linux 共享内存段研究

无论是在unix或者是linux都有命令查看主机共享内存段的使用情况，以下研究来自linux平台。
一、查看Oracle用户共享内存段使用情况
[quote][oracle@capaa ~]$ ipcs

------ Shared Memory Segments --------
key shmid owner perms bytes nattch status
0x00000000 917504 oracle 600 196608 2 dest
0x00000000 950273 oracle 600 196608 2 dest
0x00000000 983042 oracle 600 196608 2 dest
0x00000000 1015811 oracle 600 196608 2 dest
0x00000000 1081349 oracle 600 196608 2 dest
0x00000000 1179656 oracle 600 196608 2 dest
0x4cd26024 524300 oracle 640 562044928 55
0x00000000 1900559 oracle 600 12288 2 dest
0x00000506 1802260 oracle 666 16384 0
0x00000000 1835029 oracle 600 196608 2 dest

------ Semaphore Arrays --------
key semid owner perms nsems
0xc7ef5bb0 98304 oracle 640 152
0xc7ef5bb1 131073 oracle 640 152
0xc7ef5bb2 163842 oracle 640 152

------ Message Queues --------
key msqid owner perms used-bytes messages [/quote]
简单说明：
shmid：表示共享内存段id
owner：表示共享内存段属于哪个用户，需要注意的是默认情况下600表示除root用户外不允许其他用户查看
bytes：表示共享内存段大小
nattch：表示连接到共享内存段的进程数
status：表示共享内存段的状态，dest表示该内存段被标记为destroyed状态。
二、查看共享内存段进程连接信息
1、指定shmid，查看进程连接信息
查看shmid为524300的具体信息，可用如下命令
[quote][oracle@capaa ~]$ ipcs -m -i 524300

Shared memory Segment shmid=524300
uid=505 gid=505 cuid=505 cgid=505
mode=0640 access_perms=0640
bytes=562044928 lpid=15216 cpid=16790 nattch=55
att_time=Wed Dec 9 13:44:45 2009
det_time=Wed Dec 9 13:44:45 2009
change_time=Fri Nov 27 15:58:44 2009 [/quote]
2、查看所有共享内存段，进程连接信息

[quote][oracle@capaa ~]$ ipcs -pm

------ Shared Memory Creator/Last-op --------
shmid owner cpid lpid
917504 oracle 30438 30967
950273 oracle 30440 1286
983042 oracle 31234 30967
1015811 oracle 31236 30967
1081349 oracle 31708 1275
1179656 oracle 31661 30967
524300 oracle 16790 15410
1900559 oracle 1315 29959
1802260 oracle 1286 0
1835029 oracle 1315 29959[/quote]

简单说明：
cpid表示创建共享内存段进程的id
lpid表示上一次访问该共享内存的进程id

三、查看Oracle使用共享内存段使用情况
1、使用Oracle小工具查看共享内存段使用情况
Oracle软件装好之后自带一个工具（sysresv），可以清晰的看到每个实例的内存使用情况，默认安装在$ORACLE_HOME/bin目录下，以下是sysresv的输出情况
[quote][oracle@capaa ~]$ sysresv

IPC Resources for ORACLE_SID "xe" :
Shared Memory:
ID KEY
524300 0x4cd26024
Semaphores:
ID KEY
98304 0xc7ef5bb0
131073 0xc7ef5bb1
163842 0xc7ef5bb2
Oracle Instance alive for sid "xe"[/quote]
可以看到SID为xe的实例使用了shmid为524300的共享内存段
2、使用oradebug命令行查看共享内存段使用情况
1）连接到需要查看的数据库中
2）取得连接的os pid
[quote][oracle@capaa ~]$ ps -ef|grep LOCAL=YES
oracle 16497 16496 0 13:57 ? 00:00:00 oraclexe (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))
3）根据os pid利用oradebug产生跟踪文件

SQL> oradebug setospid 16497
Statement processed.
SQL> oradebug ipc
Information written to trace file.
SQL> oradebug tracefile_name
/var/oracle/admin/xe/udump/xe_ora_16497.trc[/quote]
4）打开跟踪文件可以看到该实例共享内存使用情况，其中shmid为524300
[quote][oracle@capaa ~]$ cat /var/oracle/admin/xe/udump/xe_ora_16497.trc
/var/oracle/admin/xe/udump/xe_ora_16497.trc
Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - 64bit Production
With the Partitioning, OLAP and Data Mining options
ORACLE_HOME = /var/oracle/product/10.2.0/db_1
System name: Linux
Node name: capaa
Release: 2.6.18-53.el5xen
Version: #1 SMP Mon Nov 12 02:46:57 EST 2007
Machine: x86_64
Instance name: xe
Redo thread mounted by this instance: 1
Oracle process number: 55
Unix process pid: 16497, image: oracle@capaa (TNS V1-V3)

*** 2009-12-09 14:02:27.230
*** SERVICE NAME:(SYS$USERS) 2009-12-09 14:02:27.171
*** SESSION ID:(262.2351) 2009-12-09 14:02:27.171
Dump of unix-generic skgm context
areaflags 000000a7
realmflags 0000000f
mapsize 00001000
protectsize 00001000
lcmsize 00001000
seglen 00001000
largestsize 0000001000000000
smallestsize 0000000000400000
stacklimit 0x7fff2207dac0
stackdir -1
mode 640
magic acc01ade
Handle: 0x185252d0 `/var/oracle/product/10.2.0/db_1xe'
Dump of unix-generic realm handle `/var/oracle/product/10.2.0/db_1xe', flags = 00000000
Area #0 `Fixed Size' containing Subareas 0-0
Total size 00000000001edba8 Minimum Subarea size 00000000
Area Subarea Shmid Stable Addr Actual Addr
0 0 524300 0x00000060000000 0x00000060000000
Subarea size Segment size
00000000001ee000 0000000021802000
Area #1 `Variable Size' containing Subareas 2-2
Total size 0000000021000000 Minimum Subarea size 00400000
Area Subarea Shmid Stable Addr Actual Addr
1 2 524300 0x00000060800000 0x00000060800000
Subarea size Segment size
0000000021000000 0000000021802000
Area #2 `Redo Buffers' containing Subareas 1-1
Total size 0000000000612000 Minimum Subarea size 00000000
Area Subarea Shmid Stable Addr Actual Addr
2 1 524300 0x000000601ee000 0x000000601ee000
Subarea size Segment size
0000000000612000 0000000021802000
Area #3 `skgm overhead' containing Subareas 3-3
Total size 0000000000002000 Minimum Subarea size 00000000
Area Subarea Shmid Stable Addr Actual Addr
3 3 524300 0x00000081800000 0x00000081800000
Subarea size Segment size
0000000000002000 0000000021802000
Dump of Solaris-specific skgm context
sharedmmu 00000000
shareddec 0
used region 0: start 0000000040000000 length 000000007fff80000000
Maximum processes: = 300
Number of semaphores per set: = 152
Semaphores key overhead per set: = 4
User Semaphores per set: = 148
Number of semaphore sets: = 3
Semaphore identifiers: = 3
Semaphore List=
98304
-------------- system semaphore information -------------
------ Shared Memory Segments --------
key shmid owner perms bytes nattch status
0x00000000 917504 oracle 600 196608 2 dest
0x00000000 950273 oracle 600 196608 2 dest
0x00000000 983042 oracle 600 196608 2 dest
0x00000000 1015811 oracle 600 196608 2 dest
0x00000000 1081349 oracle 600 196608 2 dest
0x00000000 1179656 oracle 600 196608 2 dest
0x4cd26024 524300 oracle 640 562044928 55
0x00000000 1900559 oracle 600 12288 2 dest
0x00000506 1802260 oracle 666 16384 0
0x00000000 1835029 oracle 600 196608 2 dest
------ Semaphore Arrays --------
key semid owner perms nsems
0xc7ef5bb0 98304 oracle 640 152
0xc7ef5bb1 131073 oracle 640 152
0xc7ef5bb2 163842 oracle 640 152
------ Message Queues --------
key msqid owner perms used-bytes messages

四、进一步研究
通过以上命令可以查看创建Oracle实例的初始进程
[oracle@capaa ~]$ ipcs -mi 524300

Shared memory Segment shmid=524300
uid=505 gid=505 cuid=505 cgid=505
mode=0640 access_perms=0640
bytes=562044928 lpid=19796 cpid=16790 nattch=54
att_time=Wed Dec 9 14:10:26 2009
det_time=Wed Dec 9 14:10:35 2009
change_time=Fri Nov 27 15:58:44 2009 [/quote]
通过cpid=16790 可以知道pid为16790启动了Oracle实例，该进程只负责启动Oracle实例，启动完成后通过“自杀”结束进程，可以从两方面得到验证
1）启动日志
[quote]PMON started with pid=2, OS id=30345
PSP0 started with pid=3, OS id=30347
MMAN started with pid=4, OS id=30349
DBW0 started with pid=5, OS id=30351
LGWR started with pid=6, OS id=30353
CKPT started with pid=7, OS id=30355
SMON started with pid=8, OS id=30357
RECO started with pid=9, OS id=30359
CJQ0 started with pid=10, OS id=30361
MMON started with pid=11, OS id=30363[/quote]
2）从v$process表中查相关信息
[quote]SQL> select addr,program from v$process where pid=1;

ADDR PROGRAM
---------------- ------------------------------------------------
00000000817DA930 PSEUDO[/quote]
PSEUDO表示该进程已被killed