueObject(key->value,field->ptr);
*valueptr = hashTypeGetValueObject(key->db,key->value,field->ptr, &isHashDeleted);
/* Currently hash-field-expiration is not exposed to modules */
serverAssert(isHashDeleted == 0);
if (*valueptr) {
robj *decoded = getDecodedObject(*valueptr);
decrRefCount(*valueptr);
@@ -11071,18 +11082,22 @@ static void moduleScanKeyCallback(void *privdata, const dictEntry *de) {
ScanKeyCBData *data = privdata;
sds key = dictGetKey(de);
robj *o = data->key->value;
robj *field = createStringObject(key, sdslen(key));
robj *field = NULL;
robj *value = NULL;
if (o->type == OBJ_SET) {
value = NULL;
} else if (o->type == OBJ_HASH) {
sds val = dictGetVal(de);
field = createStringObject(key, hfieldlen(key));
value = createStringObject(val, sdslen(val));
} else if (o->type == OBJ_ZSET) {
double *val = (double*)dictGetVal(de);
value = createStringObjectFromLongDouble(*val, 0);
}
/* if type is OBJ_HASH then key is of type hfield. Otherwise sds. */
if (!field) field = createStringObject(key, sdslen(key));
data->fn(data->key, field, value, data->user_data);
decrRefCount(field);
if (value) decrRefCount(value);
src/mstr.c
+524
View File
@@ -0,0 +1,524 @@
/*
* Copyright Redis Ltd. 2024 - present
*
* Licensed under your choice of the Redis Source Available License 2.0 (RSALv2)
* or the Server Side Public License v1 (SSPLv1).
*/
#include <string.h>
#include <assert.h>
#include "sdsalloc.h"
#include "mstr.h"
#include "stdio.h"
#define NULL_SIZE 1
static inline char mstrReqType(size_t string_size);
static inline int mstrHdrSize(char type);
static inline int mstrSumMetaLen(mstrKind *k, mstrFlags flags);
static inline size_t mstrAllocLen(const mstr s, struct mstrKind *kind);
/*** mstr API ***/
/* Create mstr without any metadata attached, based on string 'initStr'.
* - If initStr equals NULL, then only allocation will be made.
* - string of mstr is always null-terminated.
*/
mstr mstrNew(const char *initStr, size_t lenStr, int trymalloc) {
unsigned char *pInfo; /* pointer to mstr info field */
void *sh;
mstr s;
char type = mstrReqType(lenStr);
int mstrHdr = mstrHdrSize(type);
assert(lenStr + mstrHdr + 1 > lenStr); /* Catch size_t overflow */
size_t len = mstrHdr + lenStr + NULL_SIZE;
sh = trymalloc? s_trymalloc(len) : s_malloc(len);
if (sh == NULL) return NULL;
s = (char*)sh + mstrHdr;
pInfo = ((unsigned char*)s) - 1;
switch(type) {
case MSTR_TYPE_5: {
*pInfo = CREATE_MSTR_INFO(lenStr, 0 /*ismeta*/, type);
break;
}
case MSTR_TYPE_8: {
MSTR_HDR_VAR(8,s);
*pInfo = CREATE_MSTR_INFO(0 /*unused*/, 0 /*ismeta*/, type);
sh->len = lenStr;
break;
}
case MSTR_TYPE_16: {
MSTR_HDR_VAR(16,s);
*pInfo = CREATE_MSTR_INFO(0 /*unused*/, 0 /*ismeta*/, type);
sh->len = lenStr;
break;
}
case MSTR_TYPE_64: {
MSTR_HDR_VAR(64,s);
*pInfo = CREATE_MSTR_INFO(0 /*unused*/, 0 /*ismeta*/, type);
sh->len = lenStr;
break;
}
}
if (initStr && lenStr)
memcpy(s, initStr, lenStr);
s[lenStr] = '\0';
return s;
}
/* Creates mstr with given string. Reserve space for metadata.
*
* Note: mstrNew(s,l) and mstrNewWithMeta(s,l,0) are not the same. The first allocates
* just string. The second allocates a string with flags (yet without any metadata
* structures allocated).
*/
mstr mstrNewWithMeta(struct mstrKind *kind, const char *initStr, size_t lenStr, mstrFlags metaFlags, int trymalloc) {
unsigned char *pInfo; /* pointer to mstr info field */
char *allocMstr;
mstr mstrPtr;
char type = mstrReqType(lenStr);
int mstrHdr = mstrHdrSize(type);
int sumMetaLen = mstrSumMetaLen(kind, metaFlags);
/* mstrSumMetaLen() + sizeof(mstrFlags) + sizeof(mstrhdrX) + lenStr */
size_t allocLen = sumMetaLen + sizeof(mstrFlags) + mstrHdr + lenStr + NULL_SIZE;
allocMstr = trymalloc? s_trymalloc(allocLen) : s_malloc(allocLen);
if (allocMstr == NULL) return NULL;
/* metadata is located at the beginning of the allocation, then meta-flags and lastly the string */
mstrFlags *pMetaFlags = (mstrFlags *) (allocMstr + sumMetaLen) ;
mstrPtr = ((char*) pMetaFlags) + sizeof(mstrFlags) + mstrHdr;
pInfo = ((unsigned char*)mstrPtr) - 1;
switch(type) {
case MSTR_TYPE_5: {
*pInfo = CREATE_MSTR_INFO(lenStr, 1 /*ismeta*/, type);
break;
}
case MSTR_TYPE_8: {
MSTR_HDR_VAR(8, mstrPtr);
sh->len = lenStr;
*pInfo = CREATE_MSTR_INFO(0 /*unused*/, 1 /*ismeta*/, type);
break;
}
case MSTR_TYPE_16: {
MSTR_HDR_VAR(16, mstrPtr);
sh->len = lenStr;
*pInfo = CREATE_MSTR_INFO(0 /*unused*/, 1 /*ismeta*/, type);
break;
}
case MSTR_TYPE_64: {
MSTR_HDR_VAR(64, mstrPtr);
sh->len = lenStr;
*pInfo = CREATE_MSTR_INFO(0 /*unused*/, 1 /*ismeta*/, type);
break;
}
}
*pMetaFlags = metaFlags;
if (initStr != NULL) memcpy(mstrPtr, initStr, lenStr);
mstrPtr[lenStr] = '\0';
return mstrPtr;
}
/* Create copy of mstr. Flags can be modified. For each metadata flag, if
* same flag is set on both, then copy its metadata. */
mstr mstrNewCopy(struct mstrKind *kind, mstr src, mstrFlags newFlags) {
mstr dst;
/* if no flags are set, then just copy the string */
if (newFlags == 0) return mstrNew(src, mstrlen(src), 0);
dst = mstrNewWithMeta(kind, src, mstrlen(src), newFlags, 0);
memcpy(dst, src, mstrlen(src) + 1);
/* if metadata is attached to src, then selectively copy metadata */
if (mstrIsMetaAttached(src)) {
mstrFlags *pFlags1 = mstrFlagsRef(src),
*pFlags2 = mstrFlagsRef(dst);
mstrFlags flags1Shift = *pFlags1,
flags2Shift = *pFlags2;
unsigned char *at1 = ((unsigned char *) pFlags1),
*at2 = ((unsigned char *) pFlags2);
/* if the flag is set on both, then copy the metadata */
for (int i = 0; flags1Shift != 0; ++i) {
int isFlag1Set = flags1Shift & 0x1;
int isFlag2Set = flags2Shift & 0x1;
if (isFlag1Set) at1 -= kind->metaSize[i];
if (isFlag2Set) at2 -= kind->metaSize[i];
if (isFlag1Set && isFlag2Set)
memcpy(at2, at1, kind->metaSize[i]);
flags1Shift >>= 1;
flags2Shift >>= 1;
}
}
return dst;
}
/* Free mstring. Note, mstrKind is required to eval sizeof metadata and find start
* of allocation but if mstrIsMetaAttached(s) is false, you can pass NULL as well.
*/
void mstrFree(struct mstrKind *kind, mstr s) {
if (s != NULL)
s_free(mstrGetAllocPtr(kind, s));
}
/* return ref to metadata flags. Useful to modify directly flags which doesn't
* include metadata payload */
mstrFlags *mstrFlagsRef(mstr s) {
switch(s[-1]&MSTR_TYPE_MASK) {
case MSTR_TYPE_5:
return ((mstrFlags *) (s - sizeof(struct mstrhdr5))) - 1;
case MSTR_TYPE_8:
return ((mstrFlags *) (s - sizeof(struct mstrhdr8))) - 1;
case MSTR_TYPE_16:
return ((mstrFlags *) (s - sizeof(struct mstrhdr16))) - 1;
default: /* MSTR_TYPE_64: */
return ((mstrFlags *) (s - sizeof(struct mstrhdr64))) - 1;
}
}
/* Return a reference to corresponding metadata of the specified metadata flag
* index (flagIdx). If the metadata doesn't exist, it still returns a reference
* to the starting location where it would have been written among other metadatas.
* To verify if `flagIdx` of some metadata is attached, use `mstrGetFlag(s, flagIdx)`.
*/
void *mstrMetaRef(mstr s, struct mstrKind *kind, int flagIdx) {
int metaOffset = 0;
/* start iterating from flags backward */
mstrFlags *pFlags = mstrFlagsRef(s);
mstrFlags tmp = *pFlags;
for (int i = 0 ; i <= flagIdx ; ++i) {
if (tmp & 0x1) metaOffset += kind->metaSize[i];
tmp >>= 1;
}
return ((char *)pFlags) - metaOffset;
}
/* mstr layout: [meta-data#N]...[meta-data#0][mstrFlags][mstrhdr][string][null] */
void *mstrGetAllocPtr(struct mstrKind *kind, mstr str) {
if (!mstrIsMetaAttached(str))
return (char*)str - mstrHdrSize(str[-1]);
int totalMetaLen = mstrSumMetaLen(kind, *mstrFlagsRef(str));
return (char*)str - mstrHdrSize(str[-1]) - sizeof(mstrFlags) - totalMetaLen;
}
/* Prints in the following fashion:
* [0x7f8bd8816017] my_mstr: foo (strLen=3, mstrLen=11, isMeta=1, metaFlags=0x1)
* [0x7f8bd8816010] >> meta[0]: 0x78 0x56 0x34 0x12 (metaLen=4)
*/
void mstrPrint(mstr s, struct mstrKind *kind, int verbose) {
mstrFlags mflags, tmp;
int isMeta = mstrIsMetaAttached(s);
tmp = mflags = (isMeta) ? *mstrFlagsRef(s) : 0;
if (!isMeta) {
printf("[%p] %s: %s (strLen=%zu, mstrLen=%zu, isMeta=0)\n",
(void *)s, kind->name, s, mstrlen(s), mstrAllocLen(s, kind));
return;
}
printf("[%p] %s: %s (strLen=%zu, mstrLen=%zu, isMeta=1, metaFlags=0x%x)\n",
(void *)s, kind->name, s, mstrlen(s), mstrAllocLen(s, kind), mflags);
if (verbose) {
for (unsigned int i = 0 ; i < NUM_MSTR_FLAGS ; ++i) {
if (tmp & 0x1) {
int mSize = kind->metaSize[i];
void *mRef = mstrMetaRef(s, kind, i);
printf("[%p] >> meta[%d]:", mRef, i);
for (int j = 0 ; j < mSize ; ++j) {
printf(" 0x%02x", ((unsigned char *) mRef)[j]);
}
printf(" (metaLen=%d)\n", mSize);
}
tmp >>= 1;
}
}
}
/* return length of the string (ignoring metadata attached) */
size_t mstrlen(const mstr s) {
unsigned char info = s[-1];
switch(info & MSTR_TYPE_MASK) {
case MSTR_TYPE_5:
return MSTR_TYPE_5_LEN(info);
case MSTR_TYPE_8:
return MSTR_HDR(8,s)->len;
case MSTR_TYPE_16:
return MSTR_HDR(16,s)->len;
default: /* MSTR_TYPE_64: */
return MSTR_HDR(64,s)->len;
}
}
/*** mstr internals ***/
static inline int mstrSumMetaLen(mstrKind *k, mstrFlags flags) {
int total = 0;
int i = 0 ;
while (flags) {
total += (flags & 0x1) ? k->metaSize[i] : 0;
flags >>= 1;
++i;
}
return total;
}
/* mstrSumMetaLen() + sizeof(mstrFlags) + sizeof(mstrhdrX) + strlen + '\0' */
static inline size_t mstrAllocLen(const mstr s, struct mstrKind *kind) {
int hdrlen;
mstrFlags *pMetaFlags;
size_t strlen = 0;
int isMeta = mstrIsMetaAttached(s);
unsigned char info = s[-1];
switch(info & MSTR_TYPE_MASK) {
case MSTR_TYPE_5:
strlen = MSTR_TYPE_5_LEN(info);
hdrlen = sizeof(struct mstrhdr5);
pMetaFlags = ((mstrFlags *) MSTR_HDR(5, s)) - 1;
break;
case MSTR_TYPE_8:
strlen = MSTR_HDR(8,s)->len;
hdrlen = sizeof(struct mstrhdr8);
pMetaFlags = ((mstrFlags *) MSTR_HDR(8, s)) - 1;
break;
case MSTR_TYPE_16:
strlen = MSTR_HDR(16,s)->len;
hdrlen = sizeof(struct mstrhdr16);
pMetaFlags = ((mstrFlags *) MSTR_HDR(16, s)) - 1;
break;
default: /* MSTR_TYPE_64: */
strlen = MSTR_HDR(64,s)->len;
hdrlen = sizeof(struct mstrhdr64);
pMetaFlags = ((mstrFlags *) MSTR_HDR(64, s)) - 1;
break;
}
return hdrlen + strlen + NULL_SIZE + ((isMeta) ? (mstrSumMetaLen(kind, *pMetaFlags) + sizeof(mstrFlags)) : 0);
}
/* returns pointer to the beginning of malloc() of mstr */
void *mstrGetStartAlloc(mstr s, struct mstrKind *kind) {
int hdrlen;
mstrFlags *pMetaFlags;
int isMeta = mstrIsMetaAttached(s);
switch(s[-1]&MSTR_TYPE_MASK) {
case MSTR_TYPE_5:
hdrlen = sizeof(struct mstrhdr5);
pMetaFlags = ((mstrFlags *) MSTR_HDR(5, s)) - 1;
break;
case MSTR_TYPE_8:
hdrlen = sizeof(struct mstrhdr8);
pMetaFlags = ((mstrFlags *) MSTR_HDR(8, s)) - 1;
break;
case MSTR_TYPE_16:
hdrlen = sizeof(struct mstrhdr16);
pMetaFlags = ((mstrFlags *) MSTR_HDR(16, s)) - 1;
break;
default: /* MSTR_TYPE_64: */
hdrlen = sizeof(struct mstrhdr64);
pMetaFlags = ((mstrFlags *) MSTR_HDR(64, s)) - 1;
break;
}
return (char *) s - hdrlen - ((isMeta) ? (mstrSumMetaLen(kind, *pMetaFlags) + sizeof(mstrFlags)) : 0);
}
static inline int mstrHdrSize(char type) {
switch(type&MSTR_TYPE_MASK) {
case MSTR_TYPE_5:
return sizeof(struct mstrhdr5);
case MSTR_TYPE_8:
return sizeof(struct mstrhdr8);
case MSTR_TYPE_16:
return sizeof(struct mstrhdr16);
case MSTR_TYPE_64:
return sizeof(struct mstrhdr64);
}
return 0;
}
static inline char mstrReqType(size_t string_size) {
if (string_size < 1<<5)
return MSTR_TYPE_5;
if (string_size < 1<<8)
return MSTR_TYPE_8;
if (string_size < 1<<16)
return MSTR_TYPE_16;
return MSTR_TYPE_64;
}
#ifdef REDIS_TEST
#include <stdlib.h>
#include <assert.h>
#include "testhelp.h"
#include "limits.h"
#ifndef UNUSED
#define UNUSED(x) (void)(x)
#endif
/* Challenge mstr with metadata interesting enough that can include the case of hfield and hkey and more */
#define B(idx) (1<<(idx))
#define META_IDX_MYMSTR_TTL4 0
#define META_IDX_MYMSTR_TTL8 1
#define META_IDX_MYMSTR_TYPE_ENC_LRU 2 // 4Bbit type, 4bit encoding, 24bits lru
#define META_IDX_MYMSTR_VALUE_PTR 3
#define META_IDX_MYMSTR_FLAG_NO_META 4
#define TEST_CONTEXT(context) printf("\nContext: %s \n", context);
int mstrTest(int argc, char **argv, int flags) {
UNUSED(argc);
UNUSED(argv);
UNUSED(flags);
struct mstrKind kind_mymstr = {
.name = "my_mstr",
.metaSize[META_IDX_MYMSTR_TTL4] = 4,
.metaSize[META_IDX_MYMSTR_TTL8] = 8,
.metaSize[META_IDX_MYMSTR_TYPE_ENC_LRU] = 4,
.metaSize[META_IDX_MYMSTR_VALUE_PTR] = 8,
.metaSize[META_IDX_MYMSTR_FLAG_NO_META] = 0,
};
TEST_CONTEXT("Create simple short mstr")
{
char *str = "foo";
mstr s = mstrNew(str, strlen(str), 0);
size_t expStrLen = strlen(str);
test_cond("Verify str length and alloc length",
mstrAllocLen(s, NULL) == (1 + expStrLen + 1) && /* mstrhdr5 + str + null */
mstrlen(s) == expStrLen && /* expected strlen(str) */
memcmp(s, str, expStrLen + 1) == 0);
mstrFree(&kind_mymstr, s);
}
TEST_CONTEXT("Create simple 40 bytes mstr")
{
char *str = "0123456789012345678901234567890123456789"; // 40 bytes
mstr s = mstrNew(str, strlen(str), 0);
test_cond("Verify str length and alloc length",
mstrAllocLen(s, NULL) == (3 + 40 + 1) && /* mstrhdr8 + str + null */
mstrlen(s) == 40 &&
memcmp(s,str,40) == 0);
mstrFree(&kind_mymstr, s);
}
TEST_CONTEXT("Create mstr with random characters")
{
long unsigned int i;
char str[66000];
for (i = 0 ; i < sizeof(str) ; ++i) str[i] = rand() % 256;
size_t len[] = { 31, 32, 33, 255, 256, 257, 65535, 65536, 65537, 66000};
for (i = 0 ; i < sizeof(len) / sizeof(len[0]) ; ++i) {
char title[100];
mstr s = mstrNew(str, len[i], 0);
size_t mstrhdrSize = (len[i] < 1<<5) ? sizeof(struct mstrhdr5) :
(len[i] < 1<<8) ? sizeof(struct mstrhdr8) :
(len[i] < 1<<16) ? sizeof(struct mstrhdr16) :
sizeof(struct mstrhdr64);
snprintf(title, sizeof(title), "Verify string of length %zu", len[i]);
test_cond(title,
mstrAllocLen(s, NULL) == (mstrhdrSize + len[i] + 1) && /* mstrhdrX + str + null */
mstrlen(s) == len[i] &&
memcmp(s,str,len[i]) == 0);
mstrFree(&kind_mymstr, s);
}
}
TEST_CONTEXT("Create short mstr with TTL4")
{
uint32_t *ttl;
mstr s = mstrNewWithMeta(&kind_mymstr,
"foo",
strlen("foo"),
B(META_IDX_MYMSTR_TTL4), /* allocate with TTL4 metadata */
0);
ttl = mstrMetaRef(s, &kind_mymstr, META_IDX_MYMSTR_TTL4);
*ttl = 0x12345678;
test_cond("Verify memory-allocation and string lengths",
mstrAllocLen(s, &kind_mymstr) == (1 + 3 + 2 + 1 + 4) && /* mstrhdr5 + str + null + mstrFlags + TLL */
mstrlen(s) == 3);
unsigned char expMem[] = {0xFF, 0xFF, 0xFF, 0xFF, 0x01, 0x00, 0x1c, 'f', 'o', 'o', '\0' };
uint32_t value = 0x12345678;
memcpy(expMem, &value, sizeof(uint32_t));
test_cond("Verify string and TTL4 payload", memcmp(
mstrMetaRef(s, &kind_mymstr, 0) , expMem, sizeof(expMem)) == 0);
test_cond("Verify mstrIsMetaAttached() function works", mstrIsMetaAttached(s) != 0);
mstrFree(&kind_mymstr, s);
}
TEST_CONTEXT("Create short mstr with TTL4 and value ptr ")
{
mstr s = mstrNewWithMeta(&kind_mymstr, "foo", strlen("foo"),
B(META_IDX_MYMSTR_TTL4) | B(META_IDX_MYMSTR_VALUE_PTR), 0);
*((uint32_t *) (mstrMetaRef(s, &kind_mymstr,
META_IDX_MYMSTR_TTL4))) = 0x12345678;
test_cond("Verify length and alloc length",
mstrAllocLen(s, &kind_mymstr) == (1 + 3 + 1 + 2 + 4 + 8) && /* mstrhdr5 + str + null + mstrFlags + TLL + PTR */
mstrlen(s) == 3);
mstrFree(&kind_mymstr, s);
}
TEST_CONTEXT("Copy mstr and add it TTL4")
{
mstr s1 = mstrNew("foo", strlen("foo"), 0);
mstr s2 = mstrNewCopy(&kind_mymstr, s1, B(META_IDX_MYMSTR_TTL4));
*((uint32_t *) (mstrMetaRef(s2, &kind_mymstr, META_IDX_MYMSTR_TTL4))) = 0x12345678;
test_cond("Verify new mstr includes TTL4",
mstrAllocLen(s2, &kind_mymstr) == (1 + 3 + 1 + 2 + 4) && /* mstrhdr5 + str + null + mstrFlags + TTL4 */
mstrlen(s2) == 3 && /* 'foo' = 3bytes */
memcmp(s2, "foo\0", 4) == 0);
mstr s3 = mstrNewCopy(&kind_mymstr, s2, B(META_IDX_MYMSTR_TTL4));
unsigned char expMem[] = { 0xFF, 0xFF, 0xFF, 0xFF, 0x1, 0x0, 0x1c, 'f', 'o', 'o', '\0' };
uint32_t value = 0x12345678;
memcpy(expMem, &value, sizeof(uint32_t));
char *ppp = mstrGetStartAlloc(s3, &kind_mymstr);
test_cond("Verify string and TTL4 payload",
memcmp(ppp, expMem, sizeof(expMem)) == 0);
mstrPrint(s3, &kind_mymstr, 1);
mstrFree(&kind_mymstr, s1);
mstrFree(&kind_mymstr, s2);
mstrFree(&kind_mymstr, s3);
}
return 0;
}
#endif
src/mstr.h
+226
View File
@@ -0,0 +1,226 @@
/*
* Copyright Redis Ltd. 2024 - present
*
* Licensed under your choice of the Redis Source Available License 2.0 (RSALv2)
* or the Server Side Public License v1 (SSPLv1).
*
*
* WHAT IS MSTR (M-STRING)?
* ------------------------
* mstr stands for immutable string with optional metadata attached.
*
* sds string is widely used across the system and serves as a general purpose
* container to hold data. The need to optimize memory and aggregate strings
* along with metadata and store it into Redis data-structures as single bulk keep
* reoccur. One thought might be, why not to extend sds to support metadata. The
* answer is that sds is mutable string in its nature, with wide API (split, join,
* etc.). Pushing metadata logic into sds will make it very fragile, and complex
* to maintain.
*
* Another idea involved using a simple struct with flags and a dynamic buf[] at the
* end. While this could be viable, it introduces considerable complexity and would
* need maintenance across different contexts.
*
* As an alternative, we introduce a new implementation of immutable strings,
* with limited API, and with the option to attach metadata. The representation
* of the string, without any metadata, in its basic form, resembles SDS but
* without the API to manipulate the string. Only to attach metadata to it. The
* following diagram shows the memory layout of mstring (mstrhdr8) when no
* metadata is attached:
*
* +----------------------------------------------+
* | mstrhdr8 | c-string | |
* +--------------------------------+-------------+
* |8b |2b |1b |5b |?bytes |8b|
* | Len | Type |m-bit=0 | Unused | String |\0|
* +----------------------------------------------+
* ^
* |
* mstrNew() returns pointer to here --+
*
* If metadata-flag is set, depicted in diagram above as m-bit in the diagram,
* then the header will be preceded with additional 16 bits of metadata flags such
* that if i'th bit is set, then the i'th metadata structure is attached to the
* mstring. The metadata layout and their sizes are defined by mstrKind structure
* (More below).
*
* The following diagram shows the memory layout of mstr (mstrhdr8) when 3 bits in mFlags
* are set to indicate that 3 fields of metadata are attached to the mstring at the
* beginning.
*
* +-------------------------------------------------------------------------------+
* | METADATA FIELDS | mflags | mstrhdr8 | c-string | |
* +-----------------------+--------+--------------------------------+-------------+
* |?bytes |?bytes |?bytes |16b |8b |2b |1b |5b |?bytes |8b|
* | Meta3 | Meta2 | Meta0 | 0x1101 | Len | Type |m-bit=1 | Unused | String |\0|
* +-------------------------------------------------------------------------------+
* ^
* |
* mstrNewWithMeta() returns pointer to here --+
*
* mstr allows to define different kinds (groups) of mstrings, each with its
* own unique metadata layout. For example, in case of hash-fields, all instances of
* it can optionally have TTL metadata attached to it. This is achieved by first
* prototyping a single mstrKind structure that defines the metadata layout and sizes
* of this specific kind. Now each hash-field instance has still the freedom to
* attach or not attach the metadata to it, and metadata flags (mFlags) of the
* instance will reflect this decision.
*
* In the future, the keys of Redis keyspace can be another kind of mstring that
* has TTL, LRU or even dictEntry metadata embedded into. Unlike vptr in c++, this
* struct won't be attached to mstring but will be passed as yet another argument
* to API, to save memory. In addition, each instance of a given mstrkind can hold
* any subset of metadata and the 8 bits of metadata-flags will reflect it.
*
* The following example shows how to define mstrKind for possible future keyspace
* that aggregates several keyspace related metadata into one compact, singly
* allocated, mstring.
*
* typedef enum HkeyMetaFlags {
* HKEY_META_VAL_REF_COUNT = 0, // refcount
* HKEY_META_VAL_REF = 1, // Val referenced
* HKEY_META_EXPIRE = 2, // TTL and more
* HKEY_META_TYPE_ENC_LRU = 3, // TYPE + LRU + ENC
* HKEY_META_DICT_ENT_NEXT = 4, // Next dict entry
* // Following two must be together and in this order
* HKEY_META_VAL_EMBED8 = 5, // Val embedded, max 7 bytes
* HKEY_META_VAL_EMBED16 = 6, // Val embedded, max 15 bytes (23 with EMBED8)
* } HkeyMetaFlags;
*
* mstrKind hkeyKind = {
* .name = "hkey",
* .metaSize[HKEY_META_VAL_REF_COUNT] = 4,
* .metaSize[HKEY_META_VAL_REF] = 8,
* .metaSize[HKEY_META_EXPIRE] = sizeof(ExpireMeta),
* .metaSize[HKEY_META_TYPE_ENC_LRU] = 8,
* .metaSize[HKEY_META_DICT_ENT_NEXT] = 8,
* .metaSize[HKEY_META_VAL_EMBED8] = 8,
* .metaSize[HKEY_META_VAL_EMBED16] = 16,
* };
*
* MSTR-ALIGNMENT
* --------------
* There are two types of alignments to take into consideration:
* 1. Alignment of the metadata.
* 2. Alignment of returned mstr pointer
*
* 1) As the metadatas layout are reversed to their enumeration, it is recommended
* to put metadata with "better" alignment first in memory layout (enumerated
* last) and the worst, or those that simply don't require any alignment will be
* last in memory layout (enumerated first). This is similar the to the applied
* consideration when defining new struct in C. Note also that each metadata
* might either be attached to mstr or not which complicates the design phase
* of a new mstrKind a little.
*
* In the example above, HKEY_META_VAL_REF_COUNT, with worst alignment of 4
* bytes, is enumerated first, and therefore, will be last in memory layout.
*
* 2) Few optimizations in Redis rely on the fact that sds address is always an odd
* pointer. We can achieve the same with a little effort. It was already taken
* care that all headers of type mstrhdrX has odd size. With that in mind, if
* a new kind of mstr is required to be limited to odd addresses, then we must
* make sure that sizes of all related metadatas that are defined in mstrKind
* are even in size.
*/
#ifndef __MSTR_H
#define __MSTR_H
#include <sys/types.h>
#include <stdarg.h>
#include <stdint.h>
/* Selective copy of ifndef from server.h instead of including it */
#ifndef static_assert
#define static_assert(expr, lit) extern char __static_assert_failure[(expr) ? 1:-1]
#endif
#define MSTR_TYPE_5 0
#define MSTR_TYPE_8 1
#define MSTR_TYPE_16 2
#define MSTR_TYPE_64 3
#define MSTR_TYPE_MASK 3
#define MSTR_TYPE_BITS 2
#define MSTR_META_MASK 4
#define MSTR_HDR(T,s) ((struct mstrhdr##T *)((s)-(sizeof(struct mstrhdr##T))))
#define MSTR_HDR_VAR(T,s) struct mstrhdr##T *sh = (void*)((s)-(sizeof(struct mstrhdr##T)));
#define MSTR_META_BITS 1 /* is metadata attached? */
#define MSTR_TYPE_5_LEN(f) ((f) >> (MSTR_TYPE_BITS + MSTR_META_BITS))
#define CREATE_MSTR_INFO(len, ismeta, type) ( (((len<<MSTR_META_BITS) + ismeta) << (MSTR_TYPE_BITS)) | type )
/* mimic plain c-string */
typedef char *mstr;
/* Flags that can be set on mstring to indicate for attached metadata. It is
* */
typedef uint16_t mstrFlags;
struct __attribute__ ((__packed__)) mstrhdr5 {
unsigned char info; /* 2 lsb of type, 1 metadata, and 5 msb of string length */
char buf[];
};
struct __attribute__ ((__packed__)) mstrhdr8 {
uint8_t unused; /* To achieve odd size header (See comment above) */
uint8_t len;
unsigned char info; /* 2 lsb of type, 6 unused bits */
char buf[];
};
struct __attribute__ ((__packed__)) mstrhdr16 {
uint16_t len;
unsigned char info; /* 2 lsb of type, 6 unused bits */
char buf[];
};
struct __attribute__ ((__packed__)) mstrhdr64 {
uint64_t len;
unsigned char info; /* 2 lsb of type, 6 unused bits */
char buf[];
};
#define NUM_MSTR_FLAGS (sizeof(mstrFlags)*8)
/* mstrKind is used to define a kind (a group) of mstring with its own metadata layout */
typedef struct mstrKind {
const char *name;
int metaSize[NUM_MSTR_FLAGS];
} mstrKind;
mstr mstrNew(const char *initStr, size_t lenStr, int trymalloc);
mstr mstrNewWithMeta(struct mstrKind *kind, const char *initStr, size_t lenStr, mstrFlags flags, int trymalloc);
mstr mstrNewCopy(struct mstrKind *kind, mstr src, mstrFlags newFlags);
void *mstrGetAllocPtr(struct mstrKind *kind, mstr str);
void mstrFree(struct mstrKind *kind, mstr s);
mstrFlags *mstrFlagsRef(mstr s);
void *mstrMetaRef(mstr s, struct mstrKind *kind, int flagIdx);
size_t mstrlen(const mstr s);
/* return non-zero if metadata is attached to mstring */
static inline int mstrIsMetaAttached(mstr s) { return s[-1] & MSTR_META_MASK; }
/* return whether if a specific flag-index is set */
static inline int mstrGetFlag(mstr s, int flagIdx) { return *mstrFlagsRef(s) & (1 << flagIdx); }
/* DEBUG */
void mstrPrint(mstr s, struct mstrKind *kind, int verbose);
/* See comment above about MSTR-ALIGNMENT(2) */
static_assert(sizeof(struct mstrhdr5 ) % 2 == 1, "must be odd");
static_assert(sizeof(struct mstrhdr8 ) % 2 == 1, "must be odd");
static_assert(sizeof(struct mstrhdr16 ) % 2 == 1, "must be odd");
static_assert(sizeof(struct mstrhdr64 ) % 2 == 1, "must be odd");
static_assert(sizeof(mstrFlags ) % 2 == 0, "must be even to keep mstr pointer odd");
#ifdef REDIS_TEST
int mstrTest(int argc, char *argv[], int flags);
#endif
#endif
src/networking.c
+23 -4
View File
@@ -31,6 +31,14 @@ size_t sdsZmallocSize(sds s) {
return zmalloc_size(sh);
}
/* Return the size consumed from the allocator, for the specified hfield with
* metadata (mstr), including internal fragmentation. This function is used in
* order to compute the client output buffer size. */
size_t hfieldZmallocSize(hfield s) {
void *sh = hfieldGetAllocPtr(s);
return zmalloc_size(sh);
}
/* Return the amount of memory used by the sds string at object->ptr
* for a string object. This includes internal fragmentation. */
size_t getStringObjectSdsUsedMemory(robj *o) {
@@ -3749,7 +3757,9 @@ void replaceClientCommandVector(client *c, int argc, robj **argv) {
* 1. Make sure there are no "holes" and all the arguments are set.
* 2. If the original argument vector was longer than the one we
* want to end with, it's up to the caller to set c->argc and
* free the no longer used objects on c->argv. */
* free the no longer used objects on c->argv.
* 3. To remove argument at i'th index, pass NULL as new value
*/
void rewriteClientCommandArgument(client *c, int i, robj *newval) {
robj *oldval;
retainOriginalCommandVector(c);
@@ -3767,9 +3777,18 @@ void rewriteClientCommandArgument(client *c, int i, robj *newval) {
}
oldval = c->argv[i];
if (oldval) c->argv_len_sum -= getStringObjectLen(oldval);
if (newval) c->argv_len_sum += getStringObjectLen(newval);
c->argv[i] = newval;
incrRefCount(newval);
if (newval) {
c->argv[i] = newval;
incrRefCount(newval);
c->argv_len_sum += getStringObjectLen(newval);
} else {
/* move the remaining arguments one step left */
for (int j = i+1; j < c->argc; j++) {
c->argv[j-1] = c->argv[j];
}
c->argv[--c->argc] = NULL;
}
if (oldval) decrRefCount(oldval);
/* If this is the command name make sure to fix c->cmd. */
src/notify.c
+1 -1
View File
@@ -80,7 +80,7 @@ sds keyspaceEventsFlagsToString(int flags) {
* 'event' is a C string representing the event name.
* 'key' is a Redis object representing the key name.
* 'dbid' is the database ID where the key lives. */
void notifyKeyspaceEvent(int type, char *event, robj *key, int dbid) {
void notifyKeyspaceEvent(int type, const char *event, robj *key, int dbid) {
sds chan;
robj *chanobj, *eventobj;
int len = -1;
src/object.c
+12 -16
View File
@@ -333,17 +333,7 @@ void freeZsetObject(robj *o) {
}
void freeHashObject(robj *o) {
switch (o->encoding) {
case OBJ_ENCODING_HT:
dictRelease((dict*) o->ptr);
break;
case OBJ_ENCODING_LISTPACK:
lpFree(o->ptr);
break;
default:
serverPanic("Unknown hash encoding type");
break;
}
hashTypeFree(o);
}
void freeModuleObject(robj *o) {
@@ -502,6 +492,9 @@ void dismissHashObject(robj *o, size_t size_hint) {
dismissMemory(d->ht_table[1], DICTHT_SIZE(d->ht_size_exp[1])*sizeof(dictEntry*));
} else if (o->encoding == OBJ_ENCODING_LISTPACK) {
dismissMemory(o->ptr, lpBytes((unsigned char*)o->ptr));
} else if (o->encoding == OBJ_ENCODING_LISTPACK_EX) {
listpackEx *lpt = o->ptr;
dismissMemory(lpt->lp, lpBytes((unsigned char*)lpt->lp));
} else {
serverPanic("Unknown hash encoding type");
}
@@ -939,6 +932,7 @@ char *strEncoding(int encoding) {
case OBJ_ENCODING_HT: return "hashtable";
case OBJ_ENCODING_QUICKLIST: return "quicklist";
case OBJ_ENCODING_LISTPACK: return "listpack";
case OBJ_ENCODING_LISTPACK_EX: return "listpackex";
case OBJ_ENCODING_INTSET: return "intset";
case OBJ_ENCODING_SKIPLIST: return "skiplist";
case OBJ_ENCODING_EMBSTR: return "embstr";
@@ -979,7 +973,6 @@ size_t streamRadixTreeMemoryUsage(rax *rax) {
* are checked and averaged to estimate the total size. */
#define OBJ_COMPUTE_SIZE_DEF_SAMPLES 5 /* Default sample size. */
size_t objectComputeSize(robj *key, robj *o, size_t sample_size, int dbid) {
sds ele, ele2;
dict *d;
dictIterator *di;
struct dictEntry *de;
@@ -1016,7 +1009,7 @@ size_t objectComputeSize(robj *key, robj *o, size_t sample_size, int dbid) {
di = dictGetIterator(d);
asize = sizeof(*o)+sizeof(dict)+(sizeof(struct dictEntry*)*dictBuckets(d));
while((de = dictNext(di)) != NULL && samples < sample_size) {
ele = dictGetKey(de);
sds ele = dictGetKey(de);
elesize += dictEntryMemUsage() + sdsZmallocSize(ele);
samples++;
}
@@ -1052,14 +1045,17 @@ size_t objectComputeSize(robj *key, robj *o, size_t sample_size, int dbid) {
} else if (o->type == OBJ_HASH) {
if (o->encoding == OBJ_ENCODING_LISTPACK) {
asize = sizeof(*o)+zmalloc_size(o->ptr);
} else if (o->encoding == OBJ_ENCODING_LISTPACK_EX) {
listpackEx *lpt = o->ptr;
asize = sizeof(*o) + zmalloc_size(lpt) + zmalloc_size(lpt->lp);
} else if (o->encoding == OBJ_ENCODING_HT) {
d = o->ptr;
di = dictGetIterator(d);
asize = sizeof(*o)+sizeof(dict)+(sizeof(struct dictEntry*)*dictBuckets(d));
while((de = dictNext(di)) != NULL && samples < sample_size) {
ele = dictGetKey(de);
ele2 = dictGetVal(de);
elesize += sdsZmallocSize(ele) + sdsZmallocSize(ele2);
hfield ele = dictGetKey(de);
sds ele2 = dictGetVal(de);
elesize += hfieldZmallocSize(ele) + sdsZmallocSize(ele2);
elesize += dictEntryMemUsage();
samples++;
}
src/rax.c
+35 -2
View File
@@ -173,11 +173,16 @@ raxNode *raxNewNode(size_t children, int datafield) {
/* Allocate a new rax and return its pointer. On out of memory the function
* returns NULL. */
rax *raxNew(void) {
rax *rax = rax_malloc(sizeof(*rax));
return raxNewWithMetadata(0);
}
/* Allocate a new rax with metadata */
rax *raxNewWithMetadata(int metaSize) {
rax *rax = rax_malloc(sizeof(*rax) + metaSize);
if (rax == NULL) return NULL;
rax->numele = 0;
rax->numnodes = 1;
rax->head = raxNewNode(0,0);
rax->head = raxNewNode(0, 0);
if (rax->head == NULL) {
rax_free(rax);
return NULL;
@@ -1210,6 +1215,25 @@ void raxRecursiveFree(rax *rax, raxNode *n, void (*free_callback)(void*)) {
rax->numnodes--;
}
/* Same as raxRecursiveFree() with context argument */
void raxRecursiveFreeWithCtx(rax *rax, raxNode *n,
void (*free_callback)(void *item, void *ctx), void *ctx) {
debugnode("free traversing",n);
int numchildren = n->iscompr ? 1 : n->size;
raxNode **cp = raxNodeLastChildPtr(n);
while(numchildren--) {
raxNode *child;
memcpy(&child,cp,sizeof(child));
raxRecursiveFreeWithCtx(rax,child,free_callback, ctx);
cp--;
}
debugnode("free depth-first",n);
if (free_callback && n->iskey && !n->isnull)
free_callback(raxGetData(n), ctx);
rax_free(n);
rax->numnodes--;
}
/* Free a whole radix tree, calling the specified callback in order to
* free the auxiliary data. */
void raxFreeWithCallback(rax *rax, void (*free_callback)(void*)) {
@@ -1218,6 +1242,15 @@ void raxFreeWithCallback(rax *rax, void (*free_callback)(void*)) {
rax_free(rax);
}
/* Free a whole radix tree, calling the specified callback in order to
* free the auxiliary data. */
void raxFreeWithCbAndContext(rax *rax,
void (*free_callback)(void *item, void *ctx), void *ctx) {
raxRecursiveFreeWithCtx(rax,rax->head,free_callback,ctx);
assert(rax->numnodes == 0);
rax_free(rax);
}
/* Free a whole radix tree. */
void raxFree(rax *rax) {
raxFreeWithCallback(rax,NULL);
src/rax.h
+5
View File
@@ -113,6 +113,7 @@ typedef struct rax {
raxNode *head;
uint64_t numele;
uint64_t numnodes;
void *metadata[];
} rax;
/* Stack data structure used by raxLowWalk() in order to, optionally, return
@@ -166,12 +167,16 @@ typedef struct raxIterator {
/* Exported API. */
rax *raxNew(void);
rax *raxNewWithMetadata(int metaSize);
int raxInsert(rax *rax, unsigned char *s, size_t len, void *data, void **old);
int raxTryInsert(rax *rax, unsigned char *s, size_t len, void *data, void **old);
int raxRemove(rax *rax, unsigned char *s, size_t len, void **old);
int raxFind(rax *rax, unsigned char *s, size_t len, void **value);
void raxFree(rax *rax);
void raxFreeWithCallback(rax *rax, void (*free_callback)(void*));
void raxFreeWithCbAndContext(rax *rax,
void (*free_callback)(void *item, void *ctx),
void *ctx);
void raxStart(raxIterator *it, rax *rt);
int raxSeek(raxIterator *it, const char *op, unsigned char *ele, size_t len);
int raxNext(raxIterator *it);
src/rdb.c
+391 -111
View File
@@ -268,8 +268,9 @@ int rdbEncodeInteger(long long value, unsigned char *enc) {
* The returned value changes according to the flags, see
* rdbGenericLoadStringObject() for more info. */
void *rdbLoadIntegerObject(rio *rdb, int enctype, int flags, size_t *lenptr) {
int plain = flags & RDB_LOAD_PLAIN;
int sds = flags & RDB_LOAD_SDS;
int plainFlag = flags & RDB_LOAD_PLAIN;
int sdsFlag = flags & RDB_LOAD_SDS;
int hfldFlag = flags & (RDB_LOAD_HFLD|RDB_LOAD_HFLD_TTL);
int encode = flags & RDB_LOAD_ENC;
unsigned char enc[4];
long long val;
@@ -295,11 +296,17 @@ void *rdbLoadIntegerObject(rio *rdb, int enctype, int flags, size_t *lenptr) {
rdbReportCorruptRDB("Unknown RDB integer encoding type %d",enctype);
return NULL; /* Never reached. */
}
if (plain || sds) {
if (plainFlag || sdsFlag || hfldFlag) {
char buf[LONG_STR_SIZE], *p;
int len = ll2string(buf,sizeof(buf),val);
if (lenptr) *lenptr = len;
p = plain ? zmalloc(len) : sdsnewlen(SDS_NOINIT,len);
if (plainFlag) {
p = zmalloc(len);
} else if (sdsFlag) {
p = sdsnewlen(SDS_NOINIT,len);
} else { /* hfldFlag */
p = hfieldNew(NULL, len, (flags&RDB_LOAD_HFLD) ? 0 : 1);
}
memcpy(p,buf,len);
return p;
} else if (encode) {
@@ -368,8 +375,11 @@ ssize_t rdbSaveLzfStringObject(rio *rdb, unsigned char *s, size_t len) {
* changes according to 'flags'. For more info check the
* rdbGenericLoadStringObject() function. */
void *rdbLoadLzfStringObject(rio *rdb, int flags, size_t *lenptr) {
int plain = flags & RDB_LOAD_PLAIN;
int sds = flags & RDB_LOAD_SDS;
int plainFlag = flags & RDB_LOAD_PLAIN;
int sdsFlag = flags & RDB_LOAD_SDS;
int hfldFlag = flags & (RDB_LOAD_HFLD | RDB_LOAD_HFLD_TTL);
int robjFlag = (!(plainFlag || sdsFlag || hfldFlag)); /* not plain/sds/hfld */
uint64_t len, clen;
unsigned char *c = NULL;
char *val = NULL;
@@ -382,11 +392,14 @@ void *rdbLoadLzfStringObject(rio *rdb, int flags, size_t *lenptr) {
}
/* Allocate our target according to the uncompressed size. */
if (plain) {
if (plainFlag) {
val = ztrymalloc(len);
} else {
} else if (sdsFlag || robjFlag) {
val = sdstrynewlen(SDS_NOINIT,len);
} else { /* hfldFlag */
val = hfieldTryNew(NULL, len, (flags&RDB_LOAD_HFLD) ? 0 : 1);
}
if (!val) {
serverLog(isRestoreContext()? LL_VERBOSE: LL_WARNING, "rdbLoadLzfStringObject failed allocating %llu bytes", (unsigned long long)len);
goto err;
@@ -402,17 +415,17 @@ void *rdbLoadLzfStringObject(rio *rdb, int flags, size_t *lenptr) {
}
zfree(c);
if (plain || sds) {
return val;
} else {
return createObject(OBJ_STRING,val);
}
return (robjFlag) ? createObject(OBJ_STRING,val) : (void *) val;
err:
zfree(c);
if (plain)
if (plainFlag) {
zfree(val);
else
} else if (sdsFlag || robjFlag) {
sdsfree(val);
} else { /* hfldFlag*/
hfieldFree(val);
}
return NULL;
}
@@ -491,12 +504,18 @@ ssize_t rdbSaveStringObject(rio *rdb, robj *obj) {
* RDB_LOAD_PLAIN: Return a plain string allocated with zmalloc()
* instead of a Redis object with an sds in it.
* RDB_LOAD_SDS: Return an SDS string instead of a Redis object.
* RDB_LOAD_HFLD: Return a hash field object (mstr)
* RDB_LOAD_HFLD_TTL: Return a hash field with TTL metadata reserved
*
* On I/O error NULL is returned.
*/
void *rdbGenericLoadStringObject(rio *rdb, int flags, size_t *lenptr) {
int plain = flags & RDB_LOAD_PLAIN;
int sds = flags & RDB_LOAD_SDS;
void *buf;
int plainFlag = flags & RDB_LOAD_PLAIN;
int sdsFlag = flags & RDB_LOAD_SDS;
int hfldFlag = flags & (RDB_LOAD_HFLD|RDB_LOAD_HFLD_TTL);
int robjFlag = (!(plainFlag || sdsFlag || hfldFlag)); /* not plain/sds/hfld */
int isencoded;
unsigned long long len;
@@ -517,22 +536,8 @@ void *rdbGenericLoadStringObject(rio *rdb, int flags, size_t *lenptr) {
}
}
if (plain || sds) {
void *buf = plain ? ztrymalloc(len) : sdstrynewlen(SDS_NOINIT,len);
if (!buf) {
serverLog(isRestoreContext()? LL_VERBOSE: LL_WARNING, "rdbGenericLoadStringObject failed allocating %llu bytes", len);
return NULL;
}
if (lenptr) *lenptr = len;
if (len && rioRead(rdb,buf,len) == 0) {
if (plain)
zfree(buf);
else
sdsfree(buf);
return NULL;
}
return buf;
} else {
/* return robj */
if (robjFlag) {
robj *o = tryCreateStringObject(SDS_NOINIT,len);
if (!o) {
serverLog(isRestoreContext()? LL_VERBOSE: LL_WARNING, "rdbGenericLoadStringObject failed allocating %llu bytes", len);
@@ -544,6 +549,32 @@ void *rdbGenericLoadStringObject(rio *rdb, int flags, size_t *lenptr) {
}
return o;
}
/* plain/sds/hfld */
if (plainFlag) {
buf = ztrymalloc(len);
} else if (sdsFlag) {
buf = sdstrynewlen(SDS_NOINIT,len);
} else { /* hfldFlag */
buf = hfieldTryNew(NULL, len, (flags&RDB_LOAD_HFLD) ? 0 : 1);
}
if (!buf) {
serverLog(isRestoreContext()? LL_VERBOSE: LL_WARNING, "rdbGenericLoadStringObject failed allocating %llu bytes", len);
return NULL;
}
if (lenptr) *lenptr = len;
if (len && rioRead(rdb,buf,len) == 0) {
if (plainFlag)
zfree(buf);
else if (sdsFlag) {
sdsfree(buf);
} else { /* hfldFlag */
hfieldFree(buf);
}
return NULL;
}
return buf;
}
robj *rdbLoadStringObject(rio *rdb) {
@@ -665,9 +696,14 @@ int rdbSaveObjectType(rio *rdb, robj *o) {
case OBJ_HASH:
if (o->encoding == OBJ_ENCODING_LISTPACK)
return rdbSaveType(rdb,RDB_TYPE_HASH_LISTPACK);
else if (o->encoding == OBJ_ENCODING_HT)
return rdbSaveType(rdb,RDB_TYPE_HASH);
else
else if (o->encoding == OBJ_ENCODING_LISTPACK_EX)
return rdbSaveType(rdb,RDB_TYPE_HASH_LISTPACK_EX);
else if (o->encoding == OBJ_ENCODING_HT) {
if (hashTypeGetMinExpire(o) == EB_EXPIRE_TIME_INVALID)
return rdbSaveType(rdb,RDB_TYPE_HASH);
else
return rdbSaveType(rdb,RDB_TYPE_HASH_METADATA);
} else
serverPanic("Unknown hash encoding");
case OBJ_STREAM:
return rdbSaveType(rdb,RDB_TYPE_STREAM_LISTPACKS_3);
@@ -908,32 +944,58 @@ ssize_t rdbSaveObject(rio *rdb, robj *o, robj *key, int dbid) {
}
} else if (o->type == OBJ_HASH) {
/* Save a hash value */
if (o->encoding == OBJ_ENCODING_LISTPACK) {
size_t l = lpBytes((unsigned char*)o->ptr);
if ((o->encoding == OBJ_ENCODING_LISTPACK) ||
(o->encoding == OBJ_ENCODING_LISTPACK_EX))
{
unsigned char *lp_ptr = hashTypeListpackGetLp(o);
size_t l = lpBytes(lp_ptr);
if ((n = rdbSaveRawString(rdb,o->ptr,l)) == -1) return -1;
if ((n = rdbSaveRawString(rdb,lp_ptr,l)) == -1) return -1;
nwritten += n;
} else if (o->encoding == OBJ_ENCODING_HT) {
dictIterator *di = dictGetIterator(o->ptr);
dictEntry *de;
/* Determine the hash layout to use based on the presence of at least
* one field with a valid TTL. If such a field exists, employ the
* RDB_TYPE_HASH_METADATA layout, including tuples of [ttl][field][value].
* Otherwise, use the standard RDB_TYPE_HASH layout containing only
* the tuples [field][value]. */
int with_ttl = (hashTypeGetMinExpire(o) != EB_EXPIRE_TIME_INVALID);
/* save number of fields in hash */
if ((n = rdbSaveLen(rdb,dictSize((dict*)o->ptr))) == -1) {
dictReleaseIterator(di);
return -1;
}
nwritten += n;
/* save all hash fields */
while((de = dictNext(di)) != NULL) {
sds field = dictGetKey(de);
hfield field = dictGetKey(de);
sds value = dictGetVal(de);
/* save the TTL */
if (with_ttl) {
uint64_t ttl = hfieldGetExpireTime(field);
/* 0 is used to indicate no TTL is set for this field */
if (ttl == EB_EXPIRE_TIME_INVALID) ttl = 0;
if ((n = rdbSaveLen(rdb, ttl)) == -1) {
dictReleaseIterator(di);
return -1;
}
nwritten += n;
}
/* save the key */
if ((n = rdbSaveRawString(rdb,(unsigned char*)field,
sdslen(field))) == -1)
hfieldlen(field))) == -1)
{
dictReleaseIterator(di);
return -1;
}
nwritten += n;
/* save the value */
if ((n = rdbSaveRawString(rdb,(unsigned char*)value,
sdslen(value))) == -1)
{
@@ -1403,7 +1465,7 @@ werr:
return C_ERR;
}
/* This helper function is only used for diskless replication.
/* This helper function is only used for diskless replication.
* This is just a wrapper to rdbSaveRio() that additionally adds a prefix
* and a suffix to the generated RDB dump. The prefix is:
*
@@ -1753,19 +1815,20 @@ static int _listZiplistEntryConvertAndValidate(unsigned char *p, unsigned int he
/* callback for to check the listpack doesn't have duplicate records */
static int _lpEntryValidation(unsigned char *p, unsigned int head_count, void *userdata) {
struct {
int pairs;
int tuple_len;
long count;
dict *fields;
long long last_expireat;
} *data = userdata;
if (data->fields == NULL) {
data->fields = dictCreate(&hashDictType);
dictExpand(data->fields, data->pairs ? head_count/2 : head_count);
dictExpand(data->fields, head_count/data->tuple_len);
}
/* If we're checking pairs, then even records are field names. Otherwise
* we're checking all elements. Add to dict and check that's not a dup */
if (!data->pairs || ((data->count) & 1) == 0) {
if (data->count % data->tuple_len == 0) {
unsigned char *str;
int64_t slen;
unsigned char buf[LP_INTBUF_SIZE];
@@ -1779,6 +1842,19 @@ static int _lpEntryValidation(unsigned char *p, unsigned int head_count, void *u
}
}
/* Validate TTL field, only for listpackex. */
if (data->count % data->tuple_len == 2) {
long long expire_at;
/* Must be an integer. */
if (!lpGetIntegerValue(p, &expire_at)) return 0;
/* Must be less than EB_EXPIRE_TIME_MAX. */
if (expire_at < 0 || (unsigned long long)expire_at > EB_EXPIRE_TIME_MAX) return 0;
/* TTL fields are ordered. If the current field has TTL, the previous field must
* also have one, and the current TTL must be greater than the previous one. */
if (expire_at != 0 && (data->last_expireat == 0 || expire_at < data->last_expireat)) return 0;
data->last_expireat = expire_at;
}
(data->count)++;
return 1;
}
@@ -1786,23 +1862,25 @@ static int _lpEntryValidation(unsigned char *p, unsigned int head_count, void *u
/* Validate the integrity of the listpack structure.
* when `deep` is 0, only the integrity of the header is validated.
* when `deep` is 1, we scan all the entries one by one.
* when `pairs` is 0, all elements need to be unique (it's a set)
* when `pairs` is 1, odd elements need to be unique (it's a key-value map) */
int lpValidateIntegrityAndDups(unsigned char *lp, size_t size, int deep, int pairs) {
* tuple_len indicates what is a logical entry tuple size.
* Whether tuple is of size 1 (set), 2 (feild-value) or 3 (field-value[-ttl]),
* first element in the tuple must be unique */
int lpValidateIntegrityAndDups(unsigned char *lp, size_t size, int deep, int tuple_len) {
if (!deep)
return lpValidateIntegrity(lp, size, 0, NULL, NULL);
/* Keep track of the field names to locate duplicate ones */
struct {
int pairs;
int tuple_len;
long count;
dict *fields; /* Initialisation at the first callback. */
} data = {pairs, 0, NULL};
long long last_expireat; /* Last field's expiry time to ensure order in TTL fields. */
} data = {tuple_len, 0, NULL, -1};
int ret = lpValidateIntegrity(lp, size, 1, _lpEntryValidation, &data);
/* make sure we have an even number of records. */
if (pairs && data.count & 1)
/* the number of records should be a multiple of the tuple length */
if (data.count % tuple_len != 0)
ret = 0;
if (data.fields) dictRelease(data.fields);
@@ -1811,9 +1889,18 @@ int lpValidateIntegrityAndDups(unsigned char *lp, size_t size, int deep, int pai
/* Load a Redis object of the specified type from the specified file.
* On success a newly allocated object is returned, otherwise NULL.
* When the function returns NULL and if 'error' is not NULL, the
* integer pointed by 'error' is set to the type of error that occurred */
robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
*
* error - When the function returns NULL and if 'error' is not NULL, the
* integer pointed by 'error' is set to the type of error that occurred
* minExpiredField - If loading a hash with expiration on fields, then this value
* will be set to the minimum expire time found in the hash fields. If there are
* no fields with expiration or it is not a hash, then it will set be to
* EB_EXPIRE_TIME_INVALID.
*/
robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, redisDb* db, int *error,
uint64_t *minExpiredField)
{
uint64_t minExpField = EB_EXPIRE_TIME_INVALID;
robj *o = NULL, *ele, *dec;
uint64_t len;
unsigned int i;
@@ -1856,7 +1943,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
decrRefCount(ele);
}
listTypeTryConversion(o,LIST_CONV_AUTO,NULL,NULL);
listTypeTryConversion(o, LIST_CONV_AUTO, NULL, NULL);
} else if (rdbtype == RDB_TYPE_SET) {
/* Read Set value */
if ((len = rdbLoadLen(rdb,NULL)) == RDB_LENERR) return NULL;
@@ -1869,7 +1956,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
o = createSetObject();
/* It's faster to expand the dict to the right size asap in order
* to avoid rehashing */
if (len > DICT_HT_INITIAL_SIZE && dictTryExpand(o->ptr,len) != DICT_OK) {
if (len > DICT_HT_INITIAL_SIZE && dictTryExpand(o->ptr, len) != DICT_OK) {
rdbReportCorruptRDB("OOM in dictTryExpand %llu", (unsigned long long)len);
decrRefCount(o);
return NULL;
@@ -1896,7 +1983,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
/* Fetch integer value from element. */
if (isSdsRepresentableAsLongLong(sdsele,&llval) == C_OK) {
uint8_t success;
o->ptr = intsetAdd(o->ptr,llval,&success);
o->ptr = intsetAdd(o->ptr, llval, &success);
if (!success) {
rdbReportCorruptRDB("Duplicate set members detected");
decrRefCount(o);
@@ -1946,7 +2033,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
/* This will also be called when the set was just converted
* to a regular hash table encoded set. */
if (o->encoding == OBJ_ENCODING_HT) {
if (dictAdd((dict*)o->ptr,sdsele,NULL) != DICT_OK) {
if (dictAdd((dict*)o->ptr, sdsele, NULL) != DICT_OK) {
rdbReportCorruptRDB("Duplicate set members detected");
decrRefCount(o);
sdsfree(sdsele);
@@ -2024,12 +2111,13 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
maxelelen <= server.zset_max_listpack_value &&
lpSafeToAdd(NULL, totelelen))
{
zsetConvert(o,OBJ_ENCODING_LISTPACK);
zsetConvert(o, OBJ_ENCODING_LISTPACK);
}
} else if (rdbtype == RDB_TYPE_HASH) {
uint64_t len;
int ret;
sds field, value;
sds value;
hfield field;
dict *dupSearchDict = NULL;
len = rdbLoadLen(rdb, NULL);
@@ -2040,7 +2128,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
/* Too many entries? Use a hash table right from the start. */
if (len > server.hash_max_listpack_entries)
hashTypeConvert(o, OBJ_ENCODING_HT);
hashTypeConvert(o, OBJ_ENCODING_HT, NULL);
else if (deep_integrity_validation) {
/* In this mode, we need to guarantee that the server won't crash
* later when the ziplist is converted to a dict.
@@ -2049,48 +2137,50 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
dupSearchDict = dictCreate(&hashDictType);
}
/* Load every field and value into the ziplist */
/* Load every field and value into the listpack */
while (o->encoding == OBJ_ENCODING_LISTPACK && len > 0) {
len--;
/* Load raw strings */
if ((field = rdbGenericLoadStringObject(rdb,RDB_LOAD_SDS,NULL)) == NULL) {
if ((field = rdbGenericLoadStringObject(rdb,RDB_LOAD_HFLD,NULL)) == NULL) {
decrRefCount(o);
if (dupSearchDict) dictRelease(dupSearchDict);
return NULL;
}
if ((value = rdbGenericLoadStringObject(rdb,RDB_LOAD_SDS,NULL)) == NULL) {
sdsfree(field);
hfieldFree(field);
decrRefCount(o);
if (dupSearchDict) dictRelease(dupSearchDict);
return NULL;
}
if (dupSearchDict) {
sds field_dup = sdsdup(field);
sds field_dup = sdsnewlen(field, hfieldlen(field));
if (dictAdd(dupSearchDict, field_dup, NULL) != DICT_OK) {
rdbReportCorruptRDB("Hash with dup elements");
dictRelease(dupSearchDict);
decrRefCount(o);
sdsfree(field_dup);
sdsfree(field);
hfieldFree(field);
sdsfree(value);
return NULL;
}
}
/* Convert to hash table if size threshold is exceeded */
if (sdslen(field) > server.hash_max_listpack_value ||
if (hfieldlen(field) > server.hash_max_listpack_value ||
sdslen(value) > server.hash_max_listpack_value ||
!lpSafeToAdd(o->ptr, sdslen(field)+sdslen(value)))
!lpSafeToAdd(o->ptr, hfieldlen(field) + sdslen(value)))
{
hashTypeConvert(o, OBJ_ENCODING_HT);
hashTypeConvert(o, OBJ_ENCODING_HT, NULL);
dictUseStoredKeyApi((dict *)o->ptr, 1);
ret = dictAdd((dict*)o->ptr, field, value);
dictUseStoredKeyApi((dict *)o->ptr, 0);
if (ret == DICT_ERR) {
rdbReportCorruptRDB("Duplicate hash fields detected");
if (dupSearchDict) dictRelease(dupSearchDict);
sdsfree(value);
sdsfree(field);
hfieldFree(field);
decrRefCount(o);
return NULL;
}
@@ -2098,10 +2188,10 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
}
/* Add pair to listpack */
o->ptr = lpAppend(o->ptr, (unsigned char*)field, sdslen(field));
o->ptr = lpAppend(o->ptr, (unsigned char*)field, hfieldlen(field));
o->ptr = lpAppend(o->ptr, (unsigned char*)value, sdslen(value));
sdsfree(field);
hfieldFree(field);
sdsfree(value);
}
@@ -2113,7 +2203,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
}
if (o->encoding == OBJ_ENCODING_HT && len > DICT_HT_INITIAL_SIZE) {
if (dictTryExpand(o->ptr,len) != DICT_OK) {
if (dictTryExpand(o->ptr, len) != DICT_OK) {
rdbReportCorruptRDB("OOM in dictTryExpand %llu", (unsigned long long)len);
decrRefCount(o);
return NULL;
@@ -2124,22 +2214,25 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
while (o->encoding == OBJ_ENCODING_HT && len > 0) {
len--;
/* Load encoded strings */
if ((field = rdbGenericLoadStringObject(rdb,RDB_LOAD_SDS,NULL)) == NULL) {
if ((field = rdbGenericLoadStringObject(rdb,RDB_LOAD_HFLD,NULL)) == NULL) {
decrRefCount(o);
return NULL;
}
if ((value = rdbGenericLoadStringObject(rdb,RDB_LOAD_SDS,NULL)) == NULL) {
sdsfree(field);
hfieldFree(field);
decrRefCount(o);
return NULL;
}
/* Add pair to hash table */
ret = dictAdd((dict*)o->ptr, field, value);
dict *d = o->ptr;
dictUseStoredKeyApi(d, 1);
ret = dictAdd(d, field, value);
dictUseStoredKeyApi(d, 0);
if (ret == DICT_ERR) {
rdbReportCorruptRDB("Duplicate hash fields detected");
sdsfree(value);
sdsfree(field);
hfieldFree(field);
decrRefCount(o);
return NULL;
}
@@ -2147,6 +2240,149 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
/* All pairs should be read by now */
serverAssert(len == 0);
} else if (rdbtype == RDB_TYPE_HASH_METADATA) {
size_t fieldLen;
sds value, field;
uint64_t expireAt;
dict *dupSearchDict = NULL;
len = rdbLoadLen(rdb, NULL);
if (len == RDB_LENERR) return NULL;
if (len == 0) goto emptykey;
/* TODO: create listpackEx or HT directly*/
o = createHashObject();
/* Too many entries? Use a hash table right from the start. */
if (len > server.hash_max_listpack_entries) {
hashTypeConvert(o, OBJ_ENCODING_HT, NULL);
dictTypeAddMeta((dict**)&o->ptr, &mstrHashDictTypeWithHFE);
initDictExpireMetadata(key, o);
} else {
hashTypeConvert(o, OBJ_ENCODING_LISTPACK_EX, NULL);
if (deep_integrity_validation) {
/* In this mode, we need to guarantee that the server won't crash
* later when the listpack is converted to a dict.
* Create a set (dict with no values) for dup search.
* We can dismiss it as soon as we convert the listpack to a hash. */
dupSearchDict = dictCreate(&hashDictType);
}
}
while (len > 0) {
len--;
/* read the TTL */
if (rdbLoadLenByRef(rdb, NULL, &expireAt) == -1) {
serverLog(LL_WARNING, "failed reading hash TTL");
decrRefCount(o);
if (dupSearchDict != NULL) dictRelease(dupSearchDict);
return NULL;
}
if (expireAt > EB_EXPIRE_TIME_MAX) {
rdbReportCorruptRDB("invalid expireAt time: %llu", (unsigned long long)expireAt);
decrRefCount(o);
return NULL;
}
/* if needed create field with TTL metadata */
if (expireAt !=0)
field = rdbGenericLoadStringObject(rdb, RDB_LOAD_HFLD_TTL, &fieldLen);
else
field = rdbGenericLoadStringObject(rdb, RDB_LOAD_HFLD, &fieldLen);
if (field == NULL) {
serverLog(LL_WARNING, "failed reading hash field");
decrRefCount(o);
if (dupSearchDict != NULL) dictRelease(dupSearchDict);
return NULL;
}
/* read the value */
if ((value = rdbGenericLoadStringObject(rdb,RDB_LOAD_SDS,NULL)) == NULL) {
serverLog(LL_WARNING, "failed reading hash value");
decrRefCount(o);
if (dupSearchDict != NULL) dictRelease(dupSearchDict);
hfieldFree(field);
return NULL;
}
/* keep the nearest expiration to connect listpack object to db expiry */
if ((expireAt != 0) && (expireAt < minExpField)) minExpField = expireAt;
/* store the values read - either to listpack or dict */
if (o->encoding == OBJ_ENCODING_LISTPACK_EX) {
/* integrity - check for key duplication (if required) */
if (dupSearchDict) {
sds field_dup = sdsnewlen(field, hfieldlen(field));
if (dictAdd(dupSearchDict, field_dup, NULL) != DICT_OK) {
rdbReportCorruptRDB("Hash with dup elements");
dictRelease(dupSearchDict);
decrRefCount(o);
sdsfree(field_dup);
sdsfree(value);
hfieldFree(field);
return NULL;
}
}
/* check if the values can be saved to listpack (or should convert to dict encoding) */
if (hfieldlen(field) > server.hash_max_listpack_value ||
sdslen(value) > server.hash_max_listpack_value ||
!lpSafeToAdd(((listpackEx*)o->ptr)->lp, hfieldlen(field) + sdslen(value) + lpEntrySizeInteger(expireAt)))
{
/* convert to hash */
hashTypeConvert(o, OBJ_ENCODING_HT, NULL);
if (len > DICT_HT_INITIAL_SIZE) { /* TODO: this is NOT the original len, but this is also the case for simple hash, is this a bug? */
if (dictTryExpand(o->ptr, len) != DICT_OK) {
rdbReportCorruptRDB("OOM in dictTryExpand %llu", (unsigned long long)len);
decrRefCount(o);
if (dupSearchDict != NULL) dictRelease(dupSearchDict);
sdsfree(value);
hfieldFree(field);
return NULL;
}
}
/* don't add the values to the new hash: the next if will catch and the values will be added there */
} else {
listpackExAddNew(o, field, hfieldlen(field),
value, sdslen(value), expireAt);
hfieldFree(field);
sdsfree(value);
}
}
if (o->encoding == OBJ_ENCODING_HT) {
/* Add pair to hash table */
dict *d = o->ptr;
dictUseStoredKeyApi(d, 1);
int ret = dictAdd(d, field, value);
dictUseStoredKeyApi(d, 0);
/* Attach expiry to the hash field and register in hash private HFE DS */
if ((ret != DICT_ERR) && expireAt) {
dictExpireMetadata *m = (dictExpireMetadata *) dictMetadata(d);
ret = ebAdd(&m->hfe, &hashFieldExpireBucketsType, field, expireAt);
}
if (ret == DICT_ERR) {
rdbReportCorruptRDB("Duplicate hash fields detected");
sdsfree(value);
hfieldFree(field);
decrRefCount(o);
return NULL;
}
}
}
if (dupSearchDict != NULL) dictRelease(dupSearchDict);
/* check for empty key (if all fields were expired) */
if (hashTypeLength(o, 0) == 0) {
decrRefCount(o);
goto expiredHash;
}
} else if (rdbtype == RDB_TYPE_LIST_QUICKLIST || rdbtype == RDB_TYPE_LIST_QUICKLIST_2) {
if ((len = rdbLoadLen(rdb,NULL)) == RDB_LENERR) return NULL;
if (len == 0) goto emptykey;
@@ -2221,7 +2457,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
goto emptykey;
}
listTypeTryConversion(o,LIST_CONV_AUTO,NULL,NULL);
listTypeTryConversion(o, LIST_CONV_AUTO, NULL, NULL);
} else if (rdbtype == RDB_TYPE_HASH_ZIPMAP ||
rdbtype == RDB_TYPE_LIST_ZIPLIST ||
rdbtype == RDB_TYPE_SET_INTSET ||
@@ -2229,14 +2465,15 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
rdbtype == RDB_TYPE_ZSET_ZIPLIST ||
rdbtype == RDB_TYPE_ZSET_LISTPACK ||
rdbtype == RDB_TYPE_HASH_ZIPLIST ||
rdbtype == RDB_TYPE_HASH_LISTPACK)
rdbtype == RDB_TYPE_HASH_LISTPACK ||
rdbtype == RDB_TYPE_HASH_LISTPACK_EX)
{
size_t encoded_len;
unsigned char *encoded =
rdbGenericLoadStringObject(rdb,RDB_LOAD_PLAIN,&encoded_len);
if (encoded == NULL) return NULL;
o = createObject(OBJ_STRING,encoded); /* Obj type fixed below. */
o = createObject(OBJ_STRING, encoded); /* Obj type fixed below. */
/* Fix the object encoding, and make sure to convert the encoded
* data type into the base type if accordingly to the current
@@ -2292,14 +2529,14 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
o->type = OBJ_HASH;
o->encoding = OBJ_ENCODING_LISTPACK;
if (hashTypeLength(o) > server.hash_max_listpack_entries ||
if (hashTypeLength(o, 0) > server.hash_max_listpack_entries ||
maxlen > server.hash_max_listpack_value)
{
hashTypeConvert(o, OBJ_ENCODING_HT);
hashTypeConvert(o, OBJ_ENCODING_HT, NULL);
}
}
break;
case RDB_TYPE_LIST_ZIPLIST:
case RDB_TYPE_LIST_ZIPLIST:
{
quicklist *ql = quicklistNew(server.list_max_listpack_size,
server.list_compress_depth);
@@ -2341,11 +2578,11 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
o->type = OBJ_SET;
o->encoding = OBJ_ENCODING_INTSET;
if (intsetLen(o->ptr) > server.set_max_intset_entries)
setTypeConvert(o,OBJ_ENCODING_HT);
setTypeConvert(o, OBJ_ENCODING_HT);
break;
case RDB_TYPE_SET_LISTPACK:
if (deep_integrity_validation) server.stat_dump_payload_sanitizations++;
if (!lpValidateIntegrityAndDups(encoded, encoded_len, deep_integrity_validation, 0)) {
if (!lpValidateIntegrityAndDups(encoded, encoded_len, deep_integrity_validation, 1)) {
rdbReportCorruptRDB("Set listpack integrity check failed.");
zfree(encoded);
o->ptr = NULL;
@@ -2386,14 +2623,14 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
}
if (zsetLength(o) > server.zset_max_listpack_entries)
zsetConvert(o,OBJ_ENCODING_SKIPLIST);
zsetConvert(o, OBJ_ENCODING_SKIPLIST);
else
o->ptr = lpShrinkToFit(o->ptr);
break;
}
case RDB_TYPE_ZSET_LISTPACK:
if (deep_integrity_validation) server.stat_dump_payload_sanitizations++;
if (!lpValidateIntegrityAndDups(encoded, encoded_len, deep_integrity_validation, 1)) {
if (!lpValidateIntegrityAndDups(encoded, encoded_len, deep_integrity_validation, 2)) {
rdbReportCorruptRDB("Zset listpack integrity check failed.");
zfree(encoded);
o->ptr = NULL;
@@ -2408,7 +2645,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
}
if (zsetLength(o) > server.zset_max_listpack_entries)
zsetConvert(o,OBJ_ENCODING_SKIPLIST);
zsetConvert(o, OBJ_ENCODING_SKIPLIST);
break;
case RDB_TYPE_HASH_ZIPLIST:
{
@@ -2426,35 +2663,57 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
o->ptr = lp;
o->type = OBJ_HASH;
o->encoding = OBJ_ENCODING_LISTPACK;
if (hashTypeLength(o) == 0) {
if (hashTypeLength(o, 0) == 0) {
decrRefCount(o);
goto emptykey;
}
if (hashTypeLength(o) > server.hash_max_listpack_entries)
hashTypeConvert(o, OBJ_ENCODING_HT);
if (hashTypeLength(o, 0) > server.hash_max_listpack_entries)
hashTypeConvert(o, OBJ_ENCODING_HT, NULL);
else
o->ptr = lpShrinkToFit(o->ptr);
break;
}
case RDB_TYPE_HASH_LISTPACK:
case RDB_TYPE_HASH_LISTPACK_EX:
/* listpack-encoded hash with TTL requires its own struct
* pointed to by o->ptr */
o->type = OBJ_HASH;
if (rdbtype == RDB_TYPE_HASH_LISTPACK_EX) {
listpackEx *lpt = listpackExCreate();
lpt->lp = encoded;
lpt->key = key;
o->ptr = lpt;
o->encoding = OBJ_ENCODING_LISTPACK_EX;
} else
o->encoding = OBJ_ENCODING_LISTPACK;
/* tuple_len is the number of elements for each key:
* key + value for simple hash, key + value + tll for hash with TTL*/
int tuple_len = (rdbtype == RDB_TYPE_HASH_LISTPACK ? 2 : 3);
/* validate read data */
if (deep_integrity_validation) server.stat_dump_payload_sanitizations++;
if (!lpValidateIntegrityAndDups(encoded, encoded_len, deep_integrity_validation, 1)) {
if (!lpValidateIntegrityAndDups(encoded, encoded_len,
deep_integrity_validation, tuple_len)) {
rdbReportCorruptRDB("Hash listpack integrity check failed.");
zfree(encoded);
o->ptr = NULL;
decrRefCount(o);
return NULL;
}
o->type = OBJ_HASH;
o->encoding = OBJ_ENCODING_LISTPACK;
if (hashTypeLength(o) == 0) {
/* if listpack is empty, delete it */
if (hashTypeLength(o, 0) == 0) {
decrRefCount(o);
goto emptykey;
}
if (hashTypeLength(o) > server.hash_max_listpack_entries)
hashTypeConvert(o, OBJ_ENCODING_HT);
/* for TTL listpack, find the minimum expiry */
minExpField = hashTypeGetNextTimeToExpire(o);
/* Convert listpack to hash table without registering in global HFE DS,
* if has HFEs, since the listpack is not connected yet to the DB */
if (hashTypeLength(o, 0) > server.hash_max_listpack_entries)
hashTypeConvert(o, OBJ_ENCODING_HT, NULL /*db->hexpires*/);
break;
default:
/* totally unreachable */
@@ -2540,7 +2799,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
/* Load the last entry ID. */
s->last_id.ms = rdbLoadLen(rdb,NULL);
s->last_id.seq = rdbLoadLen(rdb,NULL);
if (rdbtype >= RDB_TYPE_STREAM_LISTPACKS_2) {
/* Load the first entry ID. */
s->first_id.ms = rdbLoadLen(rdb,NULL);
@@ -2559,9 +2818,9 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
s->max_deleted_entry_id.ms = 0;
s->max_deleted_entry_id.seq = 0;
s->entries_added = s->length;
/* Since the rax is already loaded, we can find the first entry's
* ID. */
* ID. */
streamGetEdgeID(s,1,1,&s->first_id);
}
@@ -2794,7 +3053,13 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
RedisModuleIO io;
robj keyobj;
initStaticStringObject(keyobj,key);
moduleInitIOContext(io,mt,rdb,&keyobj,dbid);
/* shouldn't happen since db is NULL only in RDB check mode, and
* in this mode the module load code returns few lines above after
* checking module name, few lines above. So this check is only
* for safety.
*/
if (db == NULL) return NULL;
moduleInitIOContext(io,mt,rdb,&keyobj,db->id);
/* Call the rdb_load method of the module providing the 10 bit
* encoding version in the lower 10 bits of the module ID. */
void *ptr = mt->rdb_load(&io,moduleid&1023);
@@ -2807,7 +3072,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
uint64_t eof = rdbLoadLen(rdb,NULL);
if (eof == RDB_LENERR) {
if (ptr) {
o = createModuleObject(mt,ptr); /* creating just in order to easily destroy */
o = createModuleObject(mt, ptr); /* creating just in order to easily destroy */
decrRefCount(o);
}
return NULL;
@@ -2816,7 +3081,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
rdbReportCorruptRDB("The RDB file contains module data for the module '%s' that is not terminated by "
"the proper module value EOF marker", moduleTypeModuleName(mt));
if (ptr) {
o = createModuleObject(mt,ptr); /* creating just in order to easily destroy */
o = createModuleObject(mt, ptr); /* creating just in order to easily destroy */
decrRefCount(o);
}
return NULL;
@@ -2828,17 +3093,23 @@ robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error) {
moduleTypeModuleName(mt));
return NULL;
}
o = createModuleObject(mt,ptr);
o = createModuleObject(mt, ptr);
} else {
rdbReportReadError("Unknown RDB encoding type %d",rdbtype);
return NULL;
}
if (minExpiredField) *minExpiredField = minExpField;
if (error) *error = 0;
return o;
emptykey:
if (error) *error = RDB_LOAD_ERR_EMPTY_KEY;
return NULL;
expiredHash:
if (error) *error = RDB_LOAD_ERR_EXPIRED_HASH;
return NULL;
}
/* Mark that we are loading in the global state and setup the fields
@@ -3008,6 +3279,7 @@ int rdbLoadRio(rio *rdb, int rdbflags, rdbSaveInfo *rsi) {
* currently it only allow to set db object and functionLibCtx to which the data
* will be loaded (in the future it might contains more such objects). */
int rdbLoadRioWithLoadingCtx(rio *rdb, int rdbflags, rdbSaveInfo *rsi, rdbLoadingCtx *rdb_loading_ctx) {
uint64_t minExpiredField = EB_EXPIRE_TIME_INVALID;
uint64_t dbid = 0;
int type, rdbver;
uint64_t db_size = 0, expires_size = 0;
@@ -3249,15 +3521,15 @@ int rdbLoadRioWithLoadingCtx(rio *rdb, int rdbflags, rdbSaveInfo *rsi, rdbLoadin
if ((key = rdbGenericLoadStringObject(rdb,RDB_LOAD_SDS,NULL)) == NULL)
goto eoferr;
/* Read value */
val = rdbLoadObject(type,rdb,key,db->id,&error);
val = rdbLoadObject(type,rdb,key,db,&error, &minExpiredField);
/* Check if the key already expired. This function is used when loading
* an RDB file from disk, either at startup, or when an RDB was
* received from the master. In the latter case, the master is
* responsible for key expiry. If we would expire keys here, the
* snapshot taken by the master may not be reflected on the slave.
* Similarly, if the base AOF is RDB format, we want to load all
* the keys they are, since the log of operations in the incr AOF
* Similarly, if the base AOF is RDB format, we want to load all
* the keys they are, since the log of operations in the incr AOF
* is assumed to work in the exact keyspace state. */
if (val == NULL) {
/* Since we used to have bug that could lead to empty keys
@@ -3268,6 +3540,9 @@ int rdbLoadRioWithLoadingCtx(rio *rdb, int rdbflags, rdbSaveInfo *rsi, rdbLoadin
if(empty_keys_skipped++ < 10)
serverLog(LL_NOTICE, "rdbLoadObject skipping empty key: %s", key);
sdsfree(key);
} else if (error == RDB_LOAD_ERR_EXPIRED_HASH) {
/* Valid flow. Continue. */
sdsfree(key);
} else {
sdsfree(key);
goto eoferr;
@@ -3312,6 +3587,11 @@ int rdbLoadRioWithLoadingCtx(rio *rdb, int rdbflags, rdbSaveInfo *rsi, rdbLoadin
}
}
/* If minExpiredField was set, then the object is hash with expiration
* on fields and need to register it in global HFE DS */
if (minExpiredField != EB_EXPIRE_TIME_INVALID)
hashTypeAddToExpires(db, key, val, minExpiredField);
/* Set the expire time if needed */
if (expiretime != -1) {
setExpire(NULL,db,&keyobj,expiretime);
src/rdb.h
+13 -8
View File
@@ -73,10 +73,12 @@
#define RDB_TYPE_STREAM_LISTPACKS_2 19
#define RDB_TYPE_SET_LISTPACK 20
#define RDB_TYPE_STREAM_LISTPACKS_3 21
#define RDB_TYPE_HASH_METADATA 22
#define RDB_TYPE_HASH_LISTPACK_EX 23
/* NOTE: WHEN ADDING NEW RDB TYPE, UPDATE rdbIsObjectType(), and rdb_type_string[] */
/* Test if a type is an object type. */
#define rdbIsObjectType(t) (((t) >= 0 && (t) <= 7) || ((t) >= 9 && (t) <= 21))
#define rdbIsObjectType(t) (((t) >= 0 && (t) <= 7) || ((t) >= 9 && (t) <= 23))
/* Special RDB opcodes (saved/loaded with rdbSaveType/rdbLoadType). */
#define RDB_OPCODE_SLOT_INFO 244 /* Individual slot info, such as slot id and size (cluster mode only). */
@@ -101,10 +103,12 @@
#define RDB_MODULE_OPCODE_STRING 5 /* String. */
/* rdbLoad...() functions flags. */
#define RDB_LOAD_NONE 0
#define RDB_LOAD_ENC (1<<0)
#define RDB_LOAD_PLAIN (1<<1)
#define RDB_LOAD_SDS (1<<2)
#define RDB_LOAD_NONE 0
#define RDB_LOAD_ENC (1<<0)
#define RDB_LOAD_PLAIN (1<<1)
#define RDB_LOAD_SDS (1<<2)
#define RDB_LOAD_HFLD (1<<3)
#define RDB_LOAD_HFLD_TTL (1<<4)
/* flags on the purpose of rdb save or load */
#define RDBFLAGS_NONE 0 /* No special RDB loading or saving. */
@@ -116,8 +120,9 @@
/* When rdbLoadObject() returns NULL, the err flag is
* set to hold the type of error that occurred */
#define RDB_LOAD_ERR_EMPTY_KEY 1 /* Error of empty key */
#define RDB_LOAD_ERR_OTHER 2 /* Any other errors */
#define RDB_LOAD_ERR_EMPTY_KEY 1 /* Error of empty key */
#define RDB_LOAD_ERR_EXPIRED_HASH 2 /* Expired hash since all its fields are expired */
#define RDB_LOAD_ERR_OTHER 3 /* Any other errors */
ssize_t rdbWriteRaw(rio *rdb, void *p, size_t len);
int rdbSaveType(rio *rdb, unsigned char type);
@@ -138,7 +143,7 @@ int rdbSaveToFile(const char *filename);
int rdbSave(int req, char *filename, rdbSaveInfo *rsi, int rdbflags);
ssize_t rdbSaveObject(rio *rdb, robj *o, robj *key, int dbid);
size_t rdbSavedObjectLen(robj *o, robj *key, int dbid);
robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, int dbid, int *error);
robj *rdbLoadObject(int rdbtype, rio *rdb, sds key, redisDb *db, int *error, uint64_t *minExpiredField);
void backgroundSaveDoneHandler(int exitcode, int bysignal);
int rdbSaveKeyValuePair(rio *rdb, robj *key, robj *val, long long expiretime,int dbid);
ssize_t rdbSaveSingleModuleAux(rio *rdb, int when, moduleType *mt);
src/redis-check-rdb.c
+4 -3
View File
@@ -80,6 +80,8 @@ char *rdb_type_string[] = {
"stream-v2",
"set-listpack",
"stream-v3",
"hash-hashtable-md",
"hash-listpack-md",
};
/* Show a few stats collected into 'rdbstate' */
@@ -173,7 +175,6 @@ void rdbCheckSetupSignals(void) {
* otherwise the already open file 'fp' is checked. */
int redis_check_rdb(char *rdbfilename, FILE *fp) {
uint64_t dbid;
int selected_dbid = -1;
int type, rdbver;
char buf[1024];
long long expiretime, now = mstime();
@@ -245,7 +246,6 @@ int redis_check_rdb(char *rdbfilename, FILE *fp) {
if ((dbid = rdbLoadLen(&rdb,NULL)) == RDB_LENERR)
goto eoferr;
rdbCheckInfo("Selecting DB ID %llu", (unsigned long long)dbid);
selected_dbid = dbid;
continue; /* Read type again. */
} else if (type == RDB_OPCODE_RESIZEDB) {
/* RESIZEDB: Hint about the size of the keys in the currently
@@ -331,7 +331,8 @@ int redis_check_rdb(char *rdbfilename, FILE *fp) {
rdbstate.keys++;
/* Read value */
rdbstate.doing = RDB_CHECK_DOING_READ_OBJECT_VALUE;
if ((val = rdbLoadObject(type,&rdb,key->ptr,selected_dbid,NULL)) == NULL) goto eoferr;
if ((val = rdbLoadObject(type,&rdb,key->ptr,NULL,NULL,NULL)) == NULL)
goto eoferr;
/* Check if the key already expired. */
if (expiretime != -1 && expiretime < now)
rdbstate.already_expired++;
src/server.c
+27 -3
View File
@@ -19,6 +19,8 @@
#include "syscheck.h"
#include "threads_mngr.h"
#include "fmtargs.h"
#include "mstr.h"
#include "ebuckets.h"
#include <time.h>
#include <signal.h>
@@ -281,6 +283,18 @@ int dictSdsKeyCompare(dict *d, const void *key1,
return memcmp(key1, key2, l1) == 0;
}
int dictSdsMstrKeyCompare(dict *d, const void *sdsLookup, const void *mstrStored)
{
int l1,l2;
UNUSED(d);
l1 = sdslen((sds)sdsLookup);
l2 = hfieldlen((hfield)mstrStored);
if (l1 != l2) return 0;
return memcmp(sdsLookup, mstrStored, l1) == 0;
}
/* A case insensitive version used for the command lookup table and other
* places where case insensitive non binary-safe comparison is needed. */
int dictSdsKeyCaseCompare(dict *d, const void *key1,
@@ -1945,6 +1959,8 @@ void createSharedObjects(void) {
shared.persist = createStringObject("PERSIST",7);
shared.set = createStringObject("SET",3);
shared.eval = createStringObject("EVAL",4);
shared.hpexpireat = createStringObject("HPEXPIREAT",10);
shared.hdel = createStringObject("HDEL",4);
/* Shared command argument */
shared.left = createStringObject("left",4);
@@ -2504,6 +2520,7 @@ void resetServerStats(void) {
server.stat_numcommands = 0;
server.stat_numconnections = 0;
server.stat_expiredkeys = 0;
server.stat_expired_hash_fields = 0;
server.stat_expired_stale_perc = 0;
server.stat_expired_time_cap_reached_count = 0;
server.stat_expire_cycle_time_used = 0;
@@ -2652,6 +2669,7 @@ void initServer(void) {
for (j = 0; j < server.dbnum; j++) {
server.db[j].keys = kvstoreCreate(&dbDictType, slot_count_bits, flags);
server.db[j].expires = kvstoreCreate(&dbExpiresDictType, slot_count_bits, flags);
server.db[j].hexpires = ebCreate();
server.db[j].expires_cursor = 0;
server.db[j].blocking_keys = dictCreate(&keylistDictType);
server.db[j].blocking_keys_unblock_on_nokey = dictCreate(&objectKeyPointerValueDictType);
@@ -5854,6 +5872,7 @@ sds genRedisInfoString(dict *section_dict, int all_sections, int everything) {
"sync_full:%lld\r\n", server.stat_sync_full,
"sync_partial_ok:%lld\r\n", server.stat_sync_partial_ok,
"sync_partial_err:%lld\r\n", server.stat_sync_partial_err,
"expired_hash_fields:%lld\r\n", server.stat_expired_hash_fields,
"expired_keys:%lld\r\n", server.stat_expiredkeys,
"expired_stale_perc:%.2f\r\n", server.stat_expired_stale_perc*100,
"expired_time_cap_reached_count:%lld\r\n", server.stat_expired_time_cap_reached_count,
@@ -6092,14 +6111,16 @@ sds genRedisInfoString(dict *section_dict, int all_sections, int everything) {
if (sections++) info = sdscat(info,"\r\n");
info = sdscatprintf(info, "# Keyspace\r\n");
for (j = 0; j < server.dbnum; j++) {
long long keys, vkeys;
long long keys, vkeys, hexpires;
keys = kvstoreSize(server.db[j].keys);
vkeys = kvstoreSize(server.db[j].expires);
hexpires = ebGetTotalItems(server.db[j].hexpires, &hashExpireBucketsType);
if (keys || vkeys) {
info = sdscatprintf(info,
"db%d:keys=%lld,expires=%lld,avg_ttl=%lld\r\n",
j, keys, vkeys, server.db[j].avg_ttl);
"db%d:keys=%lld,expires=%lld,avg_ttl=%lld,hashes_with_expiry_fields=%lld\r\n",
j, keys, vkeys, server.db[j].avg_ttl, hexpires);
}
}
}
@@ -6871,9 +6892,11 @@ struct redisTest {
{"crc64", crc64Test},
{"zmalloc", zmalloc_test},
{"sds", sdsTest},
{"mstr", mstrTest},
{"dict", dictTest},
{"listpack", listpackTest},
{"kvstore", kvstoreTest},
{"ebuckets", ebucketsTest},
};
redisTestProc *getTestProcByName(const char *name) {
int numtests = sizeof(redisTests)/sizeof(struct redisTest);
@@ -6900,6 +6923,7 @@ int main(int argc, char **argv) {
if (!strcasecmp(arg, "--accurate")) flags |= REDIS_TEST_ACCURATE;
else if (!strcasecmp(arg, "--large-memory")) flags |= REDIS_TEST_LARGE_MEMORY;
else if (!strcasecmp(arg, "--valgrind")) flags |= REDIS_TEST_VALGRIND;
else if (!strcasecmp(arg, "--verbose")) flags |= REDIS_TEST_VERBOSE;
}
if (!strcasecmp(argv[2], "all")) {
src/server.h
+103 -17
View File
@@ -45,6 +45,8 @@ typedef long long ustime_t; /* microsecond time type. */
#include "ae.h" /* Event driven programming library */
#include "sds.h" /* Dynamic safe strings */
#include "mstr.h" /* Immutable strings with optional metadata attached */
#include "ebuckets.h" /* expiry data structure */
#include "dict.h" /* Hash tables */
#include "kvstore.h" /* Slot-based hash table */
#include "adlist.h" /* Linked lists */
@@ -884,6 +886,7 @@ struct RedisModuleDigest {
#define OBJ_ENCODING_QUICKLIST 9 /* Encoded as linked list of listpacks */
#define OBJ_ENCODING_STREAM 10 /* Encoded as a radix tree of listpacks */
#define OBJ_ENCODING_LISTPACK 11 /* Encoded as a listpack */
#define OBJ_ENCODING_LISTPACK_EX 12 /* Encoded as listpack, extended with metadata */
#define LRU_BITS 24
#define LRU_CLOCK_MAX ((1<<LRU_BITS)-1) /* Max value of obj->lru */
@@ -960,6 +963,7 @@ typedef struct replBufBlock {
typedef struct redisDb {
kvstore *keys; /* The keyspace for this DB */
kvstore *expires; /* Timeout of keys with a timeout set */
ebuckets hexpires; /* Hash expiration DS. Single TTL per hash (of next min field to expire) */
dict *blocking_keys; /* Keys with clients waiting for data (BLPOP)*/
dict *blocking_keys_unblock_on_nokey; /* Keys with clients waiting for
* data, and should be unblocked if key is deleted (XREADEDGROUP).
@@ -1313,7 +1317,8 @@ struct sharedObjectsStruct {
*unsubscribebulk, *psubscribebulk, *punsubscribebulk, *del, *unlink,
*rpop, *lpop, *lpush, *rpoplpush, *lmove, *blmove, *zpopmin, *zpopmax,
*emptyscan, *multi, *exec, *left, *right, *hset, *srem, *xgroup, *xclaim,
*script, *replconf, *eval, *persist, *set, *pexpireat, *pexpire,
*script, *replconf, *eval, *persist, *set, *pexpireat, *pexpire,
*hdel, *hpexpireat,
*time, *pxat, *absttl, *retrycount, *force, *justid, *entriesread,
*lastid, *ping, *setid, *keepttl, *load, *createconsumer,
*getack, *special_asterick, *special_equals, *default_username, *redacted,
@@ -1646,6 +1651,7 @@ struct redisServer {
long long stat_numcommands; /* Number of processed commands */
long long stat_numconnections; /* Number of connections received */
long long stat_expiredkeys; /* Number of expired keys */
long long stat_expired_hash_fields; /* Number of expired hash-fields */
double stat_expired_stale_perc; /* Percentage of keys probably expired */
long long stat_expired_time_cap_reached_count; /* Early expire cycle stops.*/
long long stat_expire_cycle_time_used; /* Cumulative microseconds used. */
@@ -2433,7 +2439,8 @@ typedef struct {
robj *subject;
int encoding;
unsigned char *fptr, *vptr;
unsigned char *fptr, *vptr, *tptr;
uint64_t expire_time; /* Only used with OBJ_ENCODING_LISTPACK_EX */
dictIterator *di;
dictEntry *de;
@@ -2449,6 +2456,10 @@ typedef struct {
#define IO_THREADS_OP_WRITE 2
extern int io_threads_op;
/* Hash-field data type (of t_hash.c) */
typedef mstr hfield;
extern mstrKind mstrFieldKind;
/*-----------------------------------------------------------------------------
* Extern declarations
*----------------------------------------------------------------------------*/
@@ -2463,6 +2474,8 @@ extern dictType zsetDictType;
extern dictType dbDictType;
extern double R_Zero, R_PosInf, R_NegInf, R_Nan;
extern dictType hashDictType;
extern dictType mstrHashDictType;
extern dictType mstrHashDictTypeWithHFE;
extern dictType stringSetDictType;
extern dictType externalStringType;
extern dictType sdsHashDictType;
@@ -2474,6 +2487,9 @@ extern dictType sdsReplyDictType;
extern dictType keylistDictType;
extern dict *modules;
extern EbucketsType hashExpireBucketsType; /* global expires */
extern EbucketsType hashFieldExpireBucketsType; /* local per hash */
/*-----------------------------------------------------------------------------
* Functions prototypes
*----------------------------------------------------------------------------*/
@@ -2616,6 +2632,7 @@ void copyReplicaOutputBuffer(client *dst, client *src);
void addListRangeReply(client *c, robj *o, long start, long end, int reverse);
void deferredAfterErrorReply(client *c, list *errors);
size_t sdsZmallocSize(sds s);
size_t hfieldZmallocSize(hfield s);
size_t getStringObjectSdsUsedMemory(robj *o);
void freeClientReplyValue(void *o);
void *dupClientReplyValue(void *o);
@@ -3140,30 +3157,87 @@ void setTypeConvert(robj *subject, int enc);
int setTypeConvertAndExpand(robj *setobj, int enc, unsigned long cap, int panic);
robj *setTypeDup(robj *o);
/* Data structure for OBJ_ENCODING_LISTPACK_EX for hash. It contains listpack
* and metadata fields for hash field expiration.*/
typedef struct listpackEx {
ExpireMeta meta; /* To be used in order to register the hash in the
global ebuckets (i.e. db->hexpires) with next,
minimum, hash-field to expire. */
sds key; /* reference to the key, same one that stored in
db->dict. Will be used from active-expiration flow
for notification and deletion of the object, if
needed. */
void *lp; /* listpack that contains 'key-value-ttl' tuples which
are ordered by ttl. */
} listpackEx;
/* Each dict of hash object that has fields with time-Expiration will have the
* following metadata attached to dict header */
typedef struct dictExpireMetadata {
ExpireMeta expireMeta; /* embedded ExpireMeta in dict.
To be used in order to register the hash in the
global ebuckets (i.e db->hexpires) with next,
minimum, hash-field to expire */
ebuckets hfe; /* DS of Hash Fields Expiration, associated to each hash */
sds key; /* reference to the key, same one that stored in
db->dict. Will be used from active-expiration flow
for notification and deletion of the object, if
needed. */
} dictExpireMetadata;
/* Hash data type */
#define HASH_SET_TAKE_FIELD (1<<0)
#define HASH_SET_TAKE_VALUE (1<<1)
#define HASH_SET_COPY 0
void hashTypeConvert(robj *o, int enc);
void hashTypeTryConversion(robj *subject, robj **argv, int start, int end);
int hashTypeExists(robj *o, sds key);
int hashTypeDelete(robj *o, sds key);
unsigned long hashTypeLength(const robj *o);
void hashTypeConvert(robj *o, int enc, ebuckets *hexpires);
void hashTypeTryConversion(redisDb *db, robj *subject, robj **argv, int start, int end);
int hashTypeExists(redisDb *db, robj *o, sds key, int *isHashDeleted);
int hashTypeDelete(robj *o, void *key, int isSdsField);
unsigned long hashTypeLength(const robj *o, int subtractExpiredFields);
hashTypeIterator *hashTypeInitIterator(robj *subject);
void hashTypeReleaseIterator(hashTypeIterator *hi);
int hashTypeNext(hashTypeIterator *hi);
int hashTypeNext(hashTypeIterator *hi, int skipExpiredFields);
void hashTypeCurrentFromListpack(hashTypeIterator *hi, int what,
unsigned char **vstr,
unsigned int *vlen,
long long *vll);
sds hashTypeCurrentFromHashTable(hashTypeIterator *hi, int what);
void hashTypeCurrentObject(hashTypeIterator *hi, int what, unsigned char **vstr, unsigned int *vlen, long long *vll);
long long *vll,
uint64_t *expireTime);
void hashTypeCurrentFromHashTable(hashTypeIterator *hi, int what, char **str,
size_t *len, uint64_t *expireTime);
void hashTypeCurrentObject(hashTypeIterator *hi, int what, unsigned char **vstr,
unsigned int *vlen, long long *vll, uint64_t *expireTime);
sds hashTypeCurrentObjectNewSds(hashTypeIterator *hi, int what);
robj *hashTypeLookupWriteOrCreate(client *c, robj *key);
robj *hashTypeGetValueObject(robj *o, sds field);
int hashTypeSet(robj *o, sds field, sds value, int flags);
robj *hashTypeDup(robj *o);
hfield hashTypeCurrentObjectNewHfield(hashTypeIterator *hi);
robj *hashTypeGetValueObject(redisDb *db, robj *o, sds field, int *isHashDeleted);
int hashTypeSet(redisDb *db, robj *o, sds field, sds value, int flags);
robj *hashTypeDup(robj *o, sds newkey, uint64_t *minHashExpire);
uint64_t hashTypeRemoveFromExpires(ebuckets *hexpires, robj *o);
void hashTypeAddToExpires(redisDb *db, sds key, robj *hashObj, uint64_t expireTime);
void hashTypeFree(robj *o);
int hashTypeIsExpired(const robj *o, uint64_t expireAt);
uint64_t hashTypeGetMinExpire(robj *o);
unsigned char *hashTypeListpackGetLp(robj *o);
uint64_t hashTypeGetMinExpire(robj *o);
void hashTypeUpdateKeyRef(robj *o, sds newkey);
ebuckets *hashTypeGetDictMetaHFE(dict *d);
uint64_t hashTypeGetMinExpire(robj *keyObj);
uint64_t hashTypeGetNextTimeToExpire(robj *o);
void initDictExpireMetadata(sds key, robj *o);
struct listpackEx *listpackExCreate(void);
void listpackExAddNew(robj *o, char *field, size_t flen,
char *value, size_t vlen, uint64_t expireAt);
/* Hash-Field data type (of t_hash.c) */
hfield hfieldNew(const void *field, size_t fieldlen, int withExpireMeta);
hfield hfieldTryNew(const void *field, size_t fieldlen, int withExpireMeta);
int hfieldIsExpireAttached(hfield field);
int hfieldIsExpired(hfield field);
uint64_t hfieldGetExpireTime(hfield field);
static inline void hfieldFree(hfield field) { mstrFree(&mstrFieldKind, field); }
static inline void *hfieldGetAllocPtr(hfield field) { return mstrGetAllocPtr(&mstrFieldKind, field); }
static inline size_t hfieldlen(hfield field) { return mstrlen(field);}
uint64_t hfieldGetExpireTime(hfield field);
/* Pub / Sub */
int pubsubUnsubscribeAllChannels(client *c, int notify);
@@ -3182,7 +3256,7 @@ dict *getClientPubSubChannels(client *c);
dict *getClientPubSubShardChannels(client *c);
/* Keyspace events notification */
void notifyKeyspaceEvent(int type, char *event, robj *key, int dbid);
void notifyKeyspaceEvent(int type, const char *event, robj *key, int dbid);
int keyspaceEventsStringToFlags(char *classes);
sds keyspaceEventsFlagsToString(int flags);
@@ -3266,6 +3340,7 @@ int keyIsExpired(redisDb *db, robj *key);
long long getExpire(redisDb *db, robj *key);
void setExpire(client *c, redisDb *db, robj *key, long long when);
int checkAlreadyExpired(long long when);
int parseExtendedExpireArgumentsOrReply(client *c, int *flags);
robj *lookupKeyRead(redisDb *db, robj *key);
robj *lookupKeyWrite(redisDb *db, robj *key);
robj *lookupKeyReadOrReply(client *c, robj *key, robj *reply);
@@ -3284,7 +3359,7 @@ int objectSetLRUOrLFU(robj *val, long long lfu_freq, long long lru_idle,
#define LOOKUP_NOEXPIRE (1<<4) /* Avoid deleting lazy expired keys. */
#define LOOKUP_NOEFFECTS (LOOKUP_NONOTIFY | LOOKUP_NOSTATS | LOOKUP_NOTOUCH | LOOKUP_NOEXPIRE) /* Avoid any effects from fetching the key */
void dbAdd(redisDb *db, robj *key, robj *val);
dictEntry *dbAdd(redisDb *db, robj *key, robj *val);
int dbAddRDBLoad(redisDb *db, sds key, robj *val);
void dbReplaceValue(redisDb *db, robj *key, robj *val);
@@ -3439,6 +3514,7 @@ void expireSlaveKeys(void);
void rememberSlaveKeyWithExpire(redisDb *db, robj *key);
void flushSlaveKeysWithExpireList(void);
size_t getSlaveKeyWithExpireCount(void);
uint64_t hashTypeDbActiveExpire(redisDb *db, uint32_t maxFieldsToExpire);
/* evict.c -- maxmemory handling and LRU eviction. */
void evictionPoolAlloc(void);
@@ -3456,6 +3532,7 @@ void startEvictionTimeProc(void);
uint64_t dictSdsHash(const void *key);
uint64_t dictSdsCaseHash(const void *key);
int dictSdsKeyCompare(dict *d, const void *key1, const void *key2);
int dictSdsMstrKeyCompare(dict *d, const void *sdsLookup, const void *mstrStored);
int dictSdsKeyCaseCompare(dict *d, const void *key1, const void *key2);
void dictSdsDestructor(dict *d, void *val);
void dictListDestructor(dict *d, void *val);
@@ -3611,6 +3688,15 @@ void strlenCommand(client *c);
void zrankCommand(client *c);
void zrevrankCommand(client *c);
void hsetCommand(client *c);
void hpexpireCommand(client *c);
void hexpireCommand(client *c);
void hpexpireatCommand(client *c);
void hexpireatCommand(client *c);
void httlCommand(client *c);
void hpttlCommand(client *c);
void hexpiretimeCommand(client *c);
void hpexpiretimeCommand(client *c);
void hpersistCommand(client *c);
void hsetnxCommand(client *c);
void hgetCommand(client *c);
void hmgetCommand(client *c);
src/sort.c
+6 -1
View File
@@ -94,7 +94,12 @@ robj *lookupKeyByPattern(redisDb *db, robj *pattern, robj *subst) {
/* Retrieve value from hash by the field name. The returned object
* is a new object with refcount already incremented. */
o = hashTypeGetValueObject(o, fieldobj->ptr);
int isHashDeleted;
o = hashTypeGetValueObject(db, o, fieldobj->ptr, &isHashDeleted);
if (isHashDeleted)
goto noobj;
} else {
if (o->type != OBJ_STRING) goto noobj;
src/t_hash.c
+2289 -201
View File
File diff suppressed because it is too large Load Diff
src/t_set.c
+4 -4
View File
@@ -432,7 +432,7 @@ robj *setTypePopRandom(robj *set) {
if (set->encoding == OBJ_ENCODING_LISTPACK) {
/* Find random and delete it without re-seeking the listpack. */
unsigned int i = 0;
unsigned char *p = lpNextRandom(set->ptr, lpFirst(set->ptr), &i, 1, 0);
unsigned char *p = lpNextRandom(set->ptr, lpFirst(set->ptr), &i, 1, 1);
unsigned int len = 0; /* initialize to silence warning */
long long llele = 0; /* initialize to silence warning */
char *str = (char *)lpGetValue(p, &len, &llele);
@@ -815,7 +815,7 @@ void spopWithCountCommand(client *c) {
unsigned int index = 0;
unsigned char **ps = zmalloc(sizeof(char *) * count);
for (unsigned long i = 0; i < count; i++) {
p = lpNextRandom(lp, p, &index, count - i, 0);
p = lpNextRandom(lp, p, &index, count - i, 1);
unsigned int len;
str = (char *)lpGetValue(p, &len, (long long *)&llele);
@@ -877,7 +877,7 @@ void spopWithCountCommand(client *c) {
unsigned int index = 0;
unsigned char **ps = zmalloc(sizeof(char *) * remaining);
for (unsigned long i = 0; i < remaining; i++) {
p = lpNextRandom(lp, p, &index, remaining - i, 0);
p = lpNextRandom(lp, p, &index, remaining - i, 1);
unsigned int len;
str = (char *)lpGetValue(p, &len, (long long *)&llele);
setTypeAddAux(newset, str, len, llele, 0);
@@ -1103,7 +1103,7 @@ void srandmemberWithCountCommand(client *c) {
unsigned int i = 0;
addReplyArrayLen(c, count);
while (count) {
p = lpNextRandom(lp, p, &i, count--, 0);
p = lpNextRandom(lp, p, &i, count--, 1);
unsigned int len;
str = (char *)lpGetValue(p, &len, (long long *)&llele);
if (str == NULL) {
src/t_zset.c
+3 -3
View File
@@ -1754,7 +1754,7 @@ void zsetTypeRandomElement(robj *zsetobj, unsigned long zsetsize, listpackEntry
*score = *(double*)dictGetVal(de);
} else if (zsetobj->encoding == OBJ_ENCODING_LISTPACK) {
listpackEntry val;
lpRandomPair(zsetobj->ptr, zsetsize, key, &val);
lpRandomPair(zsetobj->ptr, zsetsize, key, &val, 2);
if (score) {
if (val.sval) {
*score = zzlStrtod(val.sval,val.slen);
@@ -4263,7 +4263,7 @@ void zrandmemberWithCountCommand(client *c, long l, int withscores) {
while (count) {
sample_count = count > limit ? limit : count;
count -= sample_count;
lpRandomPairs(zsetobj->ptr, sample_count, keys, vals);
lpRandomPairs(zsetobj->ptr, sample_count, keys, vals, 2);
zrandmemberReplyWithListpack(c, sample_count, keys, vals);
if (c->flags & CLIENT_CLOSE_ASAP)
break;
@@ -4317,7 +4317,7 @@ void zrandmemberWithCountCommand(client *c, long l, int withscores) {
keys = zmalloc(sizeof(listpackEntry)*count);
if (withscores)
vals = zmalloc(sizeof(listpackEntry)*count);
serverAssert(lpRandomPairsUnique(zsetobj->ptr, count, keys, vals) == count);
serverAssert(lpRandomPairsUnique(zsetobj->ptr, count, keys, vals, 2) == count);
zrandmemberReplyWithListpack(c, count, keys, vals);
zfree(keys);
zfree(vals);
src/testhelp.h
+2
View File
@@ -21,6 +21,8 @@
#define REDIS_TEST_ACCURATE (1<<0)
#define REDIS_TEST_LARGE_MEMORY (1<<1)
#define REDIS_TEST_VALGRIND (1<<2)
#define REDIS_TEST_VERBOSE (1<<3)
extern int __failed_tests;
extern int __test_num;
tests/integration/corrupt-dump-fuzzer.tcl
+3 -1
View File
@@ -19,12 +19,14 @@ proc generate_collections {suffix elements} {
# add both string values and integers
if {$j % 2 == 0} {set val $j} else {set val "_$j"}
$rd hset hash$suffix $j $val
$rd hset hashmd$suffix $j $val
$rd hexpire hashmd$suffix [expr {int(rand() * 10000)}] FIELDS 1 $j
$rd lpush list$suffix $val
$rd zadd zset$suffix $j $val
$rd sadd set$suffix $val
$rd xadd stream$suffix * item 1 value $val
}
for {set j 0} {$j < $elements * 5} {incr j} {
for {set j 0} {$j < $elements * 7} {incr j} {
$rd read ; # Discard replies
}
$rd close
tests/integration/corrupt-dump.tcl
+56
View File
@@ -59,6 +59,62 @@ test {corrupt payload: valid zipped hash header, dup records} {
}
}
test {corrupt payload: hash listpackex with invalid string TTL} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
catch {
r restore key 0 "\x17\x2d\x2d\x00\x00\x00\x09\x00\x81\x61\x02\x01\x01\xf4\xa6\x96\x18\xb8\x8f\x01\x00\x00\x09\x82\x66\x31\x03\x82\x76\x31\x03\x83\x66\x6f\x6f\x04\x82\x66\x32\x03\x82\x76\x32\x03\x00\x01\xff\x0c\x00\xde\x40\xe5\x37\x51\x1c\x12\x56" replace
} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: hash listpackex with TTL large than EB_EXPIRE_TIME_MAX} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
catch {
r restore key 0 "\x17\x33\x33\x00\x00\x00\x09\x00\x00\x01\x00\x01\xf4\x01\xc5\x89\x95\x8f\x01\x00\x00\x09\x01\x01\x82\x5f\x31\x03\xf4\x29\x94\x97\x95\x8f\x01\x00\x00\x09\x02\x01\x02\x01\xf4\x01\x5e\xaf\x95\x8f\x01\x33\x00\x09\xff\x0c\x00\x7e\x4f\xf4\x33\xe9\xc5\x3e\x56" replace
} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: hash listpackex with unordered TTL fields} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
catch {
r restore key 0 "\x17\xc3\x30\x35\x14\x35\x00\x00\x00\t\x00\x82\x66\x32\x03\x82\x76\x32\x03\xf4\x80\x73\x16\xd1\x8f\x01\x20\x12\x02\x82\x66\x31\x20\x11\x03\x31\x03\xf4\x7f\xe0\x01\x11\x00\x33\x20\x11\x04\x33\x03\x00\x01\xff\x0c\x00\xf6\x70\x29\x57\x11\x68\x9d\xe5" replace
} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: hash listpackex field without TTL should not be followed by field with TTL} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload yes
catch {
r restore key 0 "\x17\x2d\x2d\x00\x00\x00\x09\x00\x82\x66\x31\x03\x82\x76\x31\x03\x00\x01\x82\x66\x32\x03\x82\x76\x32\x03\xf4\xe0\x59\x7a\x96\x00\x00\x00\x00\x09\x82\x66\x33\x03\x82\x76\x33\x03\x00\x01\xff\x0c\x00\x42\x66\xd4\xbe\x17\xc3\x96\x72" replace
} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: hash hashtable with TTL large than EB_EXPIRE_TIME_MAX} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set hash-max-listpack-entries 0
r config set sanitize-dump-payload yes
catch {
r restore key 0 "\x16\x02\x81\x00\x01\x00\x00\x00\x00\x00\x00\x02\x66\x31\x02\x76\x31\x81\x00\x01\x00\x00\x00\x00\x00\x00\x02\x66\x32\x02\x76\x32\x0c\x00\xb9\x3c\x65\x28\x40\x94\x58\x36" replace
} err
assert_match "*Bad data format*" $err
r ping
}
}
test {corrupt payload: quicklist big ziplist prev len} {
start_server [list overrides [list loglevel verbose use-exit-on-panic yes crash-memcheck-enabled no] ] {
r config set sanitize-dump-payload no
tests/integration/psync2-master-restart.tcl
+11
View File
@@ -179,6 +179,17 @@ start_server {} {
$master set $j somevalue px 10
}
##### hash-field-expiration
# Hashes of type OBJ_ENCODING_LISTPACK_EX won't be discarded during
# RDB load, even if they are expired.
$master hset myhash1 f1 v1 f2 v2 f3 v3
$master hpexpire myhash1 10 FIELDS 3 f1 f2 f3
# Hashes of type RDB_TYPE_HASH_METADATA will be discarded during RDB load.
$master config set hash-max-listpack-entries 0
$master hset myhash2 f1 v1 f2 v2
$master hpexpire myhash2 10 FIELDS 2 f1 f2
$master config set hash-max-listpack-entries 1
after 20
wait_for_condition 500 100 {
tests/integration/rdb.tcl
+224
View File
@@ -416,4 +416,228 @@ start_server {} {
} {OK}
}
set server_path [tmpdir "server.partial-hfield-exp-test"]
# verifies writing and reading hash key with expiring and persistent fields
start_server [list overrides [list "dir" $server_path]] {
foreach {type lp_entries} {listpack 512 dict 0} {
test "HFE - save and load expired fields, expired soon after, or long after ($type)" {
r config set hash-max-listpack-entries $lp_entries
r FLUSHALL
r HMSET key a 1 b 2 c 3 d 4 e 5
# expected to be expired long after restart
r HEXPIREAT key 2524600800 FIELDS 1 a
# expected long TTL value (6 bytes) is saved and loaded correctly
r HPEXPIREAT key 188900976391764 FIELDS 1 b
# expected to be already expired after restart
r HPEXPIRE key 80 FIELDS 1 d
# expected to be expired soon after restart
r HPEXPIRE key 200 FIELDS 1 e
r save
# sleep 101 ms to make sure d will expire after restart
after 101
restart_server 0 true false
wait_done_loading r
assert_equal [lsort [r hgetall key]] "1 2 3 a b c"
assert_equal [r hpexpiretime key FIELDS 3 a b c] {2524600800000 188900976391764 -1}
assert_equal [s rdb_last_load_keys_loaded] 1
# wait until expired_hash_fields equals 2
wait_for_condition 10 100 {
[s expired_hash_fields] == 2
} else {
fail "Value of expired_hash_fields is not as expected"
}
}
}
}
set server_path [tmpdir "server.all-hfield-exp-test"]
# verifies writing hash with several expired keys, and active-expiring it on load
start_server [list overrides [list "dir" $server_path]] {
foreach {type lp_entries} {listpack 512 dict 0} {
test "HFE - save and load rdb all fields expired, ($type)" {
r config set hash-max-listpack-entries $lp_entries
r FLUSHALL
r HMSET key a 1 b 2 c 3 d 4
r HPEXPIRE key 100 FIELDS 4 a b c d
r save
# sleep 101 ms to make sure all fields will expire after restart
after 101
restart_server 0 true false
wait_done_loading r
# it is expected that no field was expired on load and the key was
# loaded, even though all its fields are actually expired.
assert_equal [s rdb_last_load_keys_loaded] 1
assert_equal [r hgetall key] {}
}
}
}
set server_path [tmpdir "server.listpack-to-dict-test"]
test "save listpack, load dict" {
start_server [list overrides [list "dir" $server_path enable-debug-command yes]] {
r config set hash-max-listpack-entries 512
r FLUSHALL
r HMSET key a 1 b 2 c 3 d 4
assert_match "*encoding:listpack*" [r debug object key]
r HPEXPIRE key 100 FIELDS 1 d
r save
# sleep 200 ms to make sure 'd' will expire after when reloading
after 200
# change configuration and reload - result should be dict-encoded key
r config set hash-max-listpack-entries 0
r debug reload nosave
# first verify d was not expired during load (no expiry when loading
# a hash that was saved listpack-encoded)
assert_equal [s rdb_last_load_keys_loaded] 1
# d should be lazy expired in hgetall
assert_equal [lsort [r hgetall key]] "1 2 3 a b c"
assert_match "*encoding:hashtable*" [r debug object key]
}
}
set server_path [tmpdir "server.dict-to-listpack-test"]
test "save dict, load listpack" {
start_server [list overrides [list "dir" $server_path enable-debug-command yes]] {
r config set hash-max-listpack-entries 0
r FLUSHALL
r HMSET key a 1 b 2 c 3 d 4
assert_match "*encoding:hashtable*" [r debug object key]
r HPEXPIRE key 200 FIELDS 1 d
r save
# sleep 201 ms to make sure 'd' will expire during reload
after 201
# change configuration and reload - result should be LP-encoded key
r config set hash-max-listpack-entries 512
r debug reload nosave
# verify d was expired during load
assert_equal [s rdb_last_load_keys_loaded] 1
assert_equal [lsort [r hgetall key]] "1 2 3 a b c"
assert_match "*encoding:listpack*" [r debug object key]
}
}
set server_path [tmpdir "server.active-expiry-after-load"]
# verifies a field is correctly expired by active expiry AFTER loading from RDB
foreach {type lp_entries} {listpack 512 dict 0} {
start_server [list overrides [list "dir" $server_path enable-debug-command yes]] {
test "active field expiry after load, ($type)" {
r config set hash-max-listpack-entries $lp_entries
r FLUSHALL
r HMSET key a 1 b 2 c 3 d 4 e 5 f 6
r HEXPIREAT key 2524600800 FIELDS 2 a b
r HPEXPIRE key 200 FIELDS 2 c d
r save
r debug reload nosave
# wait at most 2 secs to make sure 'c' and 'd' will active-expire
wait_for_condition 20 100 {
[s expired_hash_fields] == 2
} else {
fail "expired hash fields is [s expired_hash_fields] != 2"
}
assert_equal [s rdb_last_load_keys_loaded] 1
# hgetall might lazy expire fields, so it's only called after the stat asserts
assert_equal [lsort [r hgetall key]] "1 2 5 6 a b e f"
assert_equal [r hexpiretime key FIELDS 6 a b c d e f] {2524600800 2524600800 -2 -2 -1 -1}
}
}
}
set server_path [tmpdir "server.lazy-expiry-after-load"]
foreach {type lp_entries} {listpack 512 dict 0} {
start_server [list overrides [list "dir" $server_path enable-debug-command yes]] {
test "lazy field expiry after load, ($type)" {
r config set hash-max-listpack-entries $lp_entries
r debug set-active-expire 0
r FLUSHALL
r HMSET key a 1 b 2 c 3 d 4 e 5 f 6
r HEXPIREAT key 2524600800 FIELDS 2 a b
r HPEXPIRE key 200 FIELDS 2 c d
r save
r debug reload nosave
# sleep 500 msec to make sure 'c' and 'd' will lazy-expire when calling hgetall
after 500
assert_equal [s rdb_last_load_keys_loaded] 1
assert_equal [s expired_hash_fields] 0
# hgetall will lazy expire fields, so it's only called after the stat asserts
assert_equal [lsort [r hgetall key]] "1 2 5 6 a b e f"
assert_equal [r hexpiretime key FIELDS 6 a b c d e f] {2524600800 2524600800 -2 -2 -1 -1}
}
}
}
set server_path [tmpdir "server.unexpired-items-rax-list-boundary"]
foreach {type lp_entries} {listpack 512 dict 0} {
start_server [list overrides [list "dir" $server_path enable-debug-command yes]] {
test "load un-expired items below and above rax-list boundary, ($type)" {
r config set hash-max-listpack-entries $lp_entries
r flushall
set hash_sizes {15 16 17 31 32 33}
foreach h $hash_sizes {
for {set i 1} {$i <= $h} {incr i} {
r hset key$h f$i v$i
r hexpireat key$h 2524600800 FIELDS 1 f$i
}
}
r save
restart_server 0 true false
wait_done_loading r
set hash_sizes {15 16 17 31 32 33}
foreach h $hash_sizes {
for {set i 1} {$i <= $h} {incr i} {
# random expiration time
assert_equal [r hget key$h f$i] v$i
assert_equal [r hexpiretime key$h FIELDS 1 f$i] 2524600800
}
}
}
}
}
} ;# tags
tests/support/util.tcl
+15 -2
View File
@@ -293,6 +293,9 @@ proc findKeyWithType {r type} {
proc createComplexDataset {r ops {opt {}}} {
set useexpire [expr {[lsearch -exact $opt useexpire] != -1}]
# TODO: Remove usehexpire on next commit, when RDB will support replication
set usehexpire [expr {[lsearch -exact $opt usehexpire] != -1}]
if {[lsearch -exact $opt usetag] != -1} {
set tag "{t}"
} else {
@@ -386,6 +389,10 @@ proc createComplexDataset {r ops {opt {}}} {
{hash} {
randpath {{*}$r hset $k $f $v} \
{{*}$r hdel $k $f}
if { [{*}$r hexists $k $f] && $usehexpire && rand() < 0.5} {
{*}$r hexpire $k 1000 FIELDS 1 $f
}
}
}
}
@@ -438,8 +445,14 @@ proc csvdump r {
hash {
set fields [{*}$r hgetall $k]
set newfields {}
foreach {k v} $fields {
lappend newfields [list $k $v]
foreach {f v} $fields {
set expirylist [{*}$r hexpiretime $k FIELDS 1 $f]
if {$expirylist eq (-1)} {
lappend newfields [list $f $v]
} else {
set e [lindex $expirylist 0]
lappend newfields [list $f $e $v] # TODO: extract the actual ttl value from the list in $e
}
}
set fields [lsort -index 0 $newfields]
foreach kv $fields {
tests/test_helper.tcl
+1
View File
@@ -34,6 +34,7 @@ set ::all_tests {
unit/type/set
unit/type/zset
unit/type/hash
unit/type/hash-field-expire
unit/type/stream
unit/type/stream-cgroups
unit/sort
tests/unit/memefficiency.tcl
+99
View File
@@ -503,6 +503,105 @@ run_solo {defrag} {
$rd_pubsub close
}
test "Active Defrag HFE: $type" {
r flushdb
r config resetstat
r config set hz 100
r config set activedefrag no
# TODO: Lower the threshold after defraging the ebuckets.
# Now just to ensure that the reference is updated correctly.
r config set active-defrag-threshold-lower 12
r config set active-defrag-cycle-min 65
r config set active-defrag-cycle-max 75
r config set active-defrag-ignore-bytes 1500kb
r config set maxmemory 0
r config set hash-max-listpack-value 512
r config set hash-max-listpack-entries 10
# Populate memory with interleaving hash field of same size
set n 3000
set fields 16 ;# make all the fields in an eblist.
set dummy_field "[string repeat x 400]"
set rd [redis_deferring_client]
for {set i 0} {$i < $n} {incr i} {
for {set j 0} {$j < $fields} {incr j} {
$rd hset h$i f$j $dummy_field
$rd hexpire h$i 9999999 FIELDS 1 f$j
$rd set "k$i$j" $dummy_field
}
}
for {set j 0} {$j < [expr $n*$fields]} {incr j} {
$rd read ; # Discard hset replies
$rd read ; # Discard hexpire replies
$rd read ; # Discard set replies
}
# Coverage for listpackex.
r hset h_lpex f0 $dummy_field
r hexpire h_lpex 9999999 FIELDS 1 f0
assert_encoding listpackex h_lpex
after 120 ;# serverCron only updates the info once in 100ms
if {$::verbose} {
puts "used [s allocator_allocated]"
puts "rss [s allocator_active]"
puts "frag [s allocator_frag_ratio]"
puts "frag_bytes [s allocator_frag_bytes]"
}
assert_lessthan [s allocator_frag_ratio] 1.05
# Delete all the keys to create fragmentation
for {set i 0} {$i < $n} {incr i} {
for {set j 0} {$j < $fields} {incr j} {
r del "k$i$j"
}
}
$rd close
after 120 ;# serverCron only updates the info once in 100ms
if {$::verbose} {
puts "used [s allocator_allocated]"
puts "rss [s allocator_active]"
puts "frag [s allocator_frag_ratio]"
puts "frag_bytes [s allocator_frag_bytes]"
}
assert_morethan [s allocator_frag_ratio] 1.35
catch {r config set activedefrag yes} e
if {[r config get activedefrag] eq "activedefrag yes"} {
# wait for the active defrag to start working (decision once a second)
wait_for_condition 50 100 {
[s total_active_defrag_time] ne 0
} else {
after 120 ;# serverCron only updates the info once in 100ms
puts [r info memory]
puts [r info stats]
puts [r memory malloc-stats]
fail "defrag not started."
}
# wait for the active defrag to stop working
wait_for_condition 500 100 {
[s active_defrag_running] eq 0
} else {
after 120 ;# serverCron only updates the info once in 100ms
puts [r info memory]
puts [r memory malloc-stats]
fail "defrag didn't stop."
}
# test the fragmentation is lower
after 120 ;# serverCron only updates the info once in 100ms
if {$::verbose} {
puts "used [s allocator_allocated]"
puts "rss [s allocator_active]"
puts "frag [s allocator_frag_ratio]"
puts "frag_bytes [s allocator_frag_bytes]"
}
assert_lessthan_equal [s allocator_frag_ratio] 1.5
}
}
if {$type eq "standalone"} { ;# skip in cluster mode
test "Active defrag big list: $type" {
r flushdb
tests/unit/other.tcl
+2 -1
View File
@@ -124,7 +124,8 @@ start_server {tags {"other"}} {
if {$::accurate} {set numops 10000} else {set numops 1000}
test {Check consistency of different data types after a reload} {
r flushdb
createComplexDataset r $numops usetag
# TODO: integrate usehexpire following next commit that will support replication
createComplexDataset r $numops {usetag usehexpire}
if {$::ignoredigest} {
set _ 1
} else {
tests/unit/pubsub.tcl
+13 -1
View File
@@ -353,17 +353,29 @@ start_server {tags {"pubsub network"}} {
$rd1 close
}
test "Keyspace notifications: hash events test" {
foreach {type max_lp_entries} {listpackex 512 hashtable 0} {
test "Keyspace notifications: hash events test ($type)" {
r config set hash-max-listpack-entries $max_lp_entries
r config set notify-keyspace-events Kh
r del myhash
set rd1 [redis_deferring_client]
assert_equal {1} [psubscribe $rd1 *]
r hmset myhash yes 1 no 0
r hincrby myhash yes 10
r hexpire myhash 999999 FIELDS 1 yes
r hexpireat myhash [expr {[clock seconds] + 999999}] NX FIELDS 1 no
r hpexpire myhash 5 FIELDS 1 yes
r hpersist myhash FIELDS 1 yes
assert_encoding $type myhash
assert_equal "pmessage * __keyspace@${db}__:myhash hset" [$rd1 read]
assert_equal "pmessage * __keyspace@${db}__:myhash hincrby" [$rd1 read]
assert_equal "pmessage * __keyspace@${db}__:myhash hexpire" [$rd1 read]
assert_equal "pmessage * __keyspace@${db}__:myhash hexpire" [$rd1 read]
assert_equal "pmessage * __keyspace@${db}__:myhash hexpire" [$rd1 read]
assert_equal "pmessage * __keyspace@${db}__:myhash hpersist" [$rd1 read]
$rd1 close
}
} ;# foreach
test "Keyspace notifications: stream events test" {
r config set notify-keyspace-events Kt
tests/unit/scan.tcl
+26
View File
@@ -277,6 +277,32 @@ proc test_scan {type} {
set res [r hscan hash 0 count 1000 novalues]
assert_equal [lsort $keys2] [lsort [lindex $res 1]]
}
test "{$type} HSCAN with large value $enc" {
r del hash
if {$enc eq {listpack}} {
set count 60
} else {
set count 170
}
set val1 [string repeat "1" $count]
r hset hash $val1 $val1
set val2 [string repeat "2" $count]
r hset hash $val2 $val2
set res [lsort [lindex [r hscan hash 0] 1]]
assert_equal $val1 [lindex $res 0]
assert_equal $val1 [lindex $res 1]
assert_equal $val2 [lindex $res 2]
assert_equal $val2 [lindex $res 3]
set res [lsort [lindex [r hscan hash 0 novalues] 1]]
assert_equal $val1 [lindex $res 0]
assert_equal $val2 [lindex $res 1]
}
}
foreach enc {listpack skiplist} {
tests/unit/type/hash-field-expire.tcl
+1151
View File
File diff suppressed because it is too large Load Diff