sharedpool之二：freelists/sharedpoollrulist

介绍free lists及shared pool lru list. free list 空闲列表 按bucket划分，共有255个，bucket 0---bucket 254 每个bucket上挂有一个 chunk list； free lists上的都是未使用的chunk RESERVED FREE LISTS: RESERVED FREE LISTS上的bucket个数我DUMP出来的是15

介绍free lists及shared pool lru list.

free list 空闲列表

按bucket划分，共有255个，bucket 0---bucket 254
每个bucket上挂有一个 chunk list； free lists上的都是未使用的chunk
RESERVED FREE LISTS:
RESERVED FREE LISTS上的bucket个数我DUMP出来的是15个。
保留FREE LISTS，在SQL语句所需CHUNK大于4400bytes时，会在RESERVED FREE LISTS中查找空闲CHUNK。
如果SQL语句所需CHUNK不大于4400bytes时，只会在free list 中查找CHUNK。
这个是由隐含参数控制的：_shared_pool_reserved_min_alloc minimum allocation size in bytes for reserved area ，默认值4400
###################
DUMP 共享池查看free lists/bucket/RESERVED FREE LISTS结构：--用新建会话来做
alter session set events 'immediate trace name heapdump level 2';
select value from v$diag_info where name like 'De%';
/u01/diag/rdbms/bys3/bys3/trace/bys3_ora_7876.trc
查看TRACE文件内容： --找这一段的方法：VI搜索HEAP DUMP
Chunk 2bffa844 sz= 22460 freeable "character set m"
Total heap size =146798680 -- 146798680/1024/1024 --139.99813 初始化参数：shared_pool_size--140M
FREE LISTS: ------空闲列，可以明显看出bucket大小分配的规律--从小到大，共有255个buckets，从16bytes到64k，采用此方法分配内存，可以有效的减少内存碎片。每个Bucket之间都用double linked 相互连接
Bucket 0 size=16
Chunk 2bc00048 sz= 0 kghdsx
Bucket 1 size=20 20字节的Bucket 1，有很多个Chunk，节约篇幅省略了
Chunk 23a60468 sz= 20 free " "
Chunk的地址、大小、状态
Chunk 23ceb498 sz= 20 free " "
Chunk 237fcde4 sz= 20 free " "
Bucket 2 size=24 --Bucket 2 --24字节
Chunk 245b13e4 sz= 24 free " "
Chunk 23ace7c0 sz= 24 free " "
Chunk 239c5a28 sz= 24 free " "
Bucket 3 size=28
Chunk 24540e9c sz= 28 free " "
Chunk 2521209c sz= 28 free " "
Chunk 23483448 sz= 28 free " "
……
Bucket 252 size=16396
Bucket 253 size=32780
Bucket 254 size=65548
Total free space = 518232
RESERVED FREE LISTS: --保留FREE LISTS，解析方法同上。保留池中CHUNK都比较大
Reserved bucket 0 size=16
Chunk 23420320 sz= 676 R-free " "
Chunk 23427b94 sz= 3420 R-free " "
Chunk 2342618c sz= 952 R-free " "
Chunk 23800050 sz= 1040 R-free " "
Chunk 23400050 sz= 2824 R-free " "
Chunk 25bff028 sz= 4032 R-free " "
Chunk 293ff788 sz= 2144 R-free " "
Reserved bucket 1 size=4400 --这个如果用的会，4564-4400，剩下164字节会成为新CHUNK
Chunk 23430a40 sz= 4564 R-free " "
Reserved bucket 2 size=8204
Reserved bucket 3 size=8460
Reserved bucket 4 size=8464
Reserved bucket 5 size=8468
Reserved bucket 6 size=8472
Chunk 2342b988 sz= 9136 R-free " "
Reserved bucket 7 size=9296
Reserved bucket 8 size=9300
Reserved bucket 9 size=12320
Reserved bucket 10 size=12324
Reserved bucket 11 size=16396
Chunk 234161f8 sz= 16448 R-free " "
Reserved bucket 12 size=32780
Reserved bucket 13 size=65548
Chunk 23401d50 sz= 72296 R-free " "
Chunk 23815668 sz= 125312 R-free " "
Chunk 23c00050 sz= 180380 R-free " "
Reserved bucket 14 size=1990652
Total reserved free space = 6712468 --总空闲保留空间是6.4M，shared_pool_reserved_size 初始化参数大小是 7M，用了0.6M
######################

shared pool LRU链

shared pool LRU链上挂的都是recreate状态的chunk,一个SQL语句可能需要多个CHUNK，在LRU链上找到recreate状态的chunk，然后在recreate状态的chunk下再下挂freeabl状态的CHUNK，--避免全部CHUNK在LRU链上导致LRU链太长。
TRACE文件中找到关于(lru first)的一段，方法同上：
Reserved bucket 14 size=1990652
Total reserved free space = 6712468
UNPINNED RECREATABLE CHUNKS (lru first):
Chunk 246c9848 sz= 348 recreate "KGLHD " latch=(nil) --latch状态为空，Chunk SIZE是348字节，状态recreate，
Chunk 237cb10c sz= 4096 recreate "KGLH0^b9197c6e " latch=(nil)
Chunk 24bb5df0 sz= 364 recreate "KGLHD " latch=(nil)
Chunk 241aa1b8 sz= 4096 recreate "KGLH0^59449e50 " latch=(nil)
Chunk 252640a0 sz= 364 recreate "KGLHD " latch=(nil)
Chunk 23a619a0 sz= 4096 recreate "KGLH0^d5f1e0d7 " latch=(nil)
Chunk 23465600 sz= 348 recreate "KGLHD " latch=(nil)
Chunk 2346575c sz= 1036 recreate "KGLHD " latch=(nil)
Chunk 23465b68 sz= 4096 recreate "KGLH0^c6e0d102 " latch=(nil) --一个recreate状态CHUNK下的多个freeable状态CHUNK
ds 24bdecb0 sz= 4096 ct= 1
Chunk 23466b68 sz= 4096 freeable "SQLA^1536bb77 " ds=0x23db5bd8
Chunk 23467b68 sz= 144 freeable "KGLDA "
Chunk 23467bf8 sz= 4096 freeable "KGLH0^ba3f9b05 " ds=0x2425e238