OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
From: AIX Service Mail Server (aixservaustin.ibm.com)
Date: Tue May 01 2001 - 02:19:04 CDT

  • Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

    APAR: IY15339 COMPID: 5765D5101 REL: 120
    ABSTRACT: HAGS BROADCAST METHOD CAN CAUSE STORMS ON INSTALLTIONS WITH

    PROBLEM DESCRIPTION:
     When HAGS needs a broadcast, it first tries to send the msg
    out to every body (burst broadcast). Rebroadcasting to the
    undelivered nodes will be performed in every 3 seconds. This may
    cause a big overhead if the number of nodes is large. This has
    been seen to cause hags voting issues.

    PROBLEM SUMMARY:
    Whenever Group Services needs a broadcast,
    it first tries to send the messages to
    all nodes and retries the broadcast
    the messages to the unresponded nodes
    in every 3 seconds.
    This behavior may increase the overload
    to the IP stack particularly if the
    number of nodes is large, and thus
    it may increase the message drop rate
    and cause more retries which may delay
    the Group Services' protocol completion.

    PROBLEM CONCLUSION:
    On a big system (with more than 64 nodes), this
    fix will lessen the overhead to broadcast
    Group Services messages by spreading out
    the message sends.

    ------

    APAR: IY15770 COMPID: 5765C3403 REL: 430
    ABSTRACT: MAKE SIGWAIT_HANDLER VP-SAFE

    PROBLEM DESCRIPTION:
    Customer may experience process hang due to mismanaged
    signals.

    PROBLEM CONCLUSION:
    Make sigwait_handler function vp-safe.

    ------

    APAR: IY15867 COMPID: 5765C3403 REL: 433
    ABSTRACT: FTPD CORE DUMPS

    PROBLEM DESCRIPTION:
    ftpd core dumps in yyparse() .

    PROBLEM SUMMARY:
    ftpd core dumps in yyparse() .

    ------

    APAR: IY16133 COMPID: 5765E2911 REL: 430
    ABSTRACT: ENCONSOLE/ENCCP SHOULD NOT ALLOW AN OBJECT NAMED DEFAULT TO BE

    PROBLEM DESCRIPTION:
    Enconsole/enccp should not allow most operations on and object
    with the name 'default'. Previously, you could create an
    object named 'default' from enconsole or enccp, but when you
    later tried to query/show the object, you ended up with the
    default attributes for that attribute type. Other operations
    on that object would either fatal, or have unexpected results
    (because they too try to operate on the default object).

    PROBLEM SUMMARY:
    USERS AFFECTED: Encina users, Release 420
    Enconsole/enccp should not allow most operations on and object
    with the name 'default'. Previously, you could create an
    object named 'default' from enconsole or enccp, but when you
    later tried to query/show the object, you ended up with the
    default attributes for that attribute type. Other operations
    on that object would either fatal, or have unexpected results
    (because they too try to operate on the default object).

    PROBLEM CONCLUSION:
    The solution to prevent the errors and confusion is to disallow
    operations, excluding show/modify, on an object named
    'default'. Show/modify on default is allowed, since those
    commands operate on the default object. Users can no longer
    create objects named 'default'.

    ------

    APAR: IY16134 COMPID: 5765E2911 REL: 430
    ABSTRACT: HEURISTIC DAMAGE EVENT SHOULD BE WARNING

    PROBLEM DESCRIPTION:
    There is a heuristic damage trace point in the tmxa component
    that is logged as an event. This message should be logged as a
    warning.

    PROBLEM SUMMARY:
    USERS AFFECTED: Encina users, Release 420
    There is a heuristic damage trace point in the tmxa component
    that is logged as an event. This message should be logged as a
    warning.

    PROBLEM CONCLUSION:
    Changed the tmxa trace event for 'Recording heuristic damage
    for %s' to a warning.

    ------

    APAR: IY16135 COMPID: 5765E2911 REL: 430
    ABSTRACT: WSTIDL CREATED INCORRECT INTERNAL DATA TYPES FOR ARRAYS. THE

    PROBLEM DESCRIPTION:
    Wstidl created incorrect internal data types for arrays.
    The generated code caused compile errors. The data type was
    declared to be the same as the array elements instead of the
    actual array itself. This fix fully supports both Fixed arrays
    and Conformant arrays.

    PROBLEM SUMMARY:
    USERS AFFECTED: Encina users, Release 420
    Wstidl created incorrect internal data types for arrays.
    The generated code caused compile errors. The data type was
    declared to be the same as the array elements instead of the
    actual array itself. This fix fully supports both Fixed arrays
    and Conformant arrays.

    PROBLEM CONCLUSION:
    Code changes implimented to fully support both Fixed and
    Conformant arrays.

    ------

    APAR: IY16136 COMPID: 5765E2911 REL: 430
    ABSTRACT: DCE PRIN WITH $ IN THE NAME CANNOT COLD START MAS SERVERS.

    PROBLEM DESCRIPTION:
    Dce principals that have a "$" in the name could not cold start
    any type of server. The same principals could warm start
    servers once they had been cold started by another principal
    that did not have this character in the uid.

    PROBLEM SUMMARY:
    USERS AFFECTED: Encina users, Release 420
    Dce principals that have a "$" in the name could not cold start
    any type of server. The same principals could warm start
    servers once they had been cold started by another principal
    that did not have this character in the uid.

    PROBLEM CONCLUSION:
    Encina looks for reference strings in the form of $string$. In
    order to have a '$' in a direct value, it needs to be escaped
    ('$$'). Changes were made to scan direct value strings (-A for
    cold starts, etc...) for '$' and insert a second '$' for each
    occurrence.
    Note: due to operating system restrictions this fix for Encina
    4.2 is limited to supporting principal names with the $ sign at
    the end of the string only. $'s in the middle or beginning will
    result in startup failures.

    ------

    APAR: IY16137 COMPID: 5765E2911 REL: 430
    ABSTRACT: WEBSPHERE ADVANCED TO ENCINA INTEROPERABILITY COMPONENT WAS NOT

    PROBLEM DESCRIPTION:
    Previously the WebSphere Advanced to Encina Interoperability
    component, which enables Java applications based on enterprise
    beans to interoperate with Encina, was available only for the
    AIX 4.3 and Windows NT 4.0 operating systems. It was not
    available for any supported versions of Solaris.

    PROBLEM SUMMARY:
    USERS AFFECTED: Encina users, Release 420
    Previously the WebSphere Advanced to Encina Interoperability
    component, which enables Java applications based on enterprise
    beans to interoperate with Encina, was available only for the
    AIX 4.3 and Windows NT 4.0 operating systems. It was not
    available for any supported versions of Solaris.

    PROBLEM CONCLUSION:
    The WebSphere Advanced to Encina Interoperability component is
    now supported for the Solaris 7 operating system.

    ------

    APAR: IY16138 COMPID: 5765E2911 REL: 430
    ABSTRACT: OUTPUT STRING PARAMETERS IN WSTIDL GENERATED FILES ARE NOT

    PROBLEM DESCRIPTION:
    Output string parameters in Wstidl generated files are not
    handled properly. If the string is the function 'returned'
    value it worked properly but if it is defined as an out or
     inout parameter, it generated a runtime error -
    CORBA.TRANSACTION_ROLLEDBACK.

    PROBLEM SUMMARY:
    USERS AFFECTED: Encina users, Release 420
    Output string parameters in Wstidl generated files are not
    handled properly. If the string is the function 'returned'
    value it worked properly but if it is defined as an out or
     inout parameter, it generated a runtime error -
    CORBA.TRANSACTION_ROLLEDBACK.

    PROBLEM CONCLUSION:
    Code changes made to handle all parameter types.

    ------

    APAR: IY16139 COMPID: 5765E2911 REL: 430
    ABSTRACT: A FATAL ERROR COULD OCCUR WHEN A MON_CONCURRENT_SHARED PA USES

    PROBLEM DESCRIPTION:
    A fatal error could occur when a MON_CONCURRENT_SHARED pa uses
    transaction duration locking to serialize access to a single
    association RM with Encina.

    PROBLEM SUMMARY:
    USERS AFFECTED: Encina users, Release 420
    A fatal error could occur when a MON_CONCURRENT_SHARED pa uses
    transaction duration locking to serialize access to a single
    association RM with Encina.

    PROBLEM CONCLUSION:
    Code changes made, including removing thread reassociation in
    the presence of transaction locking, to correct the problem.

    ------

    APAR: IY16140 COMPID: 5765E2911 REL: 430
    ABSTRACT: TRPC_GETMANAGERFUNCTION NEEDS ENCINA_CALLING IN ITS PROTOYPE

    PROBLEM DESCRIPTION:
    A calling convention mismatch existed on the NT platform for
    the trpc function - trpc_GetManagerInfo. Attempts to use this
    function in NT applications resulted in the linker error -
    "unresolved external symbol _trpc_GetManagerInfo" and required
    user modification to the function prototype.

    PROBLEM SUMMARY:
    USERS AFFECTED: Encina users, Release 420
    A calling convention mismatch existed on the NT platform for
    the trpc function - trpc_GetManagerInfo. Attempts to use this
    function in NT applications resulted in the linker error -
    "unresolved external symbol _trpc_GetManagerInfo" and required
    user modification to the function prototype.

    PROBLEM CONCLUSION:
    Changes were made to the prototype to correct this problem.

    ------

    APAR: IY16141 COMPID: 5765E2911 REL: 430
    ABSTRACT: TRANSLATETRACEID DISLIKES LANG=C

    PROBLEM DESCRIPTION:
    The translateTraceId tool did not work correctly on the HP_UX
    platform with the environment variable LANG=C set. Certain
    function calls used to retrieve data from catalog files
    returned format strings instead of the actual data.

    PROBLEM SUMMARY:
    USERS AFFECTED: Encina users, Release 420
    The translateTraceId tool did not work correctly on the HP_UX
    platform with the environment variable LANG=C set. Certain
    function calls used to retrieve data from catalog files
    returned format strings instead of the actual data.

    PROBLEM CONCLUSION:
    Changes were made to use NLS safe function calls to retrieve
    the message data. almasy-Thu, 30 Nov 2000 14:06:44

    ------

    APAR: IY16301 COMPID: 5765E2810 REL: 430
    ABSTRACT: TXSERIES ENCINA 4.3.0.4 ON AIX 4.2.1 AND AIX 4.3.2 REFERENCE

    PROBLEM DESCRIPTION:
    WEBSPHERE ENCINA 4.3 ON AIX 4.2.1 AND AIX 4.3.2 4.3.0.4
    REFERENCE APAR.
    THIS APAR SHOULD BE USED WHEN ORDERING THE LATEST WEBSPHERE
    ENCINA 4.3 MAINTENANCE ON THE AIX 4.2.1 AND AIX 4.3.2 OPERATING
    SYSTEMS.
    THE PTF ASSOCIATED WITH THIS REFERENCE APAR CONTAINS FIXES FOR
    THE FOLLOWING LIST OF APARS WHICH WERE OPENED AGAINST COMPONENT
    ID 5765E2810:
    IY16133 IY16134 IY16135 IY16136 IY16137 IY16138 IY16139 IY16140
    IY16141 IY15088

    PROBLEM CONCLUSION:
    IY16133 IY16134 IY16135 IY16136 IY16137 IY16138 IY16139
    IY16140 IY16141 IY15088

    ------

    APAR: IY16673 COMPID: 5765C3403 REL: 430
    ABSTRACT: DBX DOES NOT STOP LOOP IN EXAMINING CORE VMM

    PROBLEM DESCRIPTION:
    dbx's read of coredump's vmm regions in PRE430 corefiles
    coredumps itself.

    PROBLEM SUMMARY:
    dbx doesnt stop looping for vmm regions correctly

    PROBLEM CONCLUSION:
    Corrected loop test.

    ------

    APAR: IY16718 COMPID: 5765C3403 REL: 430
    ABSTRACT: MPQPDD_ISA DRIVER CAN GET INTO DEADLOCK SITUATION OR CRASH

    PROBLEM DESCRIPTION:
    System crashes on call to mbuf services in the mpqpdd_isa
    device driver.

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * devices.isa.mm2.mpqp 4.3.3.26
    * devices.isa.mm2.mpqp 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * System crashes on call to mbuf services in mpqpdd_isa
    * device driver. Hang may occur waiting on a lock when
    * using the mpqpdd_isa driver.
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY16718
    ***************************************************************

    PROBLEM CONCLUSION:
    Insure that locks are not held when calling m_get and a
    wait is allowed (M_WAIT specified).

    ------

    APAR: IY16813 COMPID: 5765C3403 REL: 430
    ABSTRACT: VMGETINFO/KVMGETINFO SYSTEM CALLS IMPACT PERF.

    PROBLEM DESCRIPTION:
    vmgetinfo/kvmgetinfo system calls are taking too
    long to return. This is causing perf. impact.
    This should not be much of concern for users
    unless they are trying to obtain continuously the
    VMM statistics through vmstat or other tools which
    makes use of these system calls.

    LOCAL FIX:
    Don't use these tools if not needed.

    PROBLEM SUMMARY:
    System perf. (but only when the vmgetinfo is being
    called in a close loop).

    PROBLEM CONCLUSION:
    Add a new command option VMINFO_ABRIDGED to vmgetinfo
    and kvmgetinfo. This command when passed to upd_vminfo
    (called by vmgetinfo and kvmgetinfo) would skip the time
    consuming calculation of some the parameters in vminfo
    structure which are rarely used. Only svmon uses these
    additional stats. No other command does.
    Also make changes in the tool code to call vmgetinfo with
    VMINFO_ABRIDGED instead of VMINFO if detailed stats is not
    required.

    ------

    APAR: IY16929 COMPID: 5765C3403 REL: 430
    ABSTRACT: CANNOT LIST/CHANGE/REMOVE ETHERCHANNEL ON SMIT

    PROBLEM DESCRIPTION:
    Cannot delete or change a defined etherchannel through smit or
    smitty if LANG is set to Japanese.

    PROBLEM SUMMARY:
    The english language string 'therchannel' was being searched
    for in the device descriptions to locate Etherchannel adapters.
    This has now been fixed where the proper class and type are
    used to locate the adapters.

    PROBLEM CONCLUSION:
    This fix makes the adapter detection in smit independent of
    language environment.

    ------

    APAR: IY16965 COMPID: 5765D5100 REL: 320
    ABSTRACT: SP SWITCH 2 IP PROBLEM- POSSIBLE LARGE PACKET FAILURES AFTER

    PROBLEM DESCRIPTION:
    On the SP Switch 2, TCP/IP communication of large packets can
    fail after a large number of idrops (check with netstat -D)
    occur on the css0 interface.

    LOCAL FIX:
    reboot failing node

    PROBLEM SUMMARY:
    Properly handle returning cluster buffers back to the
    receive pool
    when the receive queue is overrun.

    PROBLEM CONCLUSION:
    when the colony ip driver overruns it's receive pool it
    doesn't recovery it's cluster buffer cause a cluster buffer
    leak.

    ------

    APAR: IY17082 COMPID: 5765C3403 REL: 430
    ABSTRACT: BACKBYNAME FAILS TO ARCHIVE DATA GREATER THAN 256GB

    PROBLEM DESCRIPTION:
    backup command in backbyname format fails to archive
    data greater then 256GB with errors.

    PROBLEM SUMMARY:
    backup command fails to archive data greater than
    256GB with errors.

    ------

    APAR: IY17120 COMPID: 5765C3403 REL: 433
    ABSTRACT: IPREPORT CORE DUMPS READING TCPDUMPS BINARIES

    PROBLEM DESCRIPTION:
     ipreport core dumps reading tcpdumps binaries.

    PROBLEM SUMMARY:
    ipreport core dumps reading tcpdump binaries

    PROBLEM CONCLUSION:
    The core dump was due to the buffer pointer
    pointing to unassigned memory. Appropriate
    changes were made to fix the problem.

    ------

    APAR: IY17121 COMPID: 5765C3403 REL: 430
    ABSTRACT: BOOTPTODHCP FAILS TO CONVERT T170 TO HEX

    PROBLEM DESCRIPTION:
    bootptodhcp fails to convert T170 to hex

    PROBLEM SUMMARY:
    The problem is fixed by changing bootptodhcp.c code.

    PROBLEM CONCLUSION:
    Done a check for the proper value taken by T170 in dhcpsd.cnf.

    ------

    APAR: IY17172 COMPID: 5765C3403 REL: 430
    ABSTRACT: CLIPPED CONCAVE POLYGONS ARE SLOW ON GXT6000

    PROBLEM DESCRIPTION:
    Dust particles are displayed in certain area of a model. Pixel
    level holes or overpaints are displayed on irregular pixel
    position for each redraw. It looks like a dust particle. It
    emerges not on whole screen but on certain area of a model.

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *el
    * PEX_PHIGS.dev.pci.14106e01 4.3.3.4d on irregular pixel
    * PEX_PHIGS.dev.pci.14106e01 4.3.3.5ke a dust particle. It
    * PEX_PHIGS.dev.pci.14106e01 4.3.3.50ain area of a model.
    * PEX_PHIGS.dev.pci.14107001 4.3.3.4this problem, but a hang
    * PEX_PHIGS.dev.pci.14107001 4.3.3.5
    * PEX_PHIGS.dev.pci.14107001 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * The following problem occurs on GXT6000P when doing
    * clipped polygons: Dust particles are displayed in
    * certain area of a model. Pixel level holes or
    * overpaints are displayed on irregular pixel position
    * for each redraw. It looks like a dust particle. It
    * emerges not on whole screen but on certain area of a
    * model. APAR IY14967 was the original fix for this
    * problem, but a hang was introduced in certain
    * situations.
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY17172
    ***************************************************************

    PROBLEM CONCLUSION:
    The original fix had performance implications. This new fix
    regains the original performance for rendering clipped poly-
    gons. The new fix also removes the hang situations caused by
    APAR IY14967. The new fix utilizes a different set of hardware
    commands when rendering clipped polygons.

    ------

    APAR: IY17188 COMPID: 5765D5100 REL: 320
    ABSTRACT: EMASTERD HANGS AT "LSSRC -A | GREP HAEM"

    PROBLEM DESCRIPTION:
    emasterd hangs in wait_for_haem loop at "/usr/bin/lssrc -a |
     /bin/grep haem > /tmp/wait_to_start.emaster_out", because
    lssrc -a does not return.
    /var/adm/SPlogs/css/emasterd.log looks like:
    ...
    emasterd_main: 2543-717 Entering wait_for_haem() loop.
    emasterd_main: 2543-719 In wait_for_haem() loop 1 hours.
    ...
    emasterd_main: 2543-719 In wait_for_haem() loop X 11 hours.

    PROBLEM SUMMARY:
    The emaster daemon loops forever in the wait_for_haem()
    function because a system library call which runs the
    command: "lssrc -a | grep haem" fails (rc = 256). This
    is due to a problem in the bourne shell which may occur
    on some levels of aix. This usually happens if the
    command we are trying to execute passes a flag. Although
    the emaster daemon is part of the SP_Switch2 support,
    this particular problem may also be seen on SP_Switch
    systems.

    PROBLEM CONCLUSION:
    The emaster daemon will set the PWD environment
    variable to the /var/adm/SPlogs/css directory before
    issuing any system() calls.

    ------

    APAR: IY17213 COMPID: 5765C3403 REL: 430
    ABSTRACT: DSI IN SYSTEM LOADER (STRLEN)

    PROBLEM DESCRIPTION:
    Under certain error conditions A DSI will occur in strlen()
    when called from the system loader.

    PROBLEM SUMMARY:
    Kernel DSI during rare error conditions.

    PROBLEM CONCLUSION:
    NULL pointers to avoid their re-use after they're free'd.

    ------

    APAR: IY17228 COMPID: 5765D5100 REL: 320
    ABSTRACT: ADAPTREVC FSD DUMP SRAM MC FAILES SHOULD NOT ESCALATE PA

    PROBLEM DESCRIPTION:
    adaptrevc fsd dump sram mc fails should not escalate PA

    PROBLEM SUMMARY:
    Transient adapter errors were escalating to
    permanent errors which in turn would take the node of the
    switch when the microcode failed to dump its cache with a
    timeout limit.

    PROBLEM CONCLUSION:
    Don't take node off switch on transient
    errors when the microcode fails to dump cache in within
    timeout limit.

    ------

    APAR: IY17269 COMPID: 5765D5100 REL: 320
    ABSTRACT: COLONY: MICROCODE / KLAPI DEADLOCK (PMR 63030)

    PROBLEM DESCRIPTION:
    The microcode is stuck on a '1b'x command in the KHAL window.
    It can make progress for other windows, but the microcode
    can't make any progress for KHAL because it needs a slot in
    the receive fifo. Unfortunately, KLAPI/KHAL won't free up
    a slot in the receive fifo, until it has a slot in the send
    command fifo. This is a classic deadlock senario.

    PROBLEM SUMMARY:
    GPFS can hang due to SFIFO/RFIFO deadlock condition
    between KLAPI and microcode on the SP_switch2 adapter.

    PROBLEM CONCLUSION:
    This fix will change the protocol handshake between
    microcode and KLAPI on the SP_switch2 adapter to
    avoid the SFIFO/RFIFO deadlock.

    ------

    APAR: IY17306 COMPID: 5765C3403 REL: 433
    ABSTRACT: ONLY ONE REMOTE LOGIN SESSION IS ALLOWED FOR ONE ID

    PROBLEM DESCRIPTION:
    Data cannot go through secure tunnel after remote clent
    re-login using same ID.

    PROBLEM SUMMARY:
    Data cannot go through secure tunnel after remote client
    re-login using same ID.

    PROBLEM CONCLUSION:
    AIX IKE will only support one login seesion for one ID,
    re-login wil replace its previous filters.

    ------

    APAR: IY17320 COMPID: 5765C3403 REL: 433
    ABSTRACT: SLOW MOTIF 1.2 MEMORY LEAK WHEN EXECUTING JAVA APPLICATION

    PROBLEM DESCRIPTION:
    Slow memory leak when executing a Motif 1.2 application. Problem
    was uncovered executing Java application.

    LOCAL FIX:
    Change client code to avoid creating and deleting multiple
    XmDisplays. The leak occurs due to incomplete memory
    destruction of XmDisplay.

    PROBLEM SUMMARY:
    The customer will experience a slow memory leak in
    some Motif 1.2 apps.

    PROBLEM CONCLUSION:
    Motif 1.2 library changed to fully free the memory of an
    XmDisplay when it is destroyed.

    ------

    APAR: IY17327 COMPID: 5765C3403 REL: 430
    ABSTRACT: RAMDISK CRASHES ON CROSS MEMORY BUFFERS FROM AIO

    PROBLEM DESCRIPTION:
    RAMDISK crashes on cross memory buffers from aio

    PROBLEM SUMMARY:
    System dump:
        MST STACK TRACE:
        0x2ff3b400
    (excpt=bfe3a000:42000000:000d047a:bfe3a000:00000106)(intpri=11)
         IAR: .panic_trap+0 (00012550): tweq r1,r1
         LR: . ramdisk:rd_strategy +150 (016e1650)
         2ff3b150: .devstrat+224 (00078af0)
         2ff3b1a0: . aio.ext:arl_rdwr +4b8 (015ed654)
         2ff3b220: . aio.ext:ardwr +5d4 (015ecfc4)
         2ff3b320: . aio.ext:kaio_rdwr +158 (015edcc8)
         2ff3b3c0: .sys_call_ret+0 (00003a90)

    PROBLEM CONCLUSION:
    Ramdisk strategy routine must be changed to split buffers which
    cross the 256MB memory boundary.

    ------

    APAR: IY17361 COMPID: 5765C3403 REL: 430
    ABSTRACT: THE SYSTEM CRASHES DURING PAGEOUT.

    PROBLEM DESCRIPTION:
    The system hangs trying to do a pageout. The stack trace
    IAR: .v_pdtsio+118 (00048aa4): lwz r0,0xc(r3)
    LR: .v_pfsio+10c (000486bc) IAR not in kernel segment.
    The system tries to do a pageout and finds that the page has
    no associated paging device (PDT) available.

    TEMPORARY FIX:
    do not mount filesystem as read only.

    ------

    APAR: IY17375 COMPID: 5765C3403 REL: 430
    ABSTRACT: CAN NOT CHANGE DEFAULT RULE LOG FLAG

    PROBLEM DESCRIPTION:
    Any attempt to change the logging for the default rule will
    not work.

    PROBLEM SUMMARY:
    can't log default filter rule

    PROBLEM CONCLUSION:
    Need to change code to make logging
    function work.

    TEMPORARY FIX:
    Put filter rule ahead of default rule
    that does the same thing, and turn
    logging on.

    ------

    APAR: IY17377 COMPID: 5765C3403 REL: 430
    ABSTRACT: PASSWORD ECHOED ON TIMEOUT OF PASSWD COMMAND

    PROBLEM DESCRIPTION:
    After typing passwd and then typing the old password but not
    hitting return and letting it time out, it comes back with an
    3004-709 Error changing password for the user. Then at the next
    line whatever has been typed in the old password field is
    echoed back. This can be recreated on 421 and 433 machines.

    PROBLEM SUMMARY:
    The old password field is always diaplayed
    when the time out is made to occur by the user,
    on the shell prompt.

    PROBLEM CONCLUSION:
    The timeout will clear the buffer containing
    the old password and it will not be displayed.

    TEMPORARY FIX:
    Never time out at the password prompt.

    ------

    APAR: IY17380 COMPID: 5765C3403 REL: 430
    ABSTRACT: NISPLUS FIXES

    PROBLEM DESCRIPTION:
    This APAR contains a collection of minor fixes to nisplus.
    Some smit screens have errors in them, password aging does
    not always work correctly and adding/removing nisplus from
    clients and servers does not clear the domainname.

    PROBLEM CONCLUSION:
    nismkuser: Add support for named groups, by adding a lookup
    name->number in the logic.
    nisplus password aging: Change logic to allow one login after
    aging has expired, just as the other authentication methods
    do.
    nisclient: clear domainname when doing unconfigure. Also,
    change the "grep" arguments to avoid adding redundant lines
    to /etc/hosts.

    ------

    APAR: IY17389 COMPID: 5765C3403 REL: 430
    ABSTRACT: M:N THREAD CREATED DETACHED PASSSES SIGNALS WHEN EXITING

    PROBLEM DESCRIPTION:
    Threaded Applications may fail to start

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * bos.adt.prof 4.3.3.28
    * bos.adt.prof 4.3.3.29
    * bos.adt.prof 4.3.3.30
    * bos.adt.prof 4.3.3.50
    * bos.rte.libpthreads 4.3.3.27
    * bos.rte.libpthreads 4.3.3.28
    * bos.rte.libpthreads 4.3.3.29
    * bos.rte.libpthreads 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * Threaded Applications may fail to start
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY17389
    ***************************************************************

    PROBLEM CONCLUSION:
    in _set_vp, check for thread state of Returned and detached
    state when setting the right sigmask for the vp to be
    put to sleep

    ------

    APAR: IY17426 COMPID: 5765C3403 REL: 430
    ABSTRACT: PATTERNED LINE CAN DISPLAY INCORRECTLY ON GXT6000P

    PROBLEM DESCRIPTION:
    A patterned line with a pattern length that exceeds 4 will
    display incorrectly on GXT6000P.

    PROBLEM SUMMARY:
    A patterned line with a pattern length greater than 4 will
    display incorrectly on GXT6000P.

    PROBLEM CONCLUSION:
    Change the shift count used to combine the pattern length into
    the Line Style Pattern commands to the hardware from 20 to 16.

    ------

    APAR: IY17434 COMPID: 5765D5100 REL: 320
    ABSTRACT: KHAL NEEDS TO IMPLEMENT HAL_STOP TO HELP DEBUG

    PROBLEM DESCRIPTION:
    KHAL needs to implement hal_stop to help debug
    problems in the field.

    PROBLEM SUMMARY:
    This is used by developers to debug KLAPI code
    when unexpected error occurred.

    PROBLEM CONCLUSION:
    Added hal_stop and time stamp functions
    to khal.

    ------

    APAR: IY17437 COMPID: 5765C3403 REL: 430
    ABSTRACT: LINK POLLING FUNCTIONALITY ADDED TO 10/100 ETHERNET ADAPTER

    PROBLEM DESCRIPTION:
    Link Down Error checks in interrupt code causes a performance
    degredation.

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * devices.pci.23100020.rte 4.3.3.13
    * devices.pci.23100020.rte 4.3.3.14
    * devices.pci.23100020.rte 4.3.3.15
    * devices.pci.23100020.rte 4.3.3.25
    * devices.pci.23100020.rte 4.3.3.26
    * devices.pci.23100020.rte 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * Link Down Error checks in interrupt code causes a
    * performance degredation.
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY17437
    ***************************************************************

    PROBLEM CONCLUSION:
    Introduced a link polling ODM attribute which can be used to
    enable link polling if required.

    ------

    APAR: IY17439 COMPID: 5765D5101 REL: 120
    ABSTRACT: ORACLE CAN NOT CONNECT HAGSD WITH MANY CLIENTS

    PROBLEM DESCRIPTION:
    Oracle can not connect hagsd with many clients.

    PROBLEM SUMMARY:
    Group Services daemon currently limits the maximum
    pending connections to 5. Therefore, if there are
    many connection requests(e.g., over 500) have been
    made to Group Services at a short time, some of
    the requests may be disallowed with "ECONNREFUSED".
    By extending the maximun pending queue length
    (called, backlog size) to the maximum configured
    queue length (i.e., no -o somaxconn), the unexpected
    connection refusal will be resolved.

    PROBLEM CONCLUSION:
    This fix will enable Group Services subsystem
    to handle many concurrent connection requests
    even if the requests have made at very short
    period.

    ------

    APAR: IY17459 COMPID: 5765C3403 REL: 430
    ABSTRACT: HTTP PORT STOPS RESPONDING WHEN FRCA USED

    PROBLEM DESCRIPTION:

    PROBLEM SUMMARY:
    When FRCA is used, http sessions could hang

    PROBLEM CONCLUSION:
    On some error path where the FRCA cannot handle the http
    requests, wake the user space process to handle all the
    queued requests.

    ------

    APAR: IY17462 COMPID: 5765D5100 REL: 320
    ABSTRACT: GPFS HANG CONDITION

    PROBLEM DESCRIPTION:
    GPFS hang condition.

    PROBLEM SUMMARY:
    When critical adapter recovery occurs there are cases
    RAMBUS resource on the CSS adapter can be incorrectly used.

    PROBLEM CONCLUSION:
    The software fix is to make sure this timing hole is
    removed and the communication protocol and the microcode
    on the adapter have the same understanding of RAMBUS usage.

    ------

    APAR: IY17528 COMPID: 5765C3403 REL: 430
    ABSTRACT: NETSTAT -AN HANGS WHILE DELETING UDP SOCKET AT THE SAME TIME

    PROBLEM DESCRIPTION:
    netstat -an command hangs while deleting
    the socket at the same time

    PROBLEM SUMMARY:
    netstat -an command hangs while deleting the socket
    at the same time is fixed by checking to see if
    the hash table linked list of inpcb is changed.
    If it's changed, we simply return to get out of
    the dead loop.

    PROBLEM CONCLUSION:
    netstat -an command hangs while deleting the socket
    at the same time is fixed by checking to see if
    the hash table linked list of inpcb is changed.
    If it's changed, we simply return to get out of
    the dead loop

    ------

    APAR: IY17529 COMPID: 5765C3403 REL: 430
    ABSTRACT: DO NOT DROP LOCK BEFORE IP_FREEMOPTIONS

    PROBLEM DESCRIPTION:
    ip_setmoptions depends on the socket lock but
    we unlock the socket before calling the
    ip_freemoptions function.

    PROBLEM SUMMARY:
    Fixed crash resulting from use of multicast options
    buffer that was freed without proper locking.

    PROBLEM CONCLUSION:
    Freeing of multicast options buffer now occurs with
    proper locking.

    ------

    APAR: IY17578 COMPID: 5765C3403 REL: 430
    ABSTRACT: PCBNOTIFY SWITCHES CHAINS

    PROBLEM DESCRIPTION:
    System crashes when we fail to detect element
    being assigned a port number.

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * bos.net.tcp.client 4.3.3.25
    * bos.net.tcp.client 4.3.3.26
    * bos.net.tcp.client 4.3.3.27
    * bos.net.tcp.client 4.3.3.28
    * bos.net.tcp.client 4.3.3.29
    * bos.net.tcp.client 4.3.3.30
    * bos.net.tcp.client 4.3.3.31
    * bos.net.tcp.client 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * System crashes when we fail to detect element being
    * assigned a port number.
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY17578
    ***************************************************************

    PROBLEM CONCLUSION:
    We will detect entry into another chain for udp.

    ------

    APAR: IY17587 COMPID: 5765C3403 REL: 430
    ABSTRACT: APPLICATIONS FAILING AS NLSPATH BLANKED OUT

    PROBLEM DESCRIPTION:
    The NLSPATH env variable is blanked out from
    a security fix and due to this the applications
    are failing as they cannot find the message catalogs.

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * bos.mp 4.3.3.30
    * bos.mp 4.3.3.31
    * bos.mp 4.3.3.32
    * bos.mp 4.3.3.50
    * bos.up 4.3.3.30
    * bos.up 4.3.3.31
    * bos.up 4.3.3.32
    * bos.up 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * The NLSPATH environment variable is blanked out due to
    * a security fix. Applications which require a NLSPATH
    * environment variable do not work properly as a
    * result.
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY17587
    ***************************************************************

    PROBLEM CONCLUSION:
    The NLSPATH is not blanked out but instead two command
    were provided to set/view the path. If the path is not
    set, it would take the default NLSPATH value which is
    /usr/lib/nls. If the applications use a different path
    than this, then chnlspath is used to set that path.
    lsnlspath is used to display the current setting.

    ------

    APAR: IY17622 COMPID: 5765D5100 REL: 320
    ABSTRACT: ASCI: ADD COMPILE AND LINK OPTIONS TO IMPROVE RAS

    PROBLEM DESCRIPTION:
    ASCI: add compile and link options to improve storage RAS

    PROBLEM SUMMARY:
    There is no external problem associated with this.

    PROBLEM CONCLUSION:
    The makefiles for the fault service daemon are changed to
    specify that read-only strings be put into read-only
    storage.

    ------

    APAR: IY17630 COMPID: 5765C3403 REL: 430
    ABSTRACT: TCP CLIENT SENDS RST FLAG TO SNMPD, CAUSED IT EXITS BECAUSE THE

    PROBLEM DESCRIPTION:
    snmpd daemon dies(not core dumped) during
    the accept() call while the tcp client sends
    snmpd a RST flag.

    ------

    APAR: IY17646 COMPID: 5765C3403 REL: 430
    ABSTRACT: QDAEMON OCCASIONALLY HANGS

    PROBLEM DESCRIPTION:
    qdaemon hangs occasionally while handling SIGTRAP.
    When it gets into that hung state, it eventually goes
    down and has to be restarted (This is usually taken care
    of by the entry in /etc/inittab).

    LOCAL FIX:
    When qdaemon gets into that hung state, stop it
    and restart it using stopsrc and startsrc.

    PROBLEM SUMMARY:
    qdaemon might occasionally hang.

    PROBLEM CONCLUSION:
    Signal handling for qdaemon has been changed to prevent this
    situation.

    ------

    APAR: IY17659 COMPID: 5765C3403 REL: 430
    ABSTRACT: COPY FROM SERVER TO LOCAL DOESNOT INHERIT READONLY ATTRIBUTE

    PROBLEM DESCRIPTION:
    Copying a file from a mapped drive to
    local drive does not inherit the readonly
    attribute.

    LOCAL FIX:
    send the correct attributes.

    PROBLEM CONCLUSION:
    Checking attributes in the print server

    ------

    APAR: IY17685 COMPID: 5765C3403 REL: 430
    ABSTRACT: SYSTEM WILL HANG WHEN FAULTY CX ADAPTER DETECTED

    PROBLEM DESCRIPTION:
    The system will hang with a cpu stuck in scxma_cmdw and the
    other cpus waiting for the board lock held by the first cpu.

    PROBLEM CONCLUSION:
    The solution is to change how the delay is done so that it
    will last only for a short while and it does not create a
    deadlock.

    ------

    APAR: IY17688 COMPID: 5765C3403 REL: 430
    ABSTRACT: DEFAULT ATTRIBUTES USED INSTEAD OF USER ATTRIBUTES IN DCE

    PROBLEM DESCRIPTION:
    The attributes specified in a stanza file like
    /etc/security/user or /etc/security/limits for a user
    are not picked up in the DCE environment and always
    default stanza attributes are used. An example is,
    if we have a sugroups=ALL in default and sugroups=XYZ
    for a root user, the dce user still will able to su to
    root even though he doesn't belong to XYZ group. This is
    because the attibute is taken from default stanza.

    LOCAL FIX:
    Remove the default stanza and it will work in some cases.

    PROBLEM SUMMARY:
    The attributes are not picked from the user
    stanza even if user stanza is present and is
    always picked up from default stanza.

    PROBLEM CONCLUSION:
    The attributes will be picked from the user
    stanza if available. If this is not present,
    attribute values are taken from default stanza.

    ------

    APAR: IY17690 COMPID: 5765D5100 REL: 320
    ABSTRACT: COLONY: ADAPTRECV ERRSLIH SHOULD PASS RESET RETCODE TO FSD RECV

    PROBLEM DESCRIPTION:
    colony:adaptrecv errslih should pass reset retcode to rsd recv

    PROBLEM SUMMARY:
    fault_service_daemon needed to know the return code from the
    adapter reset when processing a critical adapter.

    PROBLEM CONCLUSION:
    Device driver returns the errno from the adpter reset.

    ------

    APAR: IY17691 COMPID: 5765D5100 REL: 320
    ABSTRACT: COL_DUMP IS NOT WORKING PROPERLY USING BINARY INPUT FILES

    PROBLEM DESCRIPTION:
    col_dump is not working properly using binary input files

    PROBLEM SUMMARY:
    A hardware upgrade was made to the newer colony adapters
    equipping them with 4MB SRAM simms from 2MB SRAM simms. Due
    to this change, the microcode that was loaded into the
    adapter changed location. The problem with col_dump was that
    it was not able to distinguish a difference in the binary
    file size, therefore given a 4MB binary file, col_dump was
    printing out data from an improper offset or locatoin within
    the binary dump file.

    PROBLEM CONCLUSION:
    The executable col_dump had to be modified to identify the
    correct offset into the microcode binary dump file obtained
    using colread to reflect a 4MB or a 2MB dump file size. Once
    the program was able to identify between the 2 file sizes it
    became possible to use the proper offset to read the data
    from the binary files.

    ------

    APAR: IY17704 COMPID: 5765C3403 REL: 430
    ABSTRACT: ACCEPT()RETURNS ENOTCONN WHEN IT IS NOT SUPPOSED TO

    PROBLEM DESCRIPTION:
    accept() will return ENOTCONN error sometimes if you do
    multiple simultaneous accept()s and there are less number of
    connections queued compared to the the number of simultaneous
    accept()s.

    PROBLEM CONCLUSION:
    If a blocking socket, go back to sleep and wait for new
    connections. If non-blocking, return EWOULDBLOCK.

    ------

    APAR: IY17721 COMPID: 5765D5100 REL: 320
    ABSTRACT: HAL 420 STARTING A USER SPACE JOB

    PROBLEM DESCRIPTION:
    Customer receive the errors to std out:
    D1<L1>: About to call mpci_connnect
    HAL ERROR CODE = 420, (<directory_structure>)
     <program_name>.c(622):pid=86e0)
     ERROR: 0031-309 Connect failed during message passing
     initialization, task 1162, reason: Switch interface internal
     communications error
     D1<L1>: mpci_connect return code =307, task 1162
     ERROR: 0031-007 Error initializing communication subsystem:
     return code -1
    While starting a user space job.

    PROBLEM SUMMARY:
    memory usage for user-space windows was not tracked
    correctly.

    PROBLEM CONCLUSION:
    correctly handle tracking of memory usage for user-space
    windows.

    ------

    APAR: IY17729 COMPID: 5765C3403 REL: 430
    ABSTRACT: LFT NEED NOT LOG ERROR AT UNCONFIGURE ATTEMPT

    PROBLEM DESCRIPTION:
    Currently the lft will log an error when it is open and an
    attempt to unconfigure the lft is made. The unconfigure
    attempt should return with EBUSY which it does but there
    is no need to log an error. The extra log message is
    confusing customers.

    PROBLEM SUMMARY:
    Currently the lft will log an error when it is open and an
    attempt to unconfigure the lft is made. The unconfigure
    attempt should return with EBUSY which it does but there
    is no need to log an error. The extra log message is
    confusing customers.

    PROBLEM CONCLUSION:
    The solution is to remove the error.

    ------

    APAR: IY17756 COMPID: 5765C3403 REL: 430
    ABSTRACT: ADDING A ROUTE USES WRONG INTERFACE ROUTE

    PROBLEM DESCRIPTION:
    When adding a network route, the interface chosen should be
    the interface through which there exists an interface route
    that allows the gateway to be reached.

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * bos.mp 4.3.3.32
    * bos.mp 4.3.3.50
    * bos.up 4.3.3.32
    * bos.up 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * When adding a network route, the interface chosen
    * should be the interface through which there exists an
    * interface route that allows the gateway to be
    * reached.
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY17756
    ***************************************************************

    PROBLEM CONCLUSION:
    Fix ifa_ifwithroute to lookup the route to the gateway, not
    the destination.

    TEMPORARY FIX:
    Add an alias to the interface rather than adding interface
    routes using the route command.

    ------

    APAR: IY17759 COMPID: 5765C3403 REL: 430
    ABSTRACT: GETSOCKOPT CRASHES THE SYSTEM ON IA64

    PROBLEM DESCRIPTION:
    On a IA64 platform, getsockopt() can crash when the socket
    is in disconnected state.

    PROBLEM CONCLUSION:
    If the inpcb is NULL, return ENOTCONN without accessing the
    fields of inp.

    ------

    APAR: IY17761 COMPID: 5765C3403 REL: 430
    ABSTRACT: NIS+: PARTIAL SUCCESS WHEN CHANGING PASSWORDS

    PROBLEM DESCRIPTION:
    When changing password in a NIS+ environment, the user gets the
    message: "Partial success changing NIS+ password.". In spite
    of this error message, the password actually gets changed
    correctly.

    LOCAL FIX:
    None, just disregard the error message.

    PROBLEM SUMMARY:
    When changing password in a NIS+ environment, the user gets the
    message: "Partial success changing NIS+ password.". In spite
    of this error message, the password actually gets changed
    correctly.

    PROBLEM CONCLUSION:
    change the password field only. Keep gecos and shell alone.

    ------

    APAR: IY17763 COMPID: 5765C3403 REL: 430
    ABSTRACT: PSE:IOCTL AND CLOSE RACE CONDITION PANIC THE SYSTEM

    PROBLEM DESCRIPTION:
    The ioctl and close race condition in the Streams causes
    system panic.

    PROBLEM CONCLUSION:
    Cancel all pending ioctl requests from the
    osr_close_subr(). Only the close() should remain
    in the IOCTL osrq list.

    ------

    APAR: IY17815 COMPID: 5765C3403 REL: 430
    ABSTRACT: PARTIAL WHEEL MOUSE ENABLEMENT

    PROBLEM DESCRIPTION:
    IY14962, Wheel Mouse Hardware enablement implementation
    really didn't provide all of Wheel Mouse support
    this apar will.

    PROBLEM CONCLUSION:
    Add remaining software enablement for Wheel Mouse.

    ------

    APAR: IY17831 COMPID: 5765C3403 REL: 430
    ABSTRACT: FLOATING POINT DIAGNOSTICS AND EXERCISER FAIL TO IDENTIFY A

    PROBLEM DESCRIPTION:
    floating point diagnostics and exerciser fail to
    isolate the correct failing processor on multi
    processor systems.

    PROBLEM SUMMARY:
    When running the floating point exerciser on a mp machine,
    and there is a faulty cpu, the supervisor may eventually show
    all CPUs as failing thereby making it difficult to
    troubleshoot the actual cpu, which originated the problem.

    PROBLEM CONCLUSION:
    Get the is_mp() function to work under the supervisor or
    replace it.

    ------

    APAR: IY17842 COMPID: 5765B8100 REL: 220
    ABSTRACT: STOP WORD CAUSES DSR TO CRASH

    PROBLEM DESCRIPTION:
    stop word causes DSR to crash

    PROBLEM CONCLUSION:
    A number of bug fixes in the code

    ------

    APAR: IY17852 COMPID: 5765C3403 REL: 430
    ABSTRACT: CAN'T SET BREAKPOINTS IN PRIVATE LIBS

    PROBLEM DESCRIPTION:
    Can't set breakpoints in privately loaded libraries.

    PROBLEM CONCLUSION:
    Fix the laoder's VM calls.

    ------

    APAR: IY17855 COMPID: 5765C3403 REL: 430
    ABSTRACT: UUCPADM HELP OPTION DOES NOT WORK

    PROBLEM DESCRIPTION:
    UUCPADM Help Option does not work.

    PROBLEM SUMMARY:
    Increased buffer size to handle help message.

    ------

    APAR: IY17856 COMPID: 5765C3403 REL: 430
    ABSTRACT: GXT6000P/4000P FAILS TO DISPLAY CURSOR COLOR

    PROBLEM DESCRIPTION:
    When using multicolor cursor extension with the GXT4000P or
    GXT6000P graphics adapter, the cursor may display the wrong
    color.

    PROBLEM CONCLUSION:
    Shift the 16 bit RGB color components for the cursor to the
    right by 8 bits so the upper 8 bits will be used. This will
    allow the graphics adapter to use the most significant bits
    to display the cursor's color.

    ------

    APAR: IY17857 COMPID: 5765C3403 REL: 430
    ABSTRACT: THE GXT2000P MAY NOT COPY THE STENCIL BUFFER.

    PROBLEM DESCRIPTION:
    The GXT2000P graphics adapter may not copy the stencil buffer
    when windows are moved.

    PROBLEM CONCLUSION:
    Update the FRG_SWMASK register so the stencil mask will be
    set correctly.

    ------

    APAR: IY17859 COMPID: 5765C3403 REL: 430
    ABSTRACT: PROBLEMS WITH ARABIC LAYOUT ENGINE

    PROBLEM DESCRIPTION:
    On the CDE login screen the user will be able to type only
    one character in the Arabic locale

    PROBLEM CONCLUSION:
    In TextF.c file of MOTIF1.2, in RedisplayText function we have
    to put a condition for handling of the two cases, 1 byte case
    and 2 bytes case

    ------

    APAR: IY17860 COMPID: 5765C3403 REL: 430
    ABSTRACT: DIAGNOSTIC MEMORY FAULT DISPLAYING TEST RESULTS

    PROBLEM DESCRIPTION:
    Diagnostics memory faults displaying 4 digit SRN test results.

    PROBLEM CONCLUSION:
    Correct the format specification for the 4 digit SRN.

    ------

    APAR: IY17880 COMPID: 5765C3403 REL: 430
    ABSTRACT: SYSTEM CRASH DUE TO BAD THREAD LOCK COUNT

    PROBLEM DESCRIPTION:
    system crash due to t_lockcount being corrupted

    PROBLEM CONCLUSION:
    always update a threads lockcount BEFORE lock release
    SYNC operation to ensure the lockcount is updated before the
    lock is released. read/modify/write lockcount after isync
    in lock acquisition.

    ------

    APAR: IY17881 COMPID: 5765C3403 REL: 430
    ABSTRACT: TCE WRITE WORKAROUND

    PROBLEM DESCRIPTION:
    In rare situations, application failure may occur on the
    following IBM products: 9076 200Mhz Power3 SMP Thin & Wide
    Nodes, Feature Codes 2052 & 2053; 9076 375Mhz Power3 SMP Thin
    & Wide Nodes, Feature Codes 2056 & 2057; 9076 222Mhz Power3
    SMP High Node, Feature Code 2054; 9076 375Mhz Power3 SMP High
    Node, Feature Code 2058; 7043-260, 7044-270, and 7026-B80.
    This failure could, under rare circumstances, result in a
    data error.

    PROBLEM CONCLUSION:
    Eliminates translation invalidation timing problems.

    ------

    APAR: IY17886 COMPID: 5765C3403 REL: 430
    ABSTRACT: REMOVE CFGMGR_PSAFE FOR VERSION 4 OF LP9002

    PROBLEM DESCRIPTION:
    cfgmgr_psafe attribute should not exist on FC adapters
    in Version 4.

    PROBLEM CONCLUSION:
    Remove cfgmgr_psafe attribute for Version 4 FC adapters

    ------

    APAR: IY17887 COMPID: 5765D5100 REL: 320
    ABSTRACT: VALIDITY BIT SET FAILURE

    PROBLEM DESCRIPTION:
    validity bit set failure.

    PROBLEM SUMMARY:
    It can happen that the TOD validity bit can be dropped right
    after it is written. This is due to some freak hardware
    window-of-opportunity. It shouldn't happen often, bit when
    it does, the code will check the bit after of time, and then
    try again. If it fails again, the emasterd will be properly
    notified in any event.

    ------

    APAR: IY17890 COMPID: 5765C3403 REL: 430
    ABSTRACT: WHEN 3LAMPS MODEL IS LOADED ONE OF THE VIEWS IS SHADED

    PROBLEM DESCRIPTION:
    Adapter state is not carried over from view to view.

    PROBLEM CONCLUSION:
    Always make sure the adapter state is set.

    ------

    APAR: IY17900 COMPID: 5765C3403 REL: 430
    ABSTRACT: CHANGE NUD CELL REPORT TO STATISTIC

    PROBLEM DESCRIPTION:
    Error log may fill with HATM_EXCEP errors

    PROBLEM CONCLUSION:
    Eliminate error logging for these problems and count
    statitics for Bad Cells

    ------

    APAR: IY17901 COMPID: 5765C3403 REL: 430
    ABSTRACT: MY_PPDA()->CACHE_NUM IS ALWAYS 0 ON NON-CHRP

    PROBLEM DESCRIPTION:
    mbufs only show up on cpu 0 and not on the other cpus

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * bos.mp 4.3.3.25
    * bos.mp 4.3.3.26
    * bos.mp 4.3.3.27
    * bos.mp 4.3.3.28
    * bos.mp 4.3.3.29
    * bos.mp 4.3.3.30
    * bos.mp 4.3.3.31
    * bos.mp 4.3.3.32
    * bos.mp 4.3.3.50
    * bos.up 4.3.3.25
    * bos.up 4.3.3.26
    * bos.up 4.3.3.27
    * bos.up 4.3.3.28
    * bos.up 4.3.3.29
    * bos.up 4.3.3.30
    * bos.up 4.3.3.31
    * bos.up 4.3.3.32
    * bos.up 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * mbufs only show up on cpu 0 and not on the other cpus
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY17901
    ***************************************************************

    PROBLEM CONCLUSION:
    move the check for __chrp() into the init_hmt() routine
    so that the ppda.cache_num field is always set correctly.

    ------

    APAR: IY17902 COMPID: 5765C3403 REL: 430
    ABSTRACT: FC I/O HANG AFTER SHARK WARMSTART OR CABLE PULL ERROR INJECT

    PROBLEM DESCRIPTION:
    FC I/O may hang on AIX host, especially when DPO with 4 or
    more paths is used.

    PROBLEM CONCLUSION:
    Correctly preserve error recover flags on back-to-back check
    conditions as well as clear ACA flags when non-ACA error
    recovery occurs to insure halted queues are resumed.

    ------

    APAR: IY17903 COMPID: 5765C3403 REL: 430
    ABSTRACT: SYSTEMS LOSE SCREEN WHEN GOING TO HTX ERR LOG

    PROBLEM DESCRIPTION:
    system appears to hang

    PROBLEM CONCLUSION:
    fix interrupt handling

    ------

    APAR: IY17914 COMPID: 5765C3403 REL: 430
    ABSTRACT: OPENGL 64BIT STRESS CAUSES X HANG

    PROBLEM DESCRIPTION:
    OpenGL applications may dump core in tm_RestoreContext on
    the GXT4000P (only) if multiple texture mapping clients
    are running simultaneously and the working texture map sets
    exceed available texture memory.
    Besides dumping core, it is also possible that the texture
    map is not restored thereby resulting in incorrect texels
    being applied to an image.

    PROBLEM CONCLUSION:
    Correct logic that determined which textures needed to be
    restored.

    ------

    APAR: IY17915 COMPID: 5765C3403 REL: 430
    ABSTRACT: TEXTURE MAPPED APPS MAY CORE DUMP OR HANG DISPLAY ON GXT4000P

    PROBLEM DESCRIPTION:
    OpenGL texture mapped applications on the GXT4000P run slow
    and have been shown to cause clients to dump core or lock
    up the display. This problem is not readily reproducable and
    is difficult to characterize.

    PROBLEM CONCLUSION:
    Remove -DTM_DEBUG from the GXT4000P texture manager Makefile.
    This removes internal texture manager state diagnostics
    software that was not intended for ship level product usage.

    ------

    APAR: IY17920 COMPID: 5765C3403 REL: 430
    ABSTRACT: SMIT: ALLOW SPACES IN SHARE-PATH

    PROBLEM DESCRIPTION:
    smit(ty) doesnot handle spaces in path, when adding a
    share

    PROBLEM CONCLUSION:
    need to protect the string

    ------

    APAR: IY17929 COMPID: 5765D5100 REL: 320
    ABSTRACT: GPFS HANG DUE TO REVC FIFO NOT BEING UPDATED

    PROBLEM DESCRIPTION:
    We can get into a dealock situation where the Colony adapter
    microcode thinks the RFIFO is full when it actually isn't.

    PROBLEM SUMMARY:
    The problem is that col_update() asks micro code
    to update the send pointers without check if
    there is avail space in the FiFo or not.
    After the upper level protocol, such as klapi, calls
    col_update(), it may cause the sned head and tail
    in KHAL mismatch these pointers in micro code.

    PROBLEM CONCLUSION:
    Added code to check if there are spaces in the send FiFo
    in col_update().

    ------

    APAR: IY17930 COMPID: 5765C3403 REL: 430
    ABSTRACT: SK-NET FDDI PCI FLASH UPDATE UTILITY FAILED ON REV B

    PROBLEM DESCRIPTION:
    Diagnostics flash utility fails on Rev. B of
    Syskonnect adapters.

    PROBLEM CONCLUSION:
    Update diagnostics utility and driver

    ------

    APAR: IY17932 COMPID: 5765C3403 REL: 430
    ABSTRACT: SECURITY: IMAPD BUFFER OVERFLOW

    PROBLEM DESCRIPTION:
    Buffer overruns can cause coredumps and security issues.

    PROBLEM CONCLUSION:
    Do bounds checking for str* routines and sprintfs. I did not
    change the c-client library as we check bounds in imapd.c.

    ------

    APAR: IY17933 COMPID: 5765C3403 REL: 430
    ABSTRACT: DOWNLOAD MICROCODE TAKS MAY FAIL INNO-CONSOLE MODE

    PROBLEM DESCRIPTION:
    Download Microcode task fails when invoked in no-console
    mode and diskette is specified as the source.

    PROBLEM CONCLUSION:
    Update code to fix the error checking.

    ------

    APAR: IY17944 COMPID: 5765C3403 REL: 430
    ABSTRACT: PORT 996 CONFLICT IN /ETC/SERVICES

    PROBLEM DESCRIPTION:
    Call to name will point to incorrect port.

    PROBLEM CONCLUSION:
    Update services with correct name for port 996.
    Update packaging to correct version migration.

    ------

    APAR: IY17945 COMPID: 5765C3403 REL: 430
    ABSTRACT: MEMORY LEAK RELATED TO STATIC INITIALIZATION OF MUTEX AND CV

    PROBLEM DESCRIPTION:
    popserv core dumps - exceeds 2gb of heap

    PROBLEM CONCLUSION:
    turn off dbx allocations by default.

    ------

    APAR: IY17950 COMPID: 5765C3403 REL: 430
    ABSTRACT: TOKEN OBJECTS NOT PROPERLY CLEANED UP ON INIT

    PROBLEM DESCRIPTION:
    All object references from shared memory should be removed
    when the token is reinitialized.

    PROBLEM CONCLUSION:
    Added an object_mgr_destroy_token_objects() call to
    SC_InitToken to clean up correctly on initialization.

    ------

    APAR: IY17977 COMPID: 5765C3403 REL: 430
    ABSTRACT: FONTUTIL COREDUMP FOR X11.FNT.UCS.TTF PCF FONTS

    PROBLEM DESCRIPTION:
    fontutil dumps core for UCS2 encoded Japanese fonts.

    PROBLEM CONCLUSION:
    Fix to issue an error message for unsupported fonts.

    ------

    APAR: IY17979 COMPID: 5765C3403 REL: 430
    ABSTRACT: REASSOCIATION OF DMTUNNELS FAILS AFTER DELETING KMTUNNEL

    PROBLEM DESCRIPTION:
    Every once in a while, when a phase 1 tunnel is deleted, the
    pointers in the phase 2 tunnels that refer to the remaining
    phase 1 tunnels will not be updated properly.

    PROBLEM CONCLUSION:
    Fix the renumbering code in the guiapi

    ------

    APAR: IY17989 COMPID: 5765C3403 REL: 430
    ABSTRACT: GXT6000P/4000P CODE CLEAN UPS

    PROBLEM DESCRIPTION:
    Unused texture memory.

    PROBLEM CONCLUSION:
    Set the texute memory bank offsets right.

    ------

    APAR: IY17990 COMPID: 576565400 REL: 220
    ABSTRACT: CALCULATIONS FOR KERNCPU, USERCPU, CPUPCT ARE WRONG

    PROBLEM DESCRIPTION:
    kerncpu, usercpu, cpupct values exceed 100%.

    PROBLEM CONCLUSION:
    modified the code to implement the correct
    conversion factor for calculation of the above mentioned stats

    ------

    APAR: IY17992 COMPID: 5765C3403 REL: 430
    ABSTRACT: CAN'T CREATE GENERIC MKSYSB CD BECAUSE /MKCD FILLS UP

    PROBLEM DESCRIPTION:
    User will not be able to create generic mkcd CDs. The
    file systems are not created large enough.

    PROBLEM CONCLUSION:
    Add additional space to file system size to account for space
    taken up by the install images.

    ------

    APAR: IY17994 COMPID: 5765C3403 REL: 430
    ABSTRACT: SECLDAPCLNTD DOES NOT START

    PROBLEM DESCRIPTION:
    On a system configured both as a ldap server and as a ldap
    client, after reboot, the secldapclntd may not be started
    by the init process.

    PROBLEM CONCLUSION:
    Have the secldapclntd try again if the connection to the
    server fails.

    ------

    APAR: IY17995 COMPID: 5765C3403 REL: 430
    ABSTRACT: ALT_DISK_INSTALL FAILS BUT SAYS OK

    PROBLEM DESCRIPTION:
    When in verbose mode, restore errors can scroll off the
    screen.

    PROBLEM CONCLUSION:
    Modify the code to display a warning message at the end
    of the verbose output, if errors occured.

    ------

    APAR: IY18012 COMPID: 576565400 REL: 220
    ABSTRACT: SPMI PROCESS HANDLING LIMTATION

    PROBLEM DESCRIPTION:
    perfagent clients doesn't show up all the processes running
    in the machine

    PROBLEM CONCLUSION:
    Fixed the code to remove the 512 process limitation

    ------

    APAR: IY18017 COMPID: 5765D5100 REL: 320
    ABSTRACT: COLONY: DON'T ERROR LOG FOR PORTS NOT CONFIGED/NO THREAD

    PROBLEM DESCRIPTION:
    COLONY: Don't error log for ports not configed/no thread

    PROBLEM SUMMARY:
    Remove system error logging of adapter errors for non
    configured ports.

    PROBLEM CONCLUSION:
    Adapter errors on port1 were being system error logged. In
    specfic STI errors were occurring on port1 because there was
    no warp plug and both both ports are enabled regardless of
    whether or not they are configured. We want the info in the
    adapter.log for the engineers use but not in the system
    error log.

    ------

    APAR: IY18020 COMPID: 5765D5100 REL: 320
    ABSTRACT: BREAK OUT THE G_WINDOW STRUCTURE OF OF THE COLONY_UCODE.H FILE

    PROBLEM DESCRIPTION:
    break out the g_window structure out of the colony_ucode.h file

    PROBLEM SUMMARY:
    To improve code maintanability, we restructured some code
    flows.

    PROBLEM CONCLUSION:
    To improve code maintanability, we restructured some code
    flows.

    ------

    APAR: IY18028 COMPID: 5765C3403 REL: 430
    ABSTRACT: PERFORMANCE ON NETIF_BACKUP

    PROBLEM DESCRIPTION:
    failover from link status takes too long (i.e. 12-15 sec
    range)

    PROBLEM CONCLUSION:
    failover takes 2-3 sec range

    ------

    APAR: IY18029 COMPID: 5765C3403 REL: 430
    ABSTRACT: PROBLEMS WITH OPENGL RENDERING USING GL_QUAD_STRIP ON SEVERAL

    PROBLEM DESCRIPTION:
    OpenGL rendering using GL_QUAD_STRIP vertex arrays may not
    produce correct results on the GXT500, GXT500D, GXT500P,
    GXT550P, GXT800P, or GXT800M. Vertex arrays include
    glDrawArrays, glDrawElements, glDrawRangeElements,
    glMultDrawArraysEXT, and glMultiDrawElementsEXT. This problem
    should only occur when object coordinates are not required
    for the computation of color, fog, and texture.

    PROBLEM CONCLUSION:
    Don't call transform specialist when ending a quad strip
    vertex array. The buffer vertex data will get flushed by the
    vertex array specialist.

    ------

    APAR: IY18035 COMPID: 5765C3403 REL: 430
    ABSTRACT: BECAUSE MULTI-SEG .INDIRECT TOUCHES MOUNT INODE

    PROBLEM DESCRIPTION:
    system DSI v_makeiblk because the in kernel inode's mounnode
    not in memory.

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * bos.mp 4.3.3.50
    * bos.up 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * system DSI v_makeiblk because the in kernel inode's
    * mounnode not in memory.
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY18035
    ***************************************************************

    PROBLEM CONCLUSION:
    Touch the mountnode first because get into critical section.

    ------

    APAR: IY18041 COMPID: 5765C3403 REL: 430
    ABSTRACT: FTPD CORE DUMPS

    PROBLEM DESCRIPTION:
    Ftpd core dumps just before it finishes, after deleting
    dce credential files.

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * bos.net.tcp.client 4.3.3.28
    * bos.net.tcp.client 4.3.3.29
    * bos.net.tcp.client 4.3.3.30
    * bos.net.tcp.client 4.3.3.31
    * bos.net.tcp.client 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * Ftpd core dumps just before it finishes, after
    * deleting dce credential files.
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY18041
    ***************************************************************

    PROBLEM CONCLUSION:
    Compile ftpd with -lpthreads option, thereby initializing
    the pthreads library before ftpd runs.

    ------

    APAR: IY18051 COMPID: 5765C3403 REL: 430
    ABSTRACT: DRIVE UNCONFIGURED ERRORS AND COSMETIC FIXES

    PROBLEM DESCRIPTION:
    imperfect user interface.

    ------

    APAR: IY18053 COMPID: 5765C3403 REL: 430
    ABSTRACT: MACHINE CHECK OR SYSTEM HANG

    PROBLEM DESCRIPTION:
    Machine check or system hang on rspc models and some oem
    machines.

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * devices.pci.14107c00.com 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * Machine check or system hang on rspc models and some
    * oem machines.
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY18053
    ***************************************************************

    PROBLEM CONCLUSION:
    Change to use string functions instead of word sized compare
    as pointers are being compared instead of data.

    ------

    APAR: IY18068 COMPID: 5765C3403 REL: 430
    ABSTRACT: ICONV -F UTF-8 -T CT CORE DUMP WITH MALLOC DEBUG

    PROBLEM DESCRIPTION:
    iconv from UTF-8 to ct dumps core in Malloc Debug mode.

    PROBLEM CONCLUSION:
    Fix closing functions in the converter between UCS and ct.

    ------

    APAR: IY18069 COMPID: 5765C3403 REL: 430
    ABSTRACT: 64BIT APPS MAKES CORE WITH DBCS LOCALES

    PROBLEM DESCRIPTION:
    64bit apps dumps core while calling XmbResetIC() with Japanese
    IM.

    PROBLEM CONCLUSION:
    Fix Japanese IM code.

    ------

    APAR: IY18070 COMPID: 5765C3403 REL: 430
    ABSTRACT: CFGEFSCSI FAILS TO ADD A DEVICE

    PROBLEM DESCRIPTION:
    After cfgmgr is run to configure the newly added fibre channel
    device, the cfgefscsi module core dumps with a segmentation
    violation, which could be seen in the error log.

    PROBLEM CONCLUSION:
    The fix is to initialize a pointer and free it properly in
    the walk_luns() routine in the find_fcp_dev module used by
    the cfgefscsi FC protocol device config method.

    ------

    APAR: IY18072 COMPID: 5765C3403 REL: 430
    ABSTRACT: README TO INCLUDE NT FILE COPY DATE PROBLEM

    PROBLEM DESCRIPTION:
    Some latest documention is missing

    PROBLEM CONCLUSION:
    Add the new information to README file.

    ------

    APAR: IY18076 COMPID: 5765C3403 REL: 430
    ABSTRACT: OSLEVEL -R RETURNS IX78404

    PROBLEM DESCRIPTION:
    oslevel -r returns IX78404.. or it may return some other
    incorrect string which is not a ML

    PROBLEM CONCLUSION:
    Change oslevel caching to look for *_AIX_ML in a
    specific field.

    ------

    APAR: IY18083 COMPID: 5765C3403 REL: 430
    ABSTRACT: NET USER COMMAND FAILS WITH RC = 1 WITH LARGE # OF USERS

    PROBLEM DESCRIPTION:
    net user cmd cannot list big list of user names.

    PROBLEM CONCLUSION:
    net user cmd will allocate a new bigger buffer to handle
    long list of user names

    ------

    APAR: IY18084 COMPID: 5765C3403 REL: 430
    ABSTRACT: SUPPORT FOR 4 NEW JAPANESE CHARACTERS

    PROBLEM DESCRIPTION:
    Customer can not use these 4 characters for file names.

    PROBLEM CONCLUSION:
    New configuration parameter is added, incoming and outgoing
    characters are checked for these four characters and converted

    ------

    APAR: IY18085 COMPID: 5765C3403 REL: 430
    ABSTRACT: INCORRECT TIME STAMPS WITH LANMAN 2.1

    PROBLEM DESCRIPTION:
    copy a file from client on a mapped drive, creates a file
    with incorrect time stamp

    PROBLEM CONCLUSION:
    set with correct time stamp.

    ------

    APAR: IY18086 COMPID: 5765C3403 REL: 430
    ABSTRACT: DRIVE SELECT FROM FILE OPEN IN EXCEL FAILS

    PROBLEM DESCRIPTION:
    selecting a mapped drive from file open in excel returns
    error

    PROBLEM CONCLUSION:
    server needs to send the file attributes correctly.

    ------

    APAR: IY18090 COMPID: 5765C3403 REL: 430
    ABSTRACT: SYSTEM CRASH DUE TO DOUBLE FREE OF MBUF

    PROBLEM DESCRIPTION:
    Every once in a while, when an invalid packet using both AH
    and ESP encryption comes through, the machine will crash.

    PROBLEM CONCLUSION:
    The alias variable ip_packet_p was removed from
    ipsp_decapsulate() and cap6_decapsulate(); it was the failure
    of the program to properly re-initialize this variable on the
    second pass of the AH/ESP decryption that caused the problem.
    A normal exit would take care of it, but an exit on an error
    path might not.
    APAR IY11357 removed several m_freem() calls that should be
    there; these are put back now. Also cleaned up a couple of
    other memory leaks on error paths in esp_rply_decap4() and
    esp_rply_decap6().

    ------

    APAR: IY18105 COMPID: 5765C3403 REL: 430
    ABSTRACT: OGL STRESS CAUSES SYSTEM BUS TIMEOUT AND CHECKSTOP

    PROBLEM DESCRIPTION:
    A locking mechanism that would synchronize the interrupt
    handler and the FIFO drain to workaround a HW bug has a hole.
    Under the rare case, the interrupt mask might get turned on
    while the FIFO drain was still holding the lock and caused
    the bus timeout.

    PROBLEM CONCLUSION:
    Fixed the DD interrupt handler code to prevent the interrupt
    mask to be enabled before its time.

    ------

    APAR: IY18130 COMPID: 5765B8100 REL: 220
    ABSTRACT: CHANNEL WRAP FOR SS7 IN DIRECTTALK FOR AIX

    PROBLEM DESCRIPTION:
    Channel Wrap for SS7 support for DirectTalk for AIX

    PROBLEM SUMMARY:
    Channel Wrap for SS7 support for DirectTalk
    for AIX

    PROBLEM CONCLUSION:
    Removed original channel wrap implementation
     from trunkE1 because that does not work as we expected, and
     moved the function to the RTTXA and solved the problem using
     the TDM.

    ------

    APAR: IY18174 COMPID: 5765C3403 REL: 430
    ABSTRACT: UMOUNT FAILS WHILE UNMOUNTING FILE SYSTEM CONTAINING ZERO SIZED

    PROBLEM DESCRIPTION:
    umount fails while unmounting a file system which has
    a core file created with zero length.

    PROBLEM SUMMARY:
    The core file is closed before returning an error.

    PROBLEM CONCLUSION:
    The umount fails to unmount the filesystem which core
    file still open.

    ------

    APAR: IY18177 COMPID: 5765C3403 REL: 430
    ABSTRACT: BOS.NET.TCP.SERVER 4.3.3.29 NEED IFREQ TO BOS.RTE.NET 4.3.3.2

    PROBLEM DESCRIPTION:
    bos.net.tcp.server breaks named8 when bos.rte.net
    4.3.3.2 not installed

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * bos.net.tcp.server 4.3.3.27
    * bos.net.tcp.server 4.3.3.28
    * bos.net.tcp.server 4.3.3.29
    * bos.net.tcp.server 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * bos.net.tcp.server breaks named8 when bos.rte.net
    * 4.3.3.2 not installed
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY18177
    ***************************************************************

    PROBLEM CONCLUSION:
    add requisite from bos.net.tcp.server to bos.rte.net 4.3.3.2

    ------

    APAR: IY18241 COMPID: 5765D5100 REL: 320
    ABSTRACT: ADD TIMESTAMP TO COL_DUMP AND MICROCODE

    PROBLEM DESCRIPTION:
    add timestamp to col_dump and microcode

    PROBLEM SUMMARY:
    needed to be able to identify the microcode
    version

    PROBLEM CONCLUSION:
    added to the status structure a date and
    a time stamp. we then print this out during problem
    determination in col_dump

    ------

    APAR: IY18242 COMPID: 5765D5100 REL: 320
    ABSTRACT: COLONY SWITCH-SENDER HANG

    PROBLEM DESCRIPTION:
    colony switch-sender hang

    PROBLEM SUMMARY:
    an inialization of a timing structure was done
    incorrectly. This resulted in faulty navigation through one
    of our data structures.

    PROBLEM CONCLUSION:
    initialized the structure correctly.

    TEMPORARY FIX:
    could only run small jobs on the system.

    ------

    APAR: IY18243 COMPID: 5765D5100 REL: 320
    ABSTRACT: SWITCH SCALING PROBLEM

    PROBLEM DESCRIPTION:
    switch scaling problem

    PROBLEM SUMMARY:
    Primary node times out on acks from
    distribution points during Estart. This can cause Estart to be
    redriven and sometimes to fail.

    PROBLEM CONCLUSION:
    Distribution points were unnecessarily
    generating a set of routes for each leaf node during Estart.
    On a large system this took long enough that the primary node
    timed out waiting for the distribution point to ack. The fix
    for this was to eliminate the unnecessary route generation.

    ------

    APAR: IY18250 COMPID: 5765D5100 REL: 320
    ABSTRACT: ADAPTRECV LATE HW SNAP BREAKS CA ERRORS

    PROBLEM DESCRIPTION:
    adaptrecv late hw snap breaks ca errors

    PROBLEM SUMMARY:
    Critical adapter errors were not recovering
    in time when the CPU load was high. Node was getting fenced.

    PROBLEM CONCLUSION:
    Increase switch recovery timeout to
    insure node does not get fenced.

    ------

    APAR: IY18256 COMPID: 5765D5100 REL: 320
    ABSTRACT: STORE PIDS OF COMPETING PROCESSES IN RC.SWITCH.EXTRA

    PROBLEM DESCRIPTION:
    Store pids of competing processes in rc.switch.extra

    PROBLEM SUMMARY:
    During a system-wide reboot of a large system
    several nodes did not show their Fault Service daemon restarted.
    The inittab runs the rc.switch script that starts the FSD. It
    was noticed that on the nodes where the FSD did not start, the
    file /var/adm/SPlogs/css/rc.switch.extra was created. This file
    is created when rc.switch detects another process, also named
    rc.switch, executing on this node at the same time. The intent
    was to prevent two instances of rc.switch from starting two
    FSD's simultaneously. However, editing (vi, more, ps...) of
    the rc.switch.log can keep the script from executing. It is
    believed that the inittab child process that starts rc.switch
    is being detected in the "ps" command within the rc.switch
    script. This keeps rc.switch from executing and consequently
    the fSD is not started. This seems to happen only occasionally.

    PROBLEM CONCLUSION:
    This change will store the pids and names
    of the processes that keep rc.swtich from running. While it
    will not fix the problem, it will provice additional
    information to prove the theory so the problem can be fixed.

    ------

    APAR: IY18285 COMPID: 5765D5100 REL: 320
    ABSTRACT: RETRIEVE NODE DEVICE CONFIG INFO ON SNAPS

    PROBLEM DESCRIPTION:
    retrieve node device config info on snaps

    PROBLEM SUMMARY:
    The css.snap command needs to provide
    additional information.

    PROBLEM CONCLUSION:
    The css.snap command was modified to include
    output from the "lsatttr -E -l css0" and "lsattr -E -l css1"
    commands.

    ------

    APAR: IY18309 COMPID: 5765D5100 REL: 320
    ABSTRACT: HOLE IN WAY SEND_UPDATE IS DONE FOR HELD DATA

    PROBLEM DESCRIPTION:
    hole in way send_update is done for held data

    PROBLEM SUMMARY:
    Sometimes the microcode would indicate to the
    higher level protocols (KHAL/KLAPI) that a command was
    completed before it was actually completed. This only applied
    to commands, such as move_cluster_buffer, which piggy-backed
    its completion report on top of previously issued commands.
    There is an extremely small window where this could happen.

    PROBLEM CONCLUSION:
    Made sure that the microcode "piggy_backs"
    the completion reports for these special commands on the last
    completion report to be issued for the respective window.

    ------

    APAR: IY18312 COMPID: 5765D5100 REL: 320
    ABSTRACT: FIX FOR VAC COMPILER PROBLEM

    PROBLEM DESCRIPTION:
    fix for vac compiler problem

    PROBLEM SUMMARY:
    The compiler generated code that caused the
    SP Switch2 adapter to read from an illegal address in page 0
    of the address space. If non-AIX environments (i.e. microcode),
    page 0 doesn't always exist.

    PROBLEM CONCLUSION:
    The SP Switch2 microcode will be compiled
    with the -qxflag=npage0 option.

    ------

    APAR: IY18313 COMPID: 5765D5100 REL: 320
    ABSTRACT: TEMP VAR IN CSS.SNAP

    PROBLEM DESCRIPTION:
    temp var in css.snap

    PROBLEM SUMMARY:
    Some errpt data is missing from the css.snap
    file.

    PROBLEM CONCLUSION:
    The "errpt -a" data was directed to the
    wrong file; it will now be directed to the correct file.

    ------

    APAR: IY18314 COMPID: 5765D5100 REL: 320
    ABSTRACT: CSS.SNAP FAILS WHEN /TMP/CSS0 DIRECTORY EXISTS

    PROBLEM DESCRIPTION:
    css.snap fails when /tmp/css0 directory exists

    PROBLEM SUMMARY:
    css.snap fails when /tmp/css0 directory exists.

    PROBLEM CONCLUSION:
    A check was added to the css.snap command
    to see if the /tmp/css0 or /tmp/css1 directories already
    exist before creating them.

    ------

    APAR: IY18316 COMPID: 5765D5100 REL: 320
    ABSTRACT: ADD SYNC AFTER PAAM BUG DCBF

    PROBLEM DESCRIPTION:
    add sync after paam bug dcbf

    PROBLEM SUMMARY:
    Large jobs can cause cache contention which
    requires explicit cache flushes before data areas can be used
    for data transmission. Refine the way protocol code handles
    cache flushing.

    PROBLEM CONCLUSION:
    We refined the way protocol code handles
    cache flushing.

    ------

    APAR: IY18349 COMPID: 5765B9500 REL: 130
    ABSTRACT: BREAK OUT THE G_WINDOW STRUCTURE OF OF THE COLONY_UCODE.H FILE

    PROBLEM DESCRIPTION:
    break out the g_window structure out of the colony_ucode.h file

    PROBLEM SUMMARY:
    To improve code maintanability, we restructured some code
    flows.

    PROBLEM CONCLUSION:
    To improve code maintanability, we restructured some code
    flows.

    ------

    APAR: IY18350 COMPID: 5765D9300 REL: 310
    ABSTRACT: BREAK OUT THE G_WINDOW STRUCTURE OF OF THE COLONY_UCODE.H FILE

    PROBLEM DESCRIPTION:
    break out the g_window structure out of the colony_ucode.h file

    PROBLEM SUMMARY:
    To improve code maintanability, we restructured some code
    flows.

    PROBLEM CONCLUSION:
    To improve code maintanability, we restructured some code
    flows.

    ------

    APAR: IY18394 COMPID: 5765D5100 REL: 320
    ABSTRACT: PERFORMANCE PROBLEM

    PROBLEM DESCRIPTION:
    performance problems

    PROBLEM SUMMARY:
    There is a performance problem after an initial burst of >
    300MB/s on single-single; the performance drops to under 100
    MB/s due to leakage in hal resources.

    PROBLEM CONCLUSION:
    In Khal, change MAX_DATA_BUFS=64 to COL_MAX_DDMA_BUFS=2048.
    In Klapi, change POST_BUF_GRANULARITY from 4096 TO 2048.

    ------

    APAR: IY18482 COMPID: 5765C3403 REL: 430
    ABSTRACT: DSI ON CFGMGR -V AFTER REMOVING ATM ADAPTER

    PROBLEM DESCRIPTION:
    crash after remove an atm adapter in system have
    2 or more atm adapters

    PROBLEM SUMMARY:
    ***************************************************************
    *USERS AFFECTED: *
    * All users with the following filesets at these levels *
    * devices.common.IBM.atm.rte 4.3.3.50
    ***************************************************************
    *PROBLEM DESCRIPTION: *
    * crash after remove an atm adapter in system have 2 or
    * more atm adapters
    ***************************************************************
    *RECOMMENDATION: *
    * Apply apar IY18482
    ***************************************************************

    PROBLEM CONCLUSION:
    pointer to atmcm mapper need to be per adapter

    ------

    APAR: IY18512 COMPID: 576560100 REL: 210
    ABSTRACT: AIXDCE210 MAINTENANCE PACKAGE FOR PTF SET 30

    PROBLEM DESCRIPTION:
    THIS APAR CREATED FOR MAINTENANCE PACKAGE OF PTF SET 30
    aixdce210 Maintenance Package for PTF SET 30

    PROBLEM SUMMARY:
    AIXDCE210 MAINTENANCE PACKAGE FOR PTF SET 30

    PROBLEM CONCLUSION:
    AIXDCE210 MAINTENANCE PACKAGE FOR PTF SET 30

    ------

    APAR: IY18551 COMPID: 5765C3403 REL: 430
    ABSTRACT: FULL WHEEL MOUSE ENABLEMENT

    PROBLEM DESCRIPTION:
    Provide full Wheel Mouse Enablement - this includes the
    fixes for IY14962 and IY17815.

    PROBLEM CONCLUSION:
    Ship the fixes for IY14962 and IY17815 under one apar.

    ------