OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
Subject: Re: New_AIXV4_Fixes
From: AIX Service Mail Server (aixservaustin.ibm.com)
Date: Thu Dec 14 2000 - 02:14:44 CST

APAR: IY11543 COMPID: 5765D5100 REL: 311
ABSTRACT: FS_DAEMON_BCAST_CMD() BUG CAUSES NODES TO ALWAYS SYNC SWITCH TOD

PROBLEM DESCRIPTION:
fs_daemon_bcast_cmd(KLOAD_ROUTES) does not correctly manage the
pkt field "sync_tod", causing some nodes (specifically, those
following a switch router - if one - in the list of active
nodes) to sync their TBIC TODS on every KLOAD_ROUTES. An
unavoidable race condition exists when >1 node on a chip syncs
it's TOD: if a switch chip receives >1 send-set-TOD cmd at
one time, a service CRC can result (typically, but not always,
recovered by the primary). Limiting the sync TOD operation to
required times greatly limits this exposure.

PROBLEM SUMMARY:
The code change eliminates unnecessary sync-TOD operations,
thus greatly reducing the number of send-set-TOD collisions
on leaf chips. (These collisions drive fault recovery on
the primary due to a service error from the chip.)

PROBLEM CONCLUSION:
The code has been fixed to set the sync_tod field for each
packet, because the pkt storage is reused.

------

APAR: IY12286 COMPID: 5765D6100 REL: 210
ABSTRACT: DEFAULT LOADLEVELER SCHEDULER DID NOT USE ADAPTER IN

PROBLEM DESCRIPTION:
using the default loadleveler scheduler, the Adapter variable
of the requirements stanza does not correctly generate use
of the desired adapter. Upon job submission, a message is
generated stating that the adapter (switch for example)
requested has been used for the job. However, this really
does not happen and the adapter used is actually the default
adapter based on the machine name that is used in
LoadL_admin on the machine stanza.

LOCAL FIX:
use the backfill scheduler or
set the desired interface to the default as
label in the machine-stanza.

PROBLEM SUMMARY:
When LoadLeveler is set to the default scheduler,
IP jobs are not running on the switch.

PROBLEM CONCLUSION:
When LoadLeveler is set to the default scheduler,
IP jobs are now running on the switch.

------

APAR: IY12578 COMPID: 5765D5100 REL: 311
ABSTRACT: SYSCTLD CAUSED SIGSEGV IN STRCPY.STRCPY() AT MIDNIGHT.

PROBLEM DESCRIPTION:
sysctld sometime caused segmentation vioration in strcpy.strcpy
and dumped core at midnight when cleanup.logs.nodes running.
Failing function stack in the core was as follow.
 strcpy.strcpy() at 0x100080cc
 Tcl_SetVar2(??, ??, ??, ??, ??) at 0x1000fa58
 Tcl_SetVar(??, ??, ??, ??) at 0x1000fc28
 Tcl_SetCmd(??, ??, ??, ??) at 0x1000e1e4
 Tcl_Eval(??, ??) at 0x1000c320
 Tcl_IfCmd(??, ??, ??, ??) at 0x10016e10
 Tcl_Eval(??, ??) at 0x1000c320
 Tcl_WhileCmd(??, ??, ??, ??) at 0x1001c85c
 Tcl_Eval(??, ??) at 0x1000c320
 Tcl_IfCmd(??, ??, ??, ??) at 0x10016e10
 Tcl_Eval(??, ??) at 0x1000c320
 Tcl_ForeachCmd(??, ??, ??, ??) at 0x10011dd0
 Tcl_Eval(??, ??) at 0x1000c320
 InterpProc(??, ??, ??, ??) at 0x10010244
 Tcl_Eval(??, ??) at 0x1000c320
 svc_tcl_eval(??) at 0x10031980
 sysctl_svc(??) at 0x100306a8
 svc_dispatch(??) at 0x10000b34
 sysctld.svc_run() at 0x10000c78
 main(??, ??) at 0x100002f0

PROBLEM SUMMARY:
When large logs are being trimmed by either
cleanup.logs.nodes or cleanup.logs.ws, it is possible for
sysctld to core dump. The core dumps are intermittent and
are caused by the access of memory that was not allocated
successfully.

PROBLEM CONCLUSION:
logmgt.cmds was modified to use physical storage, instead
of memory when trimming log files. Checks were also added
to only proceed with the trimming of the log files if there
is enough space in the file system. As a result of these
changes sysctld will not need to allocate memory to hold
the trimmed log file and the success of the allocation of
any memory will be verified prior to accessing it.

------

APAR: IY12592 COMPID: 5765E5100 REL: 600
ABSTRACT: SMIT STILL DOES NOT WORK FOR BLANK MODES.

PROBLEM DESCRIPTION:
With an existing CPI-C side info definition, I can not
remove the mode name using SMIT. I can change the mode
name, but can not remove it completely.

PROBLEM SUMMARY:
With an existing CPI-C side info definition, I can not
 remove the mode name using SMIT. I can change the mode
 name, but can not remove it completely.

PROBLEM CONCLUSION:
We fixed SMIT so that you can remove the mode name from a CPI-C
side info definition as expected.

------

APAR: IY12762 COMPID: 5765D5100 REL: 311
ABSTRACT: SUPPER FAILS WHEN THERE IS A IP ALIAS ON EN0

PROBLEM DESCRIPTION:
If there is an IP alias of the en0 on the backup CWS, then any
that is customized while the backup CWSis active will use this
alias addressas the supper server (/etc/ssp/server_name).
This will cause supperupdates to fail ("server is busy") when
the primary CWS becomes theactive CWS again. The likelyculprit
is pssp_script - some logic needsto be added so he picks
right cws IP addr for the supper server(SDR SP cw_ipaddr
object) when there are aliases present.

PROBLEM SUMMARY:
In an hacws environment where the backup CWS contains
an IP alias for the en0 interface pssp_script, during
node customization, creates an incorrect
/etc/ssp/server_name file. The problem only occurs
if the node is customized while the backup CWS is
active. The symptom is a 'server busy' informational
message when '/var/sysman/supper update' is issued from
the node and the primary CWS is the active CWS again.

PROBLEM CONCLUSION:
The new code changes the 'cw_ipaddrs' identifier in the
SDRGetObjects call to 'control_workstation' (the new code
can be found in the 'The BIS is the control workstation'
section under the customization path). The modified
SDRGetObjects call eliminates a need for a while loop since
a list is no longer generated with the CWS's IP addresses.
The variable cws_adap_addrs now has a single value with
which the new code does a 'host' call against to retrieve
the primary CWS's IP address (which then gets entered into
the node's /etc/ssp/server_name file).
Prior to the code fix, pssp_script chose the very first IP
address from the list of addresses for the CWS. In the case
where the backup CWS has an IP alias, the first choice would
be an incorrect one ... the backup CWS IP alias entry in
the node's server_name file is what causes supper to "fail"
with a 'server busy' message if issued after customization
and when the primary CWS is the active CWS again.

------

APAR: IY12779 COMPID: 5765D5100 REL: 311
ABSTRACT: SIGNAL HANDLER SHOULD BE ALLOWED TO BE PLACED PRIOR TO MPI_INIT.

PROBLEM DESCRIPTION:
In PPE.POE 2.4 there currently exists a restriction to place any
user supplied signal handler after the MPI_Init() statement
in an MPI program. This restriction should be removed to the
extent possible. An Efix has already been tried that appears to
have addressed this problem.

PROBLEM SUMMARY:
There currently exists a restriction to place any
user supplied signal handler after the MPI_Init() statement
in an MPI program. This restriction should be removed to
the extent possible.

PROBLEM CONCLUSION:
A change was made so that the SIGIO signal from a user
supplied signal handler could be placed before the
MPI_Init call.

------

APAR: IY12846 COMPID: 5765D5101 REL: 111
ABSTRACT: HATS REPORTS CSS0 ADAPTER UP PREMATURELY AFTER REBOOT.

PROBLEM DESCRIPTION:
Problem arises at reboot when hats is initializing. hats
concludes that since the css0 adapter is configured that it is
available and attempts to start broadcasting. While the css0
adapter is configured it is not necessarily up. hats reports
to hags and the css hags groups are formed. Then, seconds later
hats reports the adapter down. This causes problems with the
subsystems that sunscribe to the css hags group(s) (rvsd and
gpfs).

PROBLEM SUMMARY:
When a node is rebooted, the Topology Services and
the Switch subsystems are initialized roughly at the
same time.
At its startup, Topology Services will initially
assume that all adapters that are configured
at their expected addresses are up. After a few
seconds, adapters that are not really up will be
considered down by Topology Services. This behavior
may result in the following scenario:
1) The Switch subsystem is initializing the switch
  adapter when Topology Services starts. The adapter
  is at the correct address but IP is still not
  working for the adapter.
2) Topology Services sees that the adapter is at the
  expected address and issues a "singleton
  adapter notification" to Group Services, which is
  propagated to VSD and GPFS.
3) The adapter is not really up, at least for IP, so
  Topology Services will declare the adapter down
  after approximately 20 seconds. When the notification
  reaches VSD and GPFS, both subsystems exit on the
  node, forcing their peers in other nodes
  to go into recovery.
4) A few seconds later the switch adapter is enabled
  for IP, and Topology Services declares the adapter
  up.
The quick "up-down-up" transition causes problems for
GPFS and VSD, and should be eliminated.

PROBLEM CONCLUSION:
With the fix, if the adapter interface flags indicate
"down" for the switch adapter, then Topology Services
will not declare the adapter up on startup. Later on,
when the adapter is finally enabled for IP, then
Topology Services will declare the adapter up.
With the fix, no quick "up-down-up" transition will
occur when the switch is being initialized at the
time the Topology Services subsystem is started.

------

APAR: IY12888 COMPID: 5765D6600 REL: 201
ABSTRACT: CUMULATIVE MAINT #01 FOR MQSERIES INTEGRATOR FOR AIX V201

PROBLEM DESCRIPTION:
CUMULATIVE MAINT #01 FOR MQSERIES INTEGRATOR FOR AIX V201

PROBLEM CONCLUSION:
Cumulative maintenance #01 contains fixes to the following
problems:-
IC27806 - DATA FLOW ENGINE WILL NOT RECOVER FROM ERRORS DURING D
IC27959 - DELETING A DATABASE REFERENCE FROM A COMPUTE NODE DOES
IC28174 - COMPOUND NODE OF A FLOW CONTAINING LABELS CANNOT BE IN
IC28401 - BIP2211E ERROR OCCURRED BY USING BYUSERID IN ORDER MOD
IY12651 - DATA FLOW ENGINE WILL NOT RECOVER FROM ERRORS DURING D
IY13621 - MQSI MRM MESSAGE IS CAUSING THE BROKER TO SPIN
f8167 - Properties windows always resizes back to default
f13255 - Properties windows divider always gets back to default
 9049 - MRMB31 Memory leak associated with datasource keyword.
11318 - Implement ICU 1.6 for enhanced EUC support
12106 - MRMB0408 Boolean tests fail
12191 - mqsistop causes DFE core in ostream_withassign (Neon?
12201 - Messages without inserts get incorrect processing of q
12331 - Usage message wrong for mqsireport
12437 - Core dump when ODBC trace active
12507 - Generate fails for Gold Messageset.
12642 - Broker spins when Create ExecutionGroup does not respo
12678 - 1812 Generated when prod. Msg Bk for Gold Message set
12698 - sample directory not removed on uninstall
12772 - Cannot delete Brokers on Uninstall
12789 - Data flow engine will not recover from errors during d
12805 - executionGroupTraceOverrideLevel broken
12848 - Apparent memory leak in MTI
12889 - MRMB3503 Dataflowengine abnormally terminated event216
12917 - NT: DB2: deploy invalid config msg -> broker terminate
12922 - mqsistop usernameserver gives BIP2015
12966 - Scribble IVP fails if no '.' in classpath
12975 - Data base update causes a memory leak on AIX using Ora
12978 - Problem with sequencing of actions in the Tour
12989 - The CWF property of a string can only be set to a fixe
12997 - DataFlowEngine locks on deploy with Sybase
13028 - MQRFH2 parser rejects valid RFH2 from AMI
13038 - DataFlowEngine locks on deploy with Sybase
13048 - Memory leak in ImbDecimal class
13121 - cniElementDecimalValue returning unexpected result.
13135 - XML white space on tree copies to NameVal parsers
13137 - cpiCreateContext() called for each message parsed
13149 - Element tree detached unexpectedly in plugin parser
13185 - Pub Sub is never going to work when access control is
13187 - COBOL / C importer name clash (PMR 36770,278,000)
13188 - Reference to removed object in compute node mapping
13221 - Allow threads in a DataFlowEngine to run concurrently
13261 - Button labels doesn't change but the text indicates th
13272 - An error message issued when using byUserID.
13300 - ODBCINI must not be set on login accounts
13356 - Valid mapping lines are removed when deleting an input
13389 - 'pow' function gives SIGSEGV on AIX
13425 - Large columns >2000 bytes corrupted on retrieval

------

APAR: IY12920 COMPID: 5765D5100 REL: 311
ABSTRACT: PCP FAILS FOR FILES > 2 GIG

PROBLEM DESCRIPTION:
The pcp script fails when files greater than 2 gig are specified
This is due to using the perl '-r' function with the pssp level
of perl which does not support files > 2 gig.

PROBLEM SUMMARY:
The PERL which we ship does 32bit addressing which
prevents processing files larger than 2 GB in size. As
a result pcp of a file larger than 2 GB would fail when
checking the existence of the file with the -r flag with
the message: Invalid file specified

PROBLEM CONCLUSION:
Modified pcp to use the system ls command instead of the
-r flag to determine the existence of a file. This allows
files larger than 2 GB to be processed.

------

APAR: IY12961 COMPID: 5765D6100 REL: 210
ABSTRACT: UNIX_GROUP NAME IS NOT BEING ACKNOWLEDGED.

PROBLEM DESCRIPTION:
If "Unix_group" is specified as the user's default group
and the include_groups lines within the class stanza
contains the user's unix group name, the unix group name is
not being acknowledged thus preventing the job from being
submitted.
Sample
 Yoshida: type = user
          default_group = UNIX_GROUP
 large: type = class
        include_groups = staff
Error
 llsubmit: 2512-080 Class "large" is not valid for group "Unix_G
           roup".
 llsubmit: 2512-051 This job has not been submitted to LoadLevel
           er.

PROBLEM SUMMARY:
If "Unix_Group" is specified as the user's default group
and the include_groups lines within the class stanza
contains the user's unix group name, the unix group name is
not being acknowledged thus preventing the job from being
submitted.

PROBLEM CONCLUSION:
The llsubmit command has been changed so that when the unix
group name is listed on either the include_groups or
exclude_groups line within the class stanza, it will be
acknowledged when the user's default group is specified as
"Unix_Group".

------

APAR: IY12989 COMPID: 5765C3403 REL: 430
ABSTRACT: X25PKT CRASHES SYSTEM

PROBLEM DESCRIPTION:
x25pkt issues a bufcall(), but then frees the
queue, thus making the bufcall_rsrv() crash
the system.

LOCAL FIX:
This happens when bosdebug is turned on.
Turning off bosdebug should work around the problem.

PROBLEM SUMMARY:
During the open() processing, x25pkt issues a bufcall(),
and then causes the queue to be freed, thus making the
bufcall_rsrv() access freed memory, and crash the system.

PROBLEM CONCLUSION:
make sure bufcall() is not called during pktopen() processing.

------

APAR: IY13040 COMPID: 5765D5101 REL: 111
ABSTRACT: HAGS DAEMON SHOULD NOT ASSERT IF IT DETECTS

PROBLEM DESCRIPTION:
The hags daemon asserts when it detects an inconsistant
instance number. The following error is seen in the hags log:
Failed assertion: !"No incarnation hash entry on delete."

PROBLEM SUMMARY:
When a switch group leader left and and rejoined the group,
it resulted in an inconsistent instance number. As a
result, the group services daemon asserted with the
following error message in the hags log:
'Failed assertion: !"No incarnation hash entry on delete."'
The group services daemon should not assert under this
condition.

PROBLEM CONCLUSION:
The group services daemon has been modified to not
assert when it encounters an inconsistent instance
number.

------

APAR: IY13073 COMPID: 5765D5101 REL: 111
ABSTRACT: TOPSVCS/GRPSVCS LOGS CREATED WITH WORLD WRITABLE

PROBLEM DESCRIPTION:
Some HACMP generated files created with world writeable
permissions
   drwxr-xrwx /var/ha/run/ nmdiag.en1 nmdiag.0.1.0
   -rw-rw-rw- root system /var/ha/log/topsvcs.02.11.llunit
   -rw-rw-rw- /sbin/cluster/server.status "S7A llup_resource"

LOCAL FIX:
change file permissons to 600
no help for newly created logs

PROBLEM SUMMARY:
The topsvcs/grpsvcs logs had world-writable permissions.

PROBLEM CONCLUSION:
The topsvcs and grpsvcs logs now have a permission
of 644.

------

APAR: IY13170 COMPID: 5765D5100 REL: 311
ABSTRACT: VSD.FENCE* DOES NOT GREP CORRECTLY WHEN /VAR/ADM/CSD/VSDFILES

PROBLEM DESCRIPTION:
vsd.FENCE* does not correctly grep for the global volume group
in the VSD_GVG file in /var/adm/csd/vsdfiles. The problem
occurred when the customer had two gvg's that were similar up
to the last character. The egrep used captured both and the
last match (which is incorrect) causing the problem. The
function this occurred in was FlushIO. Using a "-w" should
alleviate the problem.

PROBLEM SUMMARY:
The problem arose because the customer had two
vsd global volume groups that were the same except
for the last character, like those below ...
gvg_vg1
gvg_vg17
The vsd.FENCE* script was executing the FlushIO
which does the following ...
- Determines which vsd's are fenced from
 the /var/adm/csd/vsdfiles/VSD_Fence file.
 (gets the vsd minor number).
- Using the minor number it the issues a lsvsd -l
 using egrep to get the vsd name and the server
 (this egrep is ok).
- Next, if this node is the server, it searches
 the /var/adm/csd/vsdfiles/VSD_Table for the
 vsd name gotten above. It finds the volume
 group that this vsd is associated with.
- Now using the volume group found above, it will
 look thru the /var/adm/csd/vsdfiles/VSD_GVG file
 lookng for who the current server is (should be
 either the primary or backup node number).
This is where the problem occurred. Since the
customer had GVG's that were similar (up to the
last character) the egrep was returning the second
and incorrect GVG for this query.
Technically this problem could arise on the egrep
thru the VSD_Table.

PROBLEM CONCLUSION:
The solution to this problem is simple, execute
the egrep is these two occurrances with the "-w"
option. This will ensure that the value being
searched on is the one returned.

------

APAR: IY13258 COMPID: 5765D5100 REL: 311
ABSTRACT: /VAR/ADM/CSD NOT MIRRORED AUTOMATICALLY ON A MIRRORED ROOTVG

PROBLEM DESCRIPTION:
Customer has a mirrored rootvg. When they installed the vsd
code and /var/adm/csd was created, it was not automatically
mirrored.

PROBLEM SUMMARY:
When installing the vsd file sets on a node with a mirrored
rootvg, the logical volume containing the new filesystem
/var/adm/csd is not automatically mirrored.

PROBLEM CONCLUSION:
During installation of the next PTF for vsd.vsdd, a check
will be made to determine if rootvg is being mirrored on
the node. If it is being mirrored, it will be determined
if the the logical volume containing /var/adm/csd has
the correct number of copies. If not, the mklvcopy command
will be issued to create the correct number of copies.

------

APAR: IY13336 COMPID: 5765D5100 REL: 311
ABSTRACT: THE SPMIRROR/SPUNMIRROR COMMANDS SHOULD NOT SET THE BOOTLIST

PROBLEM DESCRIPTION:
The spmirror/spunmirror commands set the bootlist use en0 first
and must be changed to set the bootlist to no longer use en0
first.

LOCAL FIX:
Change bootlist after executing spmirror/spunmirror commands
via bootlist command

PROBLEM SUMMARY:
When the rootvg of a node if mirrored, using the spmirrorvg
command, or umirrored using the spunmirrorvg command,
the normal bootlist of the node is modified to use the
ent0 first. ent0 should not be added to the normal
bootlist.

PROBLEM CONCLUSION:
Modified spmirror, which is called from spmirrorvg and
spunmirror, which is called from spunmirrorvg, so that
ent0 is not added to the normal bootlist of the node.

------

APAR: IY13351 COMPID: 5765E5100 REL: 601
ABSTRACT: CUMULATIVE APAR FOR 6.0.1.0 FOR CS/AIX

PROBLEM DESCRIPTION:
Cumulative apar for 6.0.1.0 for CS/AIX.

PROBLEM SUMMARY:
Cumulative apar for 6.0.1.0.for CS/AIX

------

APAR: IY13385 COMPID: 5765D5100 REL: 311
ABSTRACT: CREATEVSD2.PERL: 0021-039 MSGS NOT BEING DISPLAYED, CUSTOMER

PROBLEM DESCRIPTION:
Customer running createvsd which runs createvsd2.perl, however,
they never see the error because it's not getting displayed.
The exact error number is 0021-039. One of the reasons
is alot of the statements in this script have 1>&2 where it
should be 2>&1.
This may be defect 50590 in in Mohonk, but I'm not sure. At
any rate we need this fixed in 311.

PROBLEM SUMMARY:
The createvsd2.perl script had in a number of places error
messages being redirected to incorrect output (1>&2).
In this particular case the following message was not being
displayed to std out when the script failed....
defvsd: 0021-039 There was a bad lv_blk0_pbn value returned
from issuing rsh vsdxlate
All other occurances of the incorrect '1>&2' redirect
statement were also modified.

PROBLEM CONCLUSION:
The new code modifies all the incorrect re-direct statements
(incorrect: 1>&2 correct: 2>&1).
As the script is currently coded, when createvsd2.perl fails
no error messages are displayed to std out for the
customer's attention.

------

APAR: IY13386 COMPID: 5765D5100 REL: 311
ABSTRACT: SWITCH MICROCODE HAS NO SOURCE CONTROL INFO (NO 'WHAT' STRINGS)

PROBLEM DESCRIPTION:
Switch microcode has no source control info (no 'what' strings),
so the source file levels that the microcode is built on cannot
be determined.

PROBLEM SUMMARY:
The switch microcode binaries have no source control
information (no 'what' strings). The version number
cannot be determined from the object file and thus it
is difficult to determine what level of microcode the
customer has on his or her system.

PROBLEM CONCLUSION:
Source control information ('what strings') has been added
to the following microcode binaries:
  /usr/lpp/ssp/css/tb3ucode
  /usr/lpp/ssp/css/tb3ucode_DEBUG
  /usr/lpp/ssp/css/tb3dump
  /usr/lpp/ssp/css/tb3mx_ucode
  /usr/lpp/ssp/css/tb3mx_ucode_DEBUG
  /usr/lpp/ssp/css/tb3mxdump
  /usr/lpp/ssp/css/tb3pci_ucode
  /usr/lpp/ssp/css/tb3pci__ucode_DEBUG
  /usr/lpp/ssp/css/tb3pcidump

------

APAR: IY13580 COMPID: 569692600 REL: 110
ABSTRACT: FRAME LAYER DOES NOT RECOVER AFTER A BAD LINE CODITION

PROBLEM DESCRIPTION:
If an RR Frame with P-bit set is lost due to bad line
condition, the Frame Layer will never recover after that.

PROBLEM SUMMARY:
***************************************************************
*USERS AFFECTED: *
* All users with the following filesets at these levels *
* sx25.rte 1.1.5.10
* sx25.rte 1.1.5.11
* sx25.rte 1.1.5.12
* sx25.rte 1.1.5.7
* sx25.rte 1.1.5.8
* sx25.rte 1.1.5.9
***************************************************************
*PROBLEM DESCRIPTION: *
* If an RR Frame with P-bit set is lost due to bad line
* condition, the Frame Layer will never recover after
* that.
***************************************************************
*RECOMMENDATION: *
* Apply apar IY13580
***************************************************************

PROBLEM CONCLUSION:
Handle the possibility of losing an polled RR Frame.

------

APAR: IY13621 COMPID: 5765D6600 REL: 201
ABSTRACT: MQSI MRM MESSAGE IS CAUSING THE BROKER TO SPIN

PROBLEM DESCRIPTION:
The problem has been isolated to the call of a 'pow' function in
the DivideDouble routine in 'cwfread.cpp'. All instances of the
'pow' function in this file have been changed to a local power
function which just uses a loop to calcluate the power. The
core problem is the linkage of MQSI with the wrong maths library
which will be corrected.

PROBLEM SUMMARY:
The 'pow' function being issued on AIX is the
libbsd 'pow' which results in a SIGSEGV.

PROBLEM CONCLUSION:
This has been replaced with the libm 'pow'.

------

APAR: IY13701 COMPID: 5765E5100 REL: 600
ABSTRACT: CAN'T CANCEL CONVERSATION WHEN RECEIVE_ALLOCATE_EX IS WAITING

PROBLEM DESCRIPTION:
Can't cancel the conversation when a RECEIVE_ALLOCATE_EX is
outstanding for an APPC conversation for BEA systems.

PROBLEM SUMMARY:
Cannot cancel RECEIVE_ALLOCATE_EX APC verb.

PROBLEM CONCLUSION:
Added code to treat the same as RECEIVE_ALLOCATE.

------

APAR: IY13870 COMPID: 5765B9501 REL: 310
ABSTRACT: NODE CRASHWHEN EXPORT NFS FILESYSTEM WHEN WRITING LARGE

PROBLEM DESCRIPTION:
node crashes when writing a large file over nfs mounted GPFS
        filesystem.

PROBLEM SUMMARY:
GPFS panic of a node which was exporting a file system

PROBLEM CONCLUSION:
Correct an error in the creation of large files from NFS

------

APAR: IY13919 COMPID: 5765E5100 REL: 600
ABSTRACT: CREATE 2 SECOND DELAY FOLLOWING TERM-SELF AWAITING UNBIND BEFORE

PROBLEM DESCRIPTION:
I notice from the trace that the mainframe fails to send an
UNBIND after CS/AIX sends a TERM-SELF. 20 seconds later we
run into the problem. This may explain why we have not
seen this problem with other systems. CS/AIX V4 only
sends TERM-SELF before allowing the deallocate from hcon
to complete. hcon can therefore exit immediately, even if the
mainframe does not send an UNBIND. Early versions of CS/AIX V5
behaved the same.
However in order to correct another problem (APAR iy09365)
the behaviour of CS/AIX V5 was changed to wait for the UNBIND
in the deallocate code, CS/AIX V6 has this changed behaviour.

LOCAL FIX:
Dev't will alter the code to wait a short time for the UNBIND
(2 sec) before completing the deallocate.

------

APAR: IY13947 COMPID: 5765E5100 REL: 600
ABSTRACT: AN ERROR MESSAGE ' FIELD PORT_NUMBER WAS SPECIFIEDMORE THAN ONCE

PROBLEM DESCRIPTION:
An error message "Field port_number was specified more than once
" when perform an add port in SNA smit menu.
Communication server V6.0

PROBLEM SUMMARY:
Cannot define sdlc port from smit, gets error message of duplica
te port number.

PROBLEM CONCLUSION:
By mistake code for CS/UNIXWARE that added the port_number to th
e sdlc port dialog was release to CS/AIX V6. The port_number is
actually calculated automatically in the CS/AIX smit code so we
ended up sending two port_numbers to the snaadmin tool which cau
sed the error. Correct the code to remove the parameter from the
 smit dialog.

------

APAR: IY13990 COMPID: 5765B9501 REL: 310
ABSTRACT: MMCONFIG FAILS WITH SDR FILE NOT FOUND

PROBLEM DESCRIPTION:
mmconfig fails with SDR File Not Found

PROBLEM SUMMARY:
mmconfig would not operate

PROBLEM CONCLUSION:
remove unnecessary check for SDR files in
mmconfig

------

APAR: IY13998 COMPID: 5765D5100 REL: 311
ABSTRACT: SETUP_SERVER FAILS ON CWS INSTALLED WITH PSSP 3.2, IF IT IS

PROBLEM DESCRIPTION:
setup_server fails at at CWS installed with PSSP 3.2, if this
CWS is kerberos client of another CWS installed with PSSP 3.1.1
(or any other PSSP code below 3.2), because setup_server tries
to execute /usr/lpp/ssp/kerberos/etc/krunitrc at the kerberos
server. Since krunitrc is a new command in PSSP 3.2, it does
not exist at the kerberos server.

LOCAL FIX:
Copy /usr/lpp/ssp/kerberos/etc/krunitrc from PSSP 3.2 CWS to
the CWS, that is kerberos server.

PROBLEM SUMMARY:
When a PSSP 3.2 CWS is setup as a kerberos client of a
kerberos server that exists on a PSSP 2.4 or PSSP 3.1.1
system, setup_CWS will fail on the PSSP 3.2 CWS.
setup_CWS invokes /usr/lpp/ssp/kerberos/etc/krunitrc on
the kerberos server, but the script does not exist on
PSSP 2.4 and PSSP 3.1.1 systems. As a result the srvtabs
will not be able to be created on the PSSP 3.2 CWS.

PROBLEM CONCLUSION:
/usr/lpp/ssp/kerberos/etc/krunitrc will now be shipped
as part of PSSP 2.4 and PSSP 3.1.1. This will allow
a PSSP 3.2 CWS to be a kerberos client of a kerberos
server on a PSSP 2.4 or PSSP 3.1.1 system.

------

APAR: IY14034 COMPID: 5765D5100 REL: 311
ABSTRACT: CHANGE IN SIZE OF VSD HEADERS CAUSED PROBLEM HANDLING LARGE

PROBLEM DESCRIPTION:
change in size of vsd headers caused problem handling large
messages in microcode

PROBLEM SUMMARY:
VSD increased the size of its headers, which caused a
problem with how large messages were handed in the
microcode.

PROBLEM CONCLUSION:
The code was altered to handle larger headers.

------

APAR: IY14203 COMPID: 5765D5100 REL: 311
ABSTRACT: SETUP_SERVER FAILURE WITH RETURN CODE OF 2 IN SUB

PROBLEM DESCRIPTION:
Defect 68863
There is an added function called check_lppsource that
was added to PSSP through APAR IY12866 (ssp.basic.3.2.0.4
or later). This function checks the supported NIM
LPPSOURCES listed on the NIM master or CWS to make sure
there are appropriate filesets avaialable, permissions
set correct, etc for Node installations.
There is a problem with the current implemtation of
this function. It currently fails when there is an
lppsource_name defined by NIM that does not have the
directory setup on the CWS.
Sample:
/usr/sbin/lsnim -t lpp_source
lppsource_aix433 resources lpp_source
lpp_source_ssp resources lpp_source
lpp_source_ssp was not created by setup_server, it was
created by Admin running 'smitty nim' ...
setup_server then fails this
setup_server: 0016-624 The lppsource directory
/spdata/sys1/install/lpp_source_ssp does not exist on
the control workstation, exiting.
setup_server: 0016-478: Error(s) found while trying
to change file modes.
Tickets destroyed.
setup_server: Processing incomplete (rc= 2).
A workaround should be a change line 876 in setup_server
from
$return_code = $SEVERE;
to
$return_code = $ATTENTION;

PROBLEM SUMMARY:
***********************************************************
* USERS AFFECTED: Users installing the PSSP system with *
* ssp.basic at level 3.2.0.4, and who have removed *
* lpp_source NIM resources from their build tree. *
***********************************************************
* PROBLEM DESCRIPTION: *
*When a lppsource directory is defined by NIM, but doesn't*
*exist, setup_server issues the 0016-624 error message and*
*returns with return code 2 (abort). In practice, this *
*condition can happen when an install is done, and the *
*installer, to conserve space, has removed an old *
*lppsource tree. The abort is an overkill. Such *
*directories should be ignored and the installation *
*continued. *
***********************************************************
* RECOMMENDATION: *
* A corrected version of setup_server is provided in a *
* PTF release. *
***********************************************************

------

APAR: IY14236 COMPID: 5765B8100 REL: 220
ABSTRACT: DTBE APP CALLING STATE TABLE IN WHICH CALL TERMINATES CAUSES

PROBLEM DESCRIPTION:
If DTBE (java) application calls a State Table in which the call
terminates (e.g. TerminateCall action) and then returns to the
java environment, the java environment is not aware that the
call has terminated and continues to attempt to play segments.
This results in an Internal channel process error. DirectTalk
error_id=1.

PROBLEM SUMMARY:
DTBE APP CALLING STATE TABLE IN WHICH CALL
TERMINATES CAUSES INTERNAL CHP ERROR_ID=1

PROBLEM CONCLUSION:
By checking for hangup before attempting
to play

------

APAR: IY14244 COMPID: 5765B9501 REL: 310
ABSTRACT: RANDOM STORAGE CLOBBER IN FSCK

PROBLEM DESCRIPTION:
random storage clobber in fsck

PROBLEM SUMMARY:
mmfsck caused a segmentation violation in the

PROBLEM CONCLUSION:
Initialize a pointer

------

APAR: IY14305 COMPID: 5765B9501 REL: 310
ABSTRACT: FIX INIT AND ERROR HANDLING IN RECLOCK HANDLING

PROBLEM DESCRIPTION:
fix init and error handling in reclock handling

PROBLEM SUMMARY:
Correct error handling in record locking discovered in
a newer release

PROBLEM CONCLUSION:
Correct error handling in record locking discovered in a
future release.

------

APAR: IY14548 COMPID: 5765A8600 REL: 420
ABSTRACT: HACMP/HAES: CLLOCKD FAILS AFTER APPLYING AIX BOS.UP.4.3.3.25

PROBLEM DESCRIPTION:
cllockd fails to startup after applying AIX bos.up.4.3.3.25
and bos.mp.4.3.3.25.

PROBLEM CONCLUSION:
Changes are required in how hacmp cllockd code accesses the
kernel virtual memory heap used by the cllockd daemon process.

------

APAR: IY14587 COMPID: 569692600 REL: 110
ABSTRACT: APAR IY12590 IS BAD

PROBLEM DESCRIPTION:
comio applications and tcp/ip over X.25 do not work correctly
with multiple adapters (2 or more).

LOCAL FIX:
You must reject IY12590.

PROBLEM SUMMARY:
***************************************************************
*USERS AFFECTED: *
* All users with the following filesets at these levels *
* sx25.adt.rte 1.1.5.3
* sx25.adt.rte 1.1.5.4
* sx25.rte 1.1.5.11
* sx25.rte 1.1.5.12
* sx25.adt.npi 1.1.5.1
* sx25.adt.npi 1.1.5.2
* sx25.server 1.1.5.3
* sx25.server 1.1.5.4
* sx25.tcpip 1.1.5.1
* sx25.tcpip 1.1.5.2
* sx25.comio 1.1.5.7
* sx25.comio 1.1.5.8
* sx25.adt.comio 1.1.5.1
* sx25.adt.comio 1.1.5.2
***************************************************************
*PROBLEM DESCRIPTION: *
* There is not a communication with the second adapter
* in the same machine.
***************************************************************
*RECOMMENDATION: *
* Apply apar IY14587
***************************************************************

PROBLEM CONCLUSION:
We are backing out the bad fix for apar IY12590.

------

APAR: IY14699 COMPID: 5765B8100 REL: 220
ABSTRACT: STE CORE WHEN INVOKESTATETABLE ON BINARYONLY ST.

PROBLEM DESCRIPTION:
When STE editor is used with InvokeStateTable on Binary only
state table, STE will core sometimes.

PROBLEM SUMMARY:
STE CORE WHEN invokes statetable on
binaryonly statetable

PROBLEM CONCLUSION:
Corrected code calls to stop problem

------

APAR: IY14739 COMPID: 5765C3403 REL: 430
ABSTRACT: JDK 1.1.8 PTF 10: A118-20001116

PROBLEM DESCRIPTION:
Fixes since 1.1.8 PTF 9 (a118-2000914):
+--------+------+-------------------------------------------+
|Build |Defect|Description |
+--------+------+-------------------------------------------+
|20000926|12506 |StackTrace from RemoteException.detail |
+--------+------+-------------------------------------------+
|20001004|12878 |Inconsistent byte array proc with JIT on |
+--------+------+-------------------------------------------+
|20001005|12899 |AWT Crash in JA_JP Locale |
+--------+------+-------------------------------------------+
|20001013|12885 |wait/notify acts as stack, not queue |
+--------+------+-------------------------------------------+
|20001101|12922 |JIT corrupts BigDecimal -> JDBC |
+--------+------+-------------------------------------------+
|20001103|12778 |SIGSEGV in libawt.a |
+--------+------+-------------------------------------------+
|20001103|12851 |NullPointerException when using Field.get()|
+--------+------+-------------------------------------------+
|20001103|12924 |Sync() call in GC causes performance hit |
+--------+------+-------------------------------------------+
|20001108|12932 |NullPointer exception only with JIT ON |
+--------+------+-------------------------------------------+

PROBLEM CONCLUSION:
All of the above defects have been fixed.

------

APAR: IY14861 COMPID: 5648C9802 REL: 430
ABSTRACT: SDK 1.3.0 PTF 3 : CA130-20001124

PROBLEM DESCRIPTION:
Fixes since PTF 2 (ca130-20001025) :
(Note: The descriptions here have been truncated.)
+--------+------+---------------------------------------------+
|Build |Defect|Description |
+--------+------+---------------------------------------------+
|20001110|10770 |JColorChooser scrollbar-HSB problem |
+--------+------+---------------------------------------------+
|20001110|24650 |Win32 BasicTableUI editCellAt(int, int) |
+--------+------+---------------------------------------------+
|20001110|25739 |JTextArea does not wrap |
+--------+------+---------------------------------------------+
|20001111|26287 |WebSM: Hang in waitForID |
+--------+------+---------------------------------------------+
|20001115|25558 |Large File Support is incomplete |
+--------+------+---------------------------------------------+
|20001115|25661 |Ctrl+H destroys highlighted text |
+--------+------+---------------------------------------------+
|20001115|25924 |Java IM window doesn't come up in PTF1 |
+--------+------+---------------------------------------------+
|20001115|8239 |Usability problem in renaming a folder in JFi|
+--------+------+---------------------------------------------+
|20001116|24456 |Roman numerals aren't displayed correctly |
+--------+------+---------------------------------------------+
|20001116|25632 |Cache coherency problem on PowerPC |
+--------+------+---------------------------------------------+
|20001116|26004 |Second fix for double-test idiom in clconstan|
+--------+------+---------------------------------------------+
|20001117|26037 |Notify() being lost - race condition? |
+--------+------+---------------------------------------------+
|20001118|26379 |Modal Dialog appears more than once |
+--------+------+---------------------------------------------+
|20001118|26444 |Popup menus displayed off screeen |
+--------+------+---------------------------------------------+
|20001121|26395 |Compiler scoping problem with inner classes |
+--------+------+---------------------------------------------+
|20001124|24494 |JTabbedPane focusGained misfires |
+--------+------+---------------------------------------------+
|20001124|26354 |JNDI plugin only sets Sun ORB initialisation |
+--------+------+---------------------------------------------+

------

APAR: IY14990 COMPID: 5765B8100 REL: 220
ABSTRACT: FIX SMALL TIMING WINDOW IN FREE_PIGGYBACK DIRECTTALK DRIVER

PROBLEM DESCRIPTION:
Fix small timing window in free_piggyback directtalk driver

PROBLEM SUMMARY:
fix small timing window in free_piggyback

PROBLEM CONCLUSION:
Restructing of code to move the WAKEUP
macro so that it is protected by the locks

------

APAR: IY15024 COMPID: 576527300 REL: 121
ABSTRACT: PTF FOR IY10580 AND IY14999

PROBLEM DESCRIPTION:
IY10580 - new printers are forced to deveice=/dev/null
IY14999 - Percent (%) sign in job-name brings down spooler

PROBLEM SUMMARY:
IY10580 - device forced to /dev/null
iy14999 Percent (%) in job-name hung spooler

------

APAR: IY15066 COMPID: 5765D5100 REL: 311
ABSTRACT: LATEST PSSP 3.1.1 FIXES AS OF DECEMBER 7, 2000

PROBLEM DESCRIPTION:
This is the lastest PSSP ptf as of December 2000.
Order this apar to get all of the ptfs as of December 2000.

PROBLEM SUMMARY:
This is a packaging apar for PSSP 3.1.1 fixes
as of December 2000.

PROBLEM CONCLUSION:
This is a packaging apar for PSSP 3.1.1
fixes as of December 2000

------

APAR: JR15131 COMPID: 5765C7100 REL: 230
ABSTRACT: FRAME RELAY FRAMES NOT RECOGNIZED BY TERMINAL EQUIPMENT DUE

PROBLEM DESCRIPTION:
User/Terminal equipment is not able to decode Frame Relay frames
and/or LMI frames due to idle characters x"FF" being send by
2220 to fill up the interframe gap between trailing FR/HDLC flag
and leading FR/HDLC flag.
The Nways Switch Control Program V2R3.0 intermittently fails
to switch the idle character from "FF" to "7E".
Associated Engineering Defect is CMVC 42098.

PROBLEM SUMMARY:
Frames Relay/ATM interworking requires that the
port adapter (line scanner) computes the CRC for each frame, as
opposed to just check the frame CRC in case of FR end-to-end.
Switching CRC management mode, the adapter code sometimes failed
setting the "idle character" to "7E" instead of "FF".

------

APAR: PQ42876 COMPID: 5765C4100 REL: 212
ABSTRACT: BLACS MIN/MAX/SUM SUBROUTINES DO NOT HANDLE CDEST=-1 PROPERLY.

PROBLEM DESCRIPTION:
For Parallel ESSL BLACS codes xGAMX2D, xGAMN2D and xGSUM2D where
x is any precision(C,D,I,S,Z), CDEST=-1 causes an error to be
reported by MPI unless RDEST=-1.

PROBLEM SUMMARY:
The BLACS standard was incorrectly interpreted.
As a result, the MIN/MAX/SUM codes only left the results on all
processors when RDEST=-1. Results should be left on all
processors when RDEST=-1 or CDEST=-1. When CDEST=-1 and RDEST
was not -1, then an MPI error message was issued.

PROBLEM CONCLUSION:
All precisions of the BLACS MIN/MAX/SUM
codes were corrected to interpret RDEST=-1 or CDEST=-1
correctly.

TEMPORARY FIX:
User may code their call such that RDEST=-1 and
CDEST=-1.

------

APAR: PQ42877 COMPID: 5765C4100 REL: 212
ABSTRACT: PBLAS FAIL TO ISSUE REASON FOR TERMINATING WHEN INTERNAL STORAGE

PROBLEM DESCRIPTION:
Applications using Parallel ESSL may terminate without issuing
a 400 error when an internal allocation fails in the PBLAS.

PROBLEM SUMMARY:
Applications using Parallel ESSL may terminate
without issuing a 400 error message when an internal allocation
fails in a call to any PBLAS subroutine.

PROBLEM CONCLUSION:
PBUFFGET(and internal code) was corrected
to properly handle the allocation failure.

------

APAR: PQ42878 COMPID: 5765C4100 REL: 212
ABSTRACT: PARALLEL ESSL HEADER FILE UPDATE TO SUPPORT VISUAL AGE

PROBLEM DESCRIPTION:
Parallel ESSL header file, pessl.h, does not work with Visual
Age C and C++ V5 compilers.

PROBLEM SUMMARY:
Parallel ESSL header file does not work with
Visual Age C and C++ V5 compilers.

PROBLEM CONCLUSION:
Header file required updating. In the C
section, __ needed to become _. In the C++ section, the
cmplx class needed changes.

------