 |
≫ |
|
|
|
パッチ名: PHCO_34510
パッチ摘要: s700_800 11.23 ksh(1)累積パッチ
作成日: 07/03/27
公開日: 07/05/09
ハードウェアプラットフォームおよびOSリリース:
s700: 11.23
s800: 11.23
現象:
PHCO_34510:
1.(SR:8606420313 CR:JAGaf80143)
関数内で"typeset -i"を使って整数型変数を定義すると、ksh(1)内でメモリリ
ークが起きます。
2.(SR:8606435240 CR:JAGaf93696)
ksh(1)のマンページでの"typeset -f"機能に関する説明が不明確です。
3.(SR:8606442247 CR:JAGag00047)
サブシェルによるコマンドリストの実行中にリスト内のあるシェル組み込みコ
マンドがエラーになると、シェルが終了してしまうため、リスト内の後続のコ
マンドが実行されません。
4.(SR:8606462549 CR:JAGag18411)
未初期化または空の変数の算術比較が行われるため、算術式から不正な結果が
返されることがあります。
問題点の説明:
PHCO_34510:
1.(SR:8606420313 CR:JAGaf80143)
関数内で"typeset -i"を使って整数型変数を定義すると、ksh(1)内でメモリリ
ークが起きていました。"ps -l"やglanceの出力によると、"typeset -i"を使
用する関数が呼び出されるたびに、整数型変数のサイズと等価なメモリリーク
が起きています。つまり、ksh(1)は、関数のリターン時に整数型変数用のメモ
リを割り当て解除していませんでした。
問題の再現手順:
1) $cat ./script
#!/usr/bin/ksh
function CallMe
{
typeset -i var1=0
}
while true
do
CallMe
done
2) 上記のスクリプトをバックグラウンドで実行します。
$/usr/bin/ksh ./script &
3) "ps -l"の出力の"SZ"フィールド(プロセスサイズ)を監視します。関数
(CallMe)を呼び出すたびに"SZ"フィールド値が増加していることがわかり
ます。
解決方法:
関数内で"typeset -i"を使って作成された整数型変数に割り当てたメモリを解
放するようにksh(1)を修正しました。
2.(SR:8606435240 CR:JAGaf93696)
スクリプトからksh(1)を呼び出すと、"typeset -f"機能は関数本体を表示しま
せん。これは設計上の制限事項です。ところが、このことがksh(1)のマンペー
ジに記載されていませんでした。"typeset -f"は対話型専用に設計された機能
です。関数の読み取り時にヒストリファイルが存在しないかnologオプション
がONの場合、ksh(1)は関数定義を表示しません。
問題の再現手順:
1) $ cat x.sh
function f {
echo "This is function f"
}
typeset -f
2) スクリプトを実行します。
$ ksh x.sh
function f
"typeset -f"機能は、マンページどおりに、関数本体を表示できません。
解決方法:
"typeset -f"機能の正しい使用法を記載するようにksh(1)のマンページを修正
しました。
3.(SR:8606442247 CR:JAGag00047)
ksh(1)サブシェルによるコマンドリストの実行中にリスト内のあるシェル組み
込みコマンドがエラーになると、シェルが終了していたため、リスト内の後続
のコマンドが実行されませんでした。
解決方法:
エラーが起きてもリスト内のすべてのコマンドを実行するようにksh(1)を修正
しました。
4.(SR:8606462549 CR:JAGag18411)
算術式の計算時に、未初期化または空のローカル変数を使用することがありま
した。そのため、算術式から不正な値が返されていました。
解決方法:
正しい結果を返すようにksh(1)を修正しました。
-----------------------------------------------------------------------------
Patch Name: PHCO_34510
Patch Description: s700_800 11.23 ksh(1) cumulative patch
Creation Date: 07/03/27
Post Date: 07/05/09
Hardware Platforms - OS Releases:
s700: 11.23
s800: 11.23
Products: N/A
Filesets:
OS-Core.CORE-ENG-A-MAN,fr=B.11.23,fa=HP-UX_B.11.23_IA/PA,v=HP
OS-Core.UX2-CORE,fr=B.11.23,fa=HP-UX_B.11.23_IA,v=HP
OS-Core.UX2-CORE,fr=B.11.23,fa=HP-UX_B.11.23_PA,v=HP
Automatic Reboot?: No
Status: General Release
Critical:
Yes
PHCO_34510: MEMORY_LEAK
Integer variables defined with 'typeset -i' inside
a function results in memory leak on function
return.
PHCO_31552: MEMORY_LEAK
Category Tags:
defect_repair general_release critical memory_leak
Path Name: /hp-ux_patches/s700_800/11.X/PHCO_34510
Symptoms:
PHCO_34510:
( SR:8606420313 CR:JAGaf80143 )
ksh(1) leaks memory when typeset -i is used to define
an integer variable in a function.
( SR:8606435240 CR:JAGaf93696 )
ksh(1) man page is not clear about the typeset -f
functionality.
( SR:8606442247 CR:JAGag00047 )
In a subshell executing a list of commands, if one of the
shell built-in command in the list encounters an error,
the shell exits thereby not executing the subsequent
commands in the list.
( SR:8606462549 CR:JAGag18411 )
The arithmetic comparisons of uninitialized or empty
variables sometimes return incorrect results.
PHCO_32856:
( SR:8606388635 CR:JAGaf48784 )
The shell variable - TMOUT, doesn't function as expected
in sub-shells, under certain circumstances.
PHCO_31552:
( SR:8606350351 CR:JAGaf11171 )
Large number of background jobs running on an
interactive ksh shell terminates it.
( SR:8606350366 CR:JAGaf11186 )
The ksh shell displays wrong pid for the signalled child.
( SR:8606369179 CR:JAGaf29723 )
In ksh shell, "set -A" leaks memory.
( SR:8606369780 CR:JAGaf30323 )
In ksh(1), calling wait on an already completed process
does not return the correct exit value.
Defect Description:
PHCO_34510:
( SR:8606420313 CR:JAGaf80143 )
ksh(1) leaks memory when typeset -i is used to define an
integer variable in a function. Output of 'ps -l' or glance
shows memory-leak equivalent to size of integer variable
every time the function using 'typeset -i' is called, ie
ksh(1) is not de-allocating memory for integer variables on
function return.
Problem reproduction:
1.$cat ./script
#!/usr/bin/ksh
function CallMe
{
typeset -i var1=0
}
while true
do
CallMe
done
2. Execute the above script in background
$/usr/bin/ksh ./script &
3. Monitor the process size by looking at the 'SZ' field
of 'ps -l' output. 'SZ' field value keeps increasing
every time the function(CallMe) is called.
Resolution:
ksh(1) has been modified to free memory allocated to
integer variables created using 'typeset -i' inside a
function.
( SR:8606435240 CR:JAGaf93696 )
ksh(1) typeset -f feature does not list the function
body when invoked through scripts. The problem is the
design limitation, that was missed in ksh(1) man page
documentation. The typeset -f is only for the
interactive case, as per the design. ksh(1) will not
display the function definitions if the history file
is lost or nolog option was on when function was read.
Steps For Problem Reproduction:
1.
$ cat x.sh
function f {
echo "This is function f"
}
typeset -f
2. Execute the script.
$ ksh x.sh
function f
The typeset -f feature fails to display the function body,
as per the man page.
Resolution:
ksh(1) man page is corrected to document the proper use of
"typeset -f" functionality.
( SR:8606442247 CR:JAGag00047 )
While executing a list of commands in ksh(1) subshell, if
any one of the built-in command fails in the list, the shell
exits thereby not executing the subsequent commands in the
list.
Resolution:
ksh(1) will now execute all the commands in a list
irrespective of the errors.
( SR:8606462549 CR:JAGag18411 )
There are instances wherein an uninitialized or empty local
variable is used while calculating the arithmetic
expression. Due to this, wrong values are returned from the
arithmetic expressions.
Resolution:
ksh(1) has been modified to return the correct results.
PHCO_32856:
( SR:8606388635 CR:JAGaf48784 )
When an interactive ksh imports a TMOUT value
from its parent and its value is changed, the new
value is not reflected in the shells behavior. It
will still continue to exhibit the behavior of
TMOUT it imported from the parent.
The problem can be reproduced as follows :
$ /usr/bin/ksh
$ export TMOUT=10
$ /usr/bin/ksh
$ export TMOUT=0
$ <Keep the shell inactive>
shell time out in 60 seconds
/usr/bin/ksh: timed out waiting for input
The TMOUT variable indicates the timeout value for
the shell. A value greater than 0 in TMOUT,
terminates the shell, if no command is issued within
the prescribed number of seconds after the PS1 prompt
is printed.
The default value for TMOUT is 0, which allows the
shell to be inactive forever.
Here, although TMOUT is set to 0 in the sub-shell,
the shell exits if it remains in-active for 10 seconds.
Resolution:
Now ksh will exit with timeout based on the recent value
set to TMOUT and not on the value it inherited from
parent.
PHCO_31552:
( SR:8606350351 CR:JAGaf11171 )
Large number of background jobs running on an
interactive ksh shell terminates it.
The problem can be reproduced as follows:
$/usr/bin/ksh
$set -o vi
$for i in `perl -e '@a=(1..50); print "@a"'`
>do
>sleep 15 &
>done.
$
Now wait for around 15 to 20 seconds without entering
any input on the standard input. After 15 or 20
seconds, the ksh shell will be terminated automatically.
Root Cause:
When the ksh shell fails to read from the standard
input due to the interruption by SIGCHLD signals
generated after the termination of each background
job, it tries to read again for a fixed number of
times. The shell exits if it doesn't succeed
within the fixed number of tries.
Resolution:
Now the code is modified so that the shell will
always try to read again from the standard input
if it fails to read due to the SIGCHLD interruption.
( SR:8606350366 CR:JAGaf11186 )
The ksh shell prints line number instead of signal number
when a signal is sent to it.
The problem can be reproduced as follows :
1. Create a test script - test.sh
$ cat test.sh
#!/usr/bin/ksh -p
# Lots of comments so
# that the lines below
# are after 10 just so
# the affect of the bug
# is easily seen.
# 8
# 9
# 10
print "mypid=$$"
ksh -c 'kill -9 $$'
ksh -c 'kill -9 $$'
ksh -c 'kill -9 $$'
ksh -c 'kill -9 $$'
2. Run the script.
$ ./test.sh
3. The output will be :
mypid=25632
test.sh[12]: 25612 Killed
test.sh[13]: 25613 Killed
test.sh[14]: 25614 Killed
test.sh[15]: 25615 Killed
This was due to the fact that, the shell would overwrite
part of pid with the line number.
Resolution:
Now the code is modified so that the shell does not
overwrite the pid value, and hence the correct value is
printed as output. Now, when the above mentioned test
script is run, the correct output is displayed :
mypid=25632
test.sh[12]: 25633 Killed
test.sh[13]: 25634 Killed
test.sh[14]: 25635 Killed
test.sh[15]: 25636 Killed
( SR:8606369179 CR:JAGaf29723 )
In ksh shell, "set -A" is used to assign an
array variable. Before assigning the new values to
the array variable, shell frees the memory
associated with the existing array elements (if any)
and allocates the memory for the new array elements.
The code for freeing the memory of the existing array
elements was returning without freeing the memory
of the last element in the array and hence
causing the memory leak.
The problem can be reproduced as follows:
1.
$cat leak.sh
#!/usr/bin/ksh
typeset -i N=0
while [[ $N -le $1 ]];do
set -A arr 1 2
(( N = N + 1 ))
done
UNIX95=1 ps -osz,comm -p$$; echo $1; exit
$
The above script will print the size of the shell.
2. Run the above script with the argument let's say
1 and 50000 and compare the size of the shell. E.g.
$/usr/bin/ksh leak.sh 1
SZ COMMAND
55 sh.org
1
$/usr/bin/ksh leak.sh 50000
SZ COMMAND
402 sh.org
50000
$
The 50000 iterations caused the shell to grow by
347*4096 bytes.
Resolution:
Now the code for freeing the memory of the
existing array elements is corrected such
that it frees even the memory of the last
array element.
( SR:8606369780 CR:JAGaf30323 )
The problem can be easily reproduced as follows :
$ cat call.sh
#invoke first child
./wait_and_exit 3 &
pid_1cld=$!
#invoke second child
./wait_and_exit 2 &
pid_2cld=$!
wait $pid_1cld
echo $?
wait $pid_2cld
echo $?
$ cat wait_and_exit.c
#include <stdlib.h>
#include <unistd.h>
main(argc, argv)
int argc;
char **argv;
{
int i;
i=atoi(*++argv);
sleep(i);
exit(i);
}
$ cc wait_and_exit.c -o wait_and_exit
Run call.sh
$ ./call.sh
3
127 <---- This is expected to be 2.
ksh(1) does not maintain an exitlist for already
completed processes. Hence, when wait is called
on a completed process, it assumes that the given
process id is invalid & hence returns 127.
Resolution:
Now an exitlist is maintained to remember the exit
values of the background processes. Hence, the correct
exit values will be returned.
Enhancement:
No
SR:
8606420313 8606435240 8606442247 8606462549 8606388635
8606350351 8606350366 8606369179 8606369780
Patch Files:
OS-Core.CORE-ENG-A-MAN,fr=B.11.23,fa=HP-UX_B.11.23_IA/PA,
v=HP:
/usr/share/man/man1.Z/ksh.1
/usr/share/man/man1.Z/rksh.1
OS-Core.UX2-CORE,fr=B.11.23,fa=HP-UX_B.11.23_IA,v=HP:
/usr/bin/ksh
/usr/bin/rksh
OS-Core.UX2-CORE,fr=B.11.23,fa=HP-UX_B.11.23_PA,v=HP:
/usr/bin/ksh
/usr/bin/rksh
what(1) Output:
OS-Core.CORE-ENG-A-MAN,fr=B.11.23,fa=HP-UX_B.11.23_IA/PA,
v=HP:
/usr/share/man/man1.Z/ksh.1:
None
/usr/share/man/man1.Z/rksh.1:
None
OS-Core.UX2-CORE,fr=B.11.23,fa=HP-UX_B.11.23_IA,v=HP:
/usr/bin/ksh:
$ B.11.23 Sep 17 2004 18:28:29 $
Version 11/16/88
error.c $Date: 2007/03/27 12:05:26 $Revision: r11.23
/1 PATCH_11.23 (PHCO_34510)
jobs.c $Date: 2005/09/02 00:07:48 $Revision: r11.23/
2 PATCH_11.23 (PHCO_32856)
adjust.c $Date: 2005/02/24 01:39:17 $Revision: r11.2
3/1 PATCH_11.23 (PHCO_32856)
streval.c $Date: 2007/03/27 12:07:12 $Revision: r11.
23/1 PATCH_11.23 (PHCO_34510)
unassign.c $Date: 2006/02/15 04:05:08 $Revision: r11
.23/2 PATCH_11.23 (PHCO_34510)
$Revision: @(#) ksh88 R11.23_BL2007_0327_2 PATCH_11.
23 PHCO_34510
/usr/bin/rksh:
$ B.11.23 Sep 17 2004 18:28:29 $
Version 11/16/88
error.c $Date: 2007/03/27 12:05:26 $Revision: r11.23
/1 PATCH_11.23 (PHCO_34510)
jobs.c $Date: 2005/09/02 00:07:48 $Revision: r11.23/
2 PATCH_11.23 (PHCO_32856)
adjust.c $Date: 2005/02/24 01:39:17 $Revision: r11.2
3/1 PATCH_11.23 (PHCO_32856)
streval.c $Date: 2007/03/27 12:07:12 $Revision: r11.
23/1 PATCH_11.23 (PHCO_34510)
unassign.c $Date: 2006/02/15 04:05:08 $Revision: r11
.23/2 PATCH_11.23 (PHCO_34510)
$Revision: @(#) ksh88 R11.23_BL2007_0327_2 PATCH_11.
23 PHCO_34510
OS-Core.UX2-CORE,fr=B.11.23,fa=HP-UX_B.11.23_PA,v=HP:
/usr/bin/ksh:
$Revision: 92453-07 linker linker crt0.o B.11.16.01
030415 $
error.c $Date: 2007/03/27 12:05:26 $Revision: r11.23
/1 PATCH_11.23 (PHCO_34510)
jobs.c $Date: 2005/09/02 00:07:48 $Revision: r11.23/
2 PATCH_11.23 (PHCO_32856)
$Revision: @(#) ksh88 R11.23_BL2007_0327_2 PATCH_11.
23 PHCO_34510
adjust.c $Date: 2005/02/24 01:39:17 $Revision: r11.2
3/1 PATCH_11.23 (PHCO_32856)
streval.c $Date: 2007/03/27 12:07:12 $Revision: r11.
23/1 PATCH_11.23 (PHCO_34510)
unassign.c $Date: 2006/02/15 04:05:08 $Revision: r11
.23/2 PATCH_11.23 (PHCO_34510)
$ B.11.23 Dec 2 2004 03:46:59 $
Version 11/16/88
/usr/bin/rksh:
$Revision: 92453-07 linker linker crt0.o B.11.16.01
030415 $
error.c $Date: 2007/03/27 12:05:26 $Revision: r11.23
/1 PATCH_11.23 (PHCO_34510)
jobs.c $Date: 2005/09/02 00:07:48 $Revision: r11.23/
2 PATCH_11.23 (PHCO_32856)
$Revision: @(#) ksh88 R11.23_BL2007_0327_2 PATCH_11.
23 PHCO_34510
adjust.c $Date: 2005/02/24 01:39:17 $Revision: r11.2
3/1 PATCH_11.23 (PHCO_32856)
streval.c $Date: 2007/03/27 12:07:12 $Revision: r11.
23/1 PATCH_11.23 (PHCO_34510)
unassign.c $Date: 2006/02/15 04:05:08 $Revision: r11
.23/2 PATCH_11.23 (PHCO_34510)
$ B.11.23 Dec 2 2004 03:46:59 $
Version 11/16/88
cksum(1) Output:
OS-Core.CORE-ENG-A-MAN,fr=B.11.23,fa=HP-UX_B.11.23_IA/PA,
v=HP:
1436246302 42408 /usr/share/man/man1.Z/ksh.1
1436246302 42408 /usr/share/man/man1.Z/rksh.1
OS-Core.UX2-CORE,fr=B.11.23,fa=HP-UX_B.11.23_IA,v=HP:
1281966804 447560 /usr/bin/ksh
1281966804 447560 /usr/bin/rksh
OS-Core.UX2-CORE,fr=B.11.23,fa=HP-UX_B.11.23_PA,v=HP:
3510868961 237568 /usr/bin/ksh
3510868961 237568 /usr/bin/rksh
Patch Conflicts: None
Patch Dependencies: None
Hardware Dependencies: None
Other Dependencies: None
Supersedes:
PHCO_31552 PHCO_32856
Equivalent Patches: None
Patch Package Size: 400 KBytes
Installation Instructions:
Please review all instructions and the Hewlett-Packard
SupportLine User Guide or your Hewlett-Packard support terms
and conditions for precautions, scope of license,
restrictions, and, limitation of liability and warranties,
before installing this patch.
------------------------------------------------------------
1. Back up your system before installing a patch.
2. Login as root.
3. Copy the patch to the /tmp directory.
4. Move to the /tmp directory and unshar the patch:
cd /tmp
sh PHCO_34510
5. Run swinstall to install the patch:
swinstall -x autoreboot=true -x patch_match_target=true \
-s /tmp/PHCO_34510.depot
By default swinstall will archive the original software in
/var/adm/sw/save/PHCO_34510. If you do not wish to retain a
copy of the original software, include the patch_save_files
option in the swinstall command above:
-x patch_save_files=false
WARNING: If patch_save_files is false when a patch is installed,
the patch cannot be deinstalled. Please be careful
when using this feature.
For future reference, the contents of the PHCO_34510.text file is
available in the product readme:
swlist -l product -a readme -d @ /tmp/PHCO_34510.depot
To put this patch on a magnetic tape and install from the
tape drive, use the command:
dd if=/tmp/PHCO_34510.depot of=/dev/rmt/0m bs=2k
Special Installation Instructions: None
|
