[20190416]process allocation latch.txt

--//看連結:

--//裡面提到:

Oracle no longer uses "repeatedly spin and sleep" approach. The process spins and waits only once. The pseudo code for

contemporary latch acquisition should looks like:

--//Oracle不再使用"反覆旋轉和睡眠"的方法。程式只旋轉一次並等待一次。當代鎖存獲取的虛擬碼應該如下所示：

Immediate latch get

Spin latch get

Add the process to the queue of latch waiters

Sleep until posted

Only "process allocation" latch shows different system calls. On Solaris this latch waits using pollsys() system call.

On Linux and HP-UX it uses select():

--//只有"process allocation"鎖存顯示不同的系統呼叫。在Solaris上，這個閂鎖使用pollsys()系統呼叫等待。在Linux和HP-UX上，它

--//使用select()：

--//先不管其它latch的改進,先看看process allocation的情況,透過測試說明問題.

--//另外說明千萬不要在生產系統做這樣的測試,阻塞process allocation 拴鎖會導致全部使用者無法登入!!

1.環境:

SCOTT@book> @ ver1

PORT_STRING VERSION BANNER

------------------------------ -------------- --------------------------------------------------------------------------------

x86_64/Linux 2.4.xx 11.2.0.4.0 Oracle Database 11g Enterprise Edition Release 11.2.0.4.0 - 64bit Production

SYS@book> @ laddr 'process allocation'

old 1: select addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch where lower(name) like '%'||lower('&&1')||'%'

new 1: select addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch where lower(name) like '%'||lower('process allocation')||'%'

ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME

---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------

0000000060009F88 process allocation 7 8 8595 32 25339 4103 15 0 0 15 206377434

0000000060011F20 OS process allocation 4 95 74998 2 0 0 0 0 0 2 0

old 1: select addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where lower(name) like '%'||lower('&&1')||'%'

new 1: select addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where lower(name) like '%'||lower('process allocation')||'%'

ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME

---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------

0000000060009F88 process allocation 7 8 8595 32 25339 4103 15 0 0 15 206377434

0000000060011F20 OS process allocation 4 95 74998 2 0 0 0 0 0 2 0

old 1: select addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_children where lower(name) like '%'||lower('&&1')||'%'

new 1: select addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_children where lower(name) like '%'||lower('process allocation')||'%'

--//可以看出這個拴鎖沒有子拴鎖.僅僅1個父拴鎖.

SYS@book> select * from exclusive_latches where name='process allocation';

VERSION LATCH# NAME S

----------- ------ ---------------------------------------- -

11.2.0.4.0 8 process allocation N

--//拴鎖型別是exclusive型別.測試可以參考連結:

http://blog.itpub.net/267265/viewspace-2641370/ => [20190415]11g下那些latch是共享的.txt

http://blog.itpub.net/267265/viewspace-2641482/ => [20190416]11g下那些latch是Exclusive的.txt

2.測試:

--//建立測試指令碼,注意測試前最好預先登入幾個會話sys使用者,避免測試後無法在登入.

$ cat p1.sh

#! /bin/bash

vdate=$(date '+%H%M%S')

echo $vdate

source peek.sh "$1" 12 | timestamp.pl >| /tmp/peekx_${vdate}.txt &

seq 12 | xargs -I{} echo -e 'sqlplus -s -l / as sysdba <<< @latch_free\nsleep 1' | bash >| /tmp/latch_free_${vdate}.txt &

sleep 1

# 引數如下: @ exclusive_latch.txt latch_name willing why where sleep_num

sqlplus /nolog @ exclusive_latch.txt "$1" 1 4 5 10 > /dev/null &

sleep 1

strace -ftT -o /tmp/pp_${vdate}.txt sqlplus -s -l scott/book <<< 'select sysdate from dual ;' &

wait

$ . p1.sh 'process allocation'

171724

SYSDATE

-------------------

2019-04-16 17:17:35

[1] Done source peek.sh "$1" 12 | timestamp.pl >|/tmp/peekx_${vdate}.txt

[3]- Done sqlplus /nolog @ exclusive_latch.txt "$1" 1 4 5 10 > /dev/null

[4]+ Done strace -ftT -o /tmp/pp_${vdate}.txt sqlplus -s -l scott/book <<< 'select sysdate from dual ;'

[2]+ Done seq 12 | xargs -I{} echo -e 'sqlplus -s -l / as sysdba <<< @latch_free\nsleep 1' | bash >|/tmp/latch_free_${vdate}.txt

$ grep -v '^.*: $' /tmp/peekx_171724.txt | cut -c10- | uniq -c

1 SYSDATE LADDR

1 ------------------- ----------------

1 2019-04-16 17:17:24 0000000060009F88

1 Statement processed.

1 [060009F88, 060009FC4) = 00000000 00000000 00002153 00020008 00000007 0000008A 8620F338 00000000 00000FEE 0000001E 0000000F 00000000 0B2D0C03 00000000 000059FC

10 [060009F88, 060009FC4) = 0000001B 00000000 00002155 00020008 00000007 00000005 00000004 00000000 00000FEF 0000001E 0000000F 00000000 0B2D0C03 00000000 000059FC

1 [060009F88, 060009FC4) = 00000000 00000000 00002159 00020008 00000007 0000008A 86212560 00000000 00000FF1 00000020 0000000F 00000000 0C4D11DA 00000000 000062FB

--//才發現我的指令碼寫的有問題,seq 12 ...那行不停登陸看latch情況也被阻塞.而且這行總是最後完成也說明這點.這樣執行效率太低..

--//不過不影響測試結論.指令碼修改如下:

$ cat p2.sh

#! /bin/bash

vdate=$(date '+%H%M%S')

echo $vdate

source peek.sh "$1" 12 | timestamp.pl >| /tmp/peekx_${vdate}.txt &

sqlplus -s -l / as sysdba <<EOF >| /tmp/latch_free_${vdate}.txt &

$(seq 12 | xargs -I {} echo -e '@latch_free \n host sleep 1')

EOF

sleep 1

# 引數如下: @ exclusive_latch.txt latch_name willing why where sleep_num

sqlplus /nolog @ exclusive_latch.txt "$1" 1 4 5 10 > /dev/null &

sleep 1

strace -ftT -o /tmp/pp_${vdate}.txt sqlplus -s -l scott/book <<< 'select sysdate from dual ;' &

wait

----------------

$ awk '{print $3}' /tmp/pp_171724.txt | uniq -c | sort -nr | head

1080 select(0,

174 getrusage(RUSAGE_SELF,

79 getrusage(RUSAGE_SELF,

49 getrusage(RUSAGE_SELF,

30 read(4,

29 read(4,

27 getrusage(RUSAGE_SELF,

--//出現1080次呼叫

$ grep 'select(' /tmp/pp_171724.txt | head

18887 17:17:26 select(0, [], [], [], {0, 8000}) = 0 (Timeout) <0.008083>

18887 17:17:26 select(0, [], [], [], {0, 8000}) = 0 (Timeout) <0.008114>

18887 17:17:26 select(0, [], [], [], {0, 8000}) = 0 (Timeout) <0.008117>

18887 17:17:26 select(0, [], [], [], {0, 8000}) = 0 (Timeout) <0.008148>

18887 17:17:26 select(0, [], [], [], {0, 8000}) = 0 (Timeout) <0.008121>

18887 17:17:26 select(0, [], [], [], {0, 8000}) = 0 (Timeout) <0.008119>

18887 17:17:26 select(0, [], [], [], {0, 8000}) = 0 (Timeout) <0.008126>

18887 17:17:26 select(0, [], [], [], {0, 8000}) = 0 (Timeout) <0.008120>

18887 17:17:26 select(0, [], [], [], {0, 8000}) = 0 (Timeout) <0.008119>

--//你可以發現語句執行時間在'2019-04-16 17:17:35',與select第1次呼叫相差9秒.而且每次呼叫間隔是8000微秒.

--//順便看看select 呼叫總共花了多少時間.

$ awk '/select/ {print $NF}' /tmp/pp_171724.txt | tr -d '<>' | xargs | sed 's/ /+/g' | bc -l

8.787933

--//基本符合.

3.oracle為什麼這樣設計:

--//我想應該是登入儘可能快.採用這樣設計延遲更小一些.

SYS@book> @ pt2 'select * from x$kslltr where KSLLTNAM=''process allocation''';

old 6: from table(xmlsequence(cursor( &1 )))

new 6: from table(xmlsequence(cursor( select * from x$kslltr where KSLLTNAM='process allocation' )))

ROW_NUM COL_NUM COL_NAME COL_VALUE

------- ------- --------------- -------------------

1 1 ADDR 00007F0270051660

2 INDX 8

3 INST_ID 1

4 KSLLTADDR 0000000060009F88

5 KSLLTNUM 8

6 KSLLTNGT 4096

7 KSLLTNFA 15

8 KSLLTWGT 8574

9 KSLLTWFF 32

10 KSLLTWSL 25339

11 KSLLTHST0 15

12 KSLLTCNM 0

13 KSLLTWHR 138

14 KSLLTWHY 2250314920

15 KSLLTWTT 206377434

16 CLASS_KSLLT 2

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

17 KSLLTLVL 7

18 KSLLTHSH 2600548697

19 KSLLTNAM process allocation

20 KSLLTWKC 0

21 KSLLTWTH 0

22 KSLLTMSX 0

23 KSLLTMXS 0

24 KSLLTMSW 0

25 KSLLTWSX 0

26 KSLLTWXS 0

27 KSLLTWSW 0

28 KSLLTHST1 0

29 KSLLTHST2 0

30 KSLLTHST3 0

31 KSLLTHST4 0

32 KSLLTHST5 0

33 KSLLTHST6 0

34 KSLLTHST7 0

35 KSLLTHST8 0

36 KSLLTHST9 0

37 KSLLTHST10 0

38 KSLLTHST11 0

39 KSLLTHDT 0

40 KSLLTDNT 0

41 KSLLTWTW 0

42 YIELDS_KSLLT 0

43 MISSES_WL_KSLLT 0

44 YIELDS_WL_KSLLT 0

45 SLEEPS_WL_KSLLT 0

45 rows selected.

--//latch_name='process allocation'使用拴鎖類CLASS_KSLLT=2.

SYS@book> select CLASS_KSLLT,count(*) from x$kslltr group by CLASS_KSLLT;

CLASS_KSLLT COUNT(*)

----------- ----------

2 1

0 581

--//可以發現僅僅1個CLASS_KSLLT=2的情況.就是'process allocation'.

SYS@book> select * from x$ksllclass ;

ADDR INDX INST_ID SPIN YIELD WAITTIME SLEEP0 SLEEP1 SLEEP2 SLEEP3 SLEEP4 SLEEP5 SLEEP6 SLEEP7

---------------- ---- ------- ----- ----- -------- ------ ------ ------ ------ ------ ------ ------ ------

00000000861986C0 0 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

00000000861986EC 1 1 20000 0 1 1000 1000 1000 1000 1000 1000 1000 1000

0000000086198718 2 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

0000000086198744 3 1 20000 0 1 1000 1000 1000 1000 1000 1000 1000 1000

0000000086198770 4 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

000000008619879C 5 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

00000000861987C8 6 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

00000000861987F4 7 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

8 rows selected.

3.更改拴鎖類看看:

--//前面已經知道name=process allocation'的latch#=8.安全起見建立pfile,修改它啟動看看.

--//alter system set "_latch_classes"='8:3' scope=spfile;

SYS@book> create pfile='/tmp/@.ora' from spfile ;

File created.

--//修改/tmp/book.ora檔案加入如下:

*._latch_classes='8:3'

--//關閉資料庫並使用該引數檔案重啟資料庫

SYS@book> startup pfile=/tmp/@.ora

ORACLE instance started.

Total System Global Area 643084288 bytes

Fixed Size 2255872 bytes

Variable Size 205521920 bytes

Database Buffers 427819008 bytes

Redo Buffers 7487488 bytes

Database mounted.

Database opened.

SYS@book> select CLASS_KSLLT,KSLLTNAM from x$kslltr where KSLLTNAM='process allocation';

CLASS_KSLLT KSLLTNAM

----------- ------------------

3 process allocation

--//現在是3.

$ . p1.sh 'process allocation'

174817

SYSDATE

-------------------

2019-04-16 17:48:29

[1] Done source peek.sh "$1" 12 | timestamp.pl >|/tmp/peekx_${vdate}.txt

[3]- Done sqlplus /nolog @ exclusive_latch.txt "$1" 1 4 5 10 > /dev/null

[4]+ Done strace -ftT -o /tmp/pp_${vdate}.txt sqlplus -s -l scott/book <<< 'select sysdate from dual ;'

[2]+ Done seq 12 | xargs -I{} echo -e 'sqlplus -s -l / as sysdba <<< @latch_free\nsleep 1' | bash >|/tmp/latch_free_${vdate}.txt

$ grep -v '^.*: $' /tmp/peekx_174817.txt | cut -c10- | uniq -c

1 SYSDATE LADDR

1 ------------------- ----------------

1 2019-04-16 17:48:18 0000000060009F88

1 Statement processed.

1 [060009F88, 060009FC4) = 00000000 00000000 0000004E 00030008 00000007 00000081 00000000 00000000 00000024 00000000 00000000 00000000 00000000 00000000 00000000

10 [060009F88, 060009FC4) = 00000018 00000000 00000051 00030008 00000007 00000005 00000004 00000000 00000025 00000000 00000000 00000000 00000000 00000000 00000000

1 [060009F88, 060009FC4) = 00000000 00000000 00000055 00030008 00000007 0000008A 862103F0 00000000 00000026 00000002 00000001 00000000 011F5A13 00000000 00003F80

$ awk '/select/ {print $NF}' /tmp/pp_174817.txt | wc

7356 7356 80916

--//呼叫次數更多對比前面的測試.

$ grep 'select' /tmp/pp_174817.txt | head

19323 17:48:20 select(0, [], [], [], {0, 1000}) = 0 (Timeout) <0.001103>

19323 17:48:20 select(0, [], [], [], {0, 1000}) = 0 (Timeout) <0.001107>

19323 17:48:20 select(0, [], [], [], {0, 1000}) = 0 (Timeout) <0.001080>

19323 17:48:20 select(0, [], [], [], {0, 1000}) = 0 (Timeout) <0.001103>

19323 17:48:20 select(0, [], [], [], {0, 1000}) = 0 (Timeout) <0.001102>

19323 17:48:20 select(0, [], [], [], {0, 1000}) = 0 (Timeout) <0.001089>

19323 17:48:20 select(0, [], [], [], {0, 1000}) = 0 (Timeout) <0.001056>

19323 17:48:20 select(0, [], [], [], {0, 1000}) = 0 (Timeout) <0.001087>

19323 17:48:20 select(0, [], [], [], {0, 1000}) = 0 (Timeout) <0.001085>

19323 17:48:20 select(0, [], [], [], {0, 1000}) = 0 (Timeout) <0.001080>

--//現在的時間間隔是0.001000秒.

$ awk '/select/ {print $NF}' /tmp/pp_174817.txt | tr -d '<>' | xargs | sed 's/ /+/g' | bc -l

8.048958

--//select呼叫次數多了,但是時間消耗上比前面少(8.787933-8.048958 = .738975),

--//可以看出spin的消耗更大.spin次數也增加了.我的理解在每次select之間做spin,次數設計上不是2000次而是20000次.

--//不成功呼叫select,如何反覆.

--//看select * from x$ksllclass ;的顯示.

4.實際上還可以定製sleep的引數:

--//alter system set "_latch_class_3"="100 0 1 10000 20000 30000 40000 50000 60000 70000 80000" scope=spfile;

--//修改/tmp/book.ora檔案加入如下:

*._latch_classes='8:3'

*._latch_class_3='100 0 1 10000 20000 30000 40000 50000 60000 70000 80000'

--//關閉資料庫並使用該引數檔案重啟資料庫

SYS@book> startup pfile=/tmp/@.ora

ORACLE instance started.

Total System Global Area 643084288 bytes

Fixed Size 2255872 bytes

Variable Size 205521920 bytes

Database Buffers 427819008 bytes

Redo Buffers 7487488 bytes

Database mounted.

Database opened.

SYS@book> show parameter latch

NAME TYPE VALUE

-------------- ------ -------------------------------------------------------

_latch_class_3 string 100 0 1 10000 20000 30000 40000 50000 60000 70000 80000

_latch_classes string 8:3

SYS@book> select * from x$ksllclass ;

ADDR INDX INST_ID SPIN YIELD WAITTIME SLEEP0 SLEEP1 SLEEP2 SLEEP3 SLEEP4 SLEEP5 SLEEP6 SLEEP7

---------------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ----------

00000000861986C0 0 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

00000000861986EC 1 1 20000 0 1 1000 1000 1000 1000 1000 1000 1000 1000

0000000086198718 2 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

0000000086198744 3 1 100 0 1 10000 20000 30000 40000 50000 60000 70000 80000

0000000086198770 4 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

000000008619879C 5 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

00000000861987C8 6 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

00000000861987F4 7 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000

8 rows selected.

$ . p2.sh 'process allocation'

085808

[1] Done sqlplus -s -l / as sysdba >|/tmp/latch_free.txt <<EOF

$(seq 12 | xargs -I {} echo -e '@latch_free \n host sleep 1')

EOF

SYSDATE

-------------------

2019-04-17 08:58:19

[2] Done source peek.sh "$1" 12 | timestamp.pl >|/tmp/peekx_${vdate}.txt

[4]- Done sqlplus /nolog @ exclusive_latch.txt "$1" 1 4 5 10 > /dev/null

[5]+ Done strace -ftT -o /tmp/pp_${vdate}.txt sqlplus -s -l scott/book <<< 'select sysdate from dual ;'

[3]+ Done sqlplus -s -l / as sysdba >|/tmp/latch_free_${vdate}.txt <<EOF

$(seq 12 | xargs -I {} echo -e '@latch_free \n host sleep 1')

EOF

$ grep -v '^.*: $' /tmp/peekx_085808.txt | cut -c10- | uniq -c

1 SYSDATE LADDR

1 ------------------- ----------------

1 2019-04-17 08:58:08 0000000060009F88

1 Statement processed.

1 [060009F88, 060009FC4) = 00000000 00000000 00000058 00030008 00000007 00000081 00000000 00000000 00000028 00000000 00000000 00000000 00000000 00000000 00000000

10 [060009F88, 060009FC4) = 0000001B 00000000 0000005A 00030008 00000007 00000005 00000004 00000000 00000029 00000000 00000000 00000000 00000000 00000000 00000000

1 [060009F88, 060009FC4) = 00000000 00000000 0000005C 00030008 00000007 00000081 00000000 00000000 0000002A 00000001 00000000 00000000 00889761 00000000 00000073

$ grep 'select' /tmp/pp_085808.txt | head -20

23504 08:58:10 select(0, [], [], [], {0, 10000}) = 0 (Timeout) <0.010084>

23504 08:58:10 select(0, [], [], [], {0, 20000}) = 0 (Timeout) <0.020129>

23504 08:58:10 select(0, [], [], [], {0, 30000}) = 0 (Timeout) <0.030121>

23504 08:58:10 select(0, [], [], [], {0, 40000}) = 0 (Timeout) <0.040120>

23504 08:58:10 select(0, [], [], [], {0, 50000}) = 0 (Timeout) <0.050132>

23504 08:58:10 select(0, [], [], [], {0, 60000}) = 0 (Timeout) <0.060148>

23504 08:58:10 select(0, [], [], [], {0, 70000}) = 0 (Timeout) <0.070153>

23504 08:58:10 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080179>

23504 08:58:10 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080151>

23504 08:58:10 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080155>

23504 08:58:10 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080160>

23504 08:58:11 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080155>

23504 08:58:11 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080158>

23504 08:58:11 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080153>

23504 08:58:11 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080163>

23504 08:58:11 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080165>

23504 08:58:11 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080163>

23504 08:58:11 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080157>

23504 08:58:11 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080173>

23504 08:58:11 select(0, [], [], [], {0, 80000}) = 0 (Timeout) <0.080152>

...

--//開始select時間是0.01,0.02,0.03,0.04,0.05,0.06,0.07,0.08,0.08....

$ awk '/select/ {print $NF}' /tmp/pp_085808.txt | tr -d '<>' | xargs | sed 's/ /+/g' | bc -l

8.938015

--//非常接近9秒,說明spin消耗很小,每次select之間spin次數是100.而且休眠時間比原來長.

5.繼續測試,修改裡面引數yield對應x$ksllclass.yield看看:

--//修改如下:

*._latch_class_3='100 1 1 10000 20000 30000 40000 50000 60000 70000 80000'

SYS@book> select * from x$ksllclass where indx=3;

ADDR INDX INST_ID SPIN YIELD WAITTIME SLEEP0 SLEEP1 SLEEP2 SLEEP3 SLEEP4 SLEEP5 SLEEP6 SLEEP7

---------------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ----------

0000000086198744 3 1 100 1 1 10000 20000 30000 40000 50000 60000 70000 80000

--//yield=1

--//執行p2.sh.僅僅貼出測試結果.

--// /tmp/pp_090552.txt

23747 09:05:54 sched_yield() = 0 <0.000019>

23747 09:05:54 select(0, [], [], [], {0, 10000}) = 0 (Timeout) <0.010112>

23747 09:05:54 sched_yield() = 0 <0.000027>

23747 09:05:54 select(0, [], [], [], {0, 20000}) = 0 (Timeout) <0.020130>

23747 09:05:54 sched_yield() = 0 <0.000027>

23747 09:05:54 select(0, [], [], [], {0, 30000}) = 0 (Timeout) <0.030124>

23747 09:05:55 sched_yield() = 0 <0.000032>

23747 09:05:55 select(0, [], [], [], {0, 40000}) = 0 (Timeout) <0.040122>

23747 09:05:55 sched_yield() = 0 <0.000032>

23747 09:05:55 select(0, [], [], [], {0, 50000}) = 0 (Timeout) <0.050173>

23747 09:05:55 sched_yield() = 0 <0.000032>

23747 09:05:55 select(0, [], [], [], {0, 60000}) = 0 (Timeout) <0.060134>