IBM Support

PM92185: DEVELOPMENT APAR

A fix is available

Subscribe

You can track all active APARs for this component.

 

APAR status

  • Closed as unreproducible in next release.

Error description

  • Devlopment APAR
    

Local fix

Problem summary

  • ****************************************************************
    * USERS AFFECTED:                                              *
    * V2R1 Users                                                   *
    * E2827/K                                                      *
    ****************************************************************
    * PROBLEM DESCRIPTION:                                         *
    * (16238) R130618O AB/S00C4 0011 TCPSVT V2R1 : Job(GEATTA1 )   *
    * EZBTCUTL +0128D4                                             *
    * (15369) SNA messages needed for QP failure                   *
    * (16271) R130620O AB/S00C4 0011 EZBITTUB(HIP6210              *
    * 13.064)+000576                                               *
    * (16303) R130621O Connections link inactive due to internal   *
    * errors.                                                      *
    * (16305) R130623O AB/S0C4 EZBTLRDQ +B9A FTPATT Sweden         *
    * (16289) AB/S080 6B000121 TCPSVT on shutdown                  *
    * (16376) Performance enhancement                              *
    * (16473) R130625Q AB/S00C4 11 Job(SMONITOR) EZBNMGTC+0012EC   *
    * (15845) ZZ020531 MP65 GA2 RoCE - AB/S00C4 010 ISTSRIQ2       *
    * (16472) R130625Q AB/S00C4 0038 Job(TNPROCC ) EZBSRUTL+007E48 *
    * (16503) R130627O + Constr CSM AB/4C5 0402 EZBSRUTL+00C8D6    *
    * S4C5/743004                                                  *
    * (16486)  sysplex distributor optlocal connection uses new    *
    * link                                                         *
    * (16506) AB/S04c5 402 EZBSRUTL(BOOTSNT 13.177)+005338         *
    * (16500) R130627O + Constr CSM AB/4c5 2001 EZBITKOB+000752    *
    * S4C5/74F520                                                  *
    ****************************************************************
    * RECOMMENDATION:                                              *
    * Apply Fix                                                    *
    ****************************************************************
    (16238) An abend may occur when running link connections due to
    storage for a TCP connection control block being freed
    prematurely.
    (15369) Failover can occur to another Queue Pair on the same or
    a different adapter without any alert to the operator.
    (16271) EZBSRUTL deleted a link while a Test Link timer was
    still running for that link. This caused the TID to be deleted
    prematurely.
    (16303) Alternate link was never being freed up when the LG was
    being terminated. The alternate link was left in a state
    awaiting a CM_Setup_Cnf response from VTAM and therefore the LG
    could not accept new requests for TCP connection to utilize
    links.
    (16305) EZBTLRDQ was called to process SSL data but the
    TLSX_recv_q was empty.
    EZBTCRD did not have the inbound data pointer set correctly
    after a link decline was received.
    (16289) The stack terminates with fixed private storage
    allocated with the DETACHFIXED(NO) option. This causes the
    ABEND080 during termination.
    (16376) Performance affected by unnecessary cursor updates.
    (16473) The NMI code does not properly handle the case where
    link data arrives before the link Confirm message is received
    from the partner. This can lead to an ABEND0C4 in EZBNMGTC.
    (15845) ZZ020531 MP65 GA2 - AB/S00C4 010 ISTSRIQ2
    (16472) EZBSRLLC processed an inbound CRK signal which was not
    part of an active CRK exchange and this caused a corrupted
    control block chain.
    (16503) Incorrectly trigger an ?itAbend for a valid scenario
    where connections are lost due to remote TCP/IP stack
    termination
    (16486) The stack incorrectly establishes a new link connection
    when using DRVIPAs and OPTLOCAL when the connection could use
    fast local sockets.
    (16506) The stack did not cancel a timer when a connection fell
    back from new link to TCP. This could lead to an abend in
    EZBSRUTL.
    (16500) The stack does not correctly clear a pointer to a
    control block when the control block is freed. This can cause
    abends when the pointer is used after the control block has been
    freed.
    

Problem conclusion

  • (16238) The code will be changed to ensure the TCP connection
    control block storage is not freed prematurely.
    (15369)  Issue new message group IST2406I when a Queue Pair
    failover occurs for reasons other than the deactivation or
    inoperative condition of an interface.
    (16271) Change EZBSRUTL to ensure the Test Link timer is
    cancelled before deleting a the link.
    (16303) Change EZBSRUTLto correctly alter the FSM state when a
    CM_Setup_Cnf indicates that a failure was detected. If the FSM
    state is not altered then the code added to EZBSRCRQ an
    Free_..._Link will not allow for the link termination. Since the
    link state indicates that we are still awaiting a CM_Setup_Cnf
    !!!
    (16305) EZBTCRD is updated to reset the inbound pointer if link
    negotiation completes and AT-TLS is started on the connection.
    (16289) The storage allocation for private and staging buffers
    was changed to DETACHFIXED(NO) because we MUST deregister
    staging buffers before freeing them. There may be fixed private
    buffers in the storage caches when the stack terminates. These
    buffers should be allowed to be detached fixed since they are
    not being registered to the interface.
    (16376) Code has been modified to prevent the unnecessary cursor
    updates and improve CPU utilization.
    (16473) The code in EZBNMGTC was taking an else leg invalidly
    when there was data in the storage but the tcb_Status was still
    zero. This can happen when link data beats the Confirm message
    back from the partner. A DO-END will be added around the link
    section and the Status check will be changed to check tcb_smccb@
    to handle the case when link data beats the Confirm.
    (15845) Force Close processing has been modified to "age" the
    64-bit control blocks anchored off the PFCTE.  These control
    blocks use to be immediately freed during Close or Force Close
    processing.  During Force Close processing, outbound (ISTSRIWI)
    or inbound (ISTSRIDQ) SRBs may be active which reference these
    control blocks.  Aging these control blocks prior to being freed
    allows time for these SRBs to exit prior to the storage being
    freed.
    (16472) Change EZBSRLLC to properly validate that we are in the
    correct state for an inbound CRK signal and reset appropriate
    fields when a CRK exchange completes.
    (16503) Change EZBSRUTL to allow for cases where a remote
    storage does not exist for an existing TCP connection. These
    connections will be terminated during fail-over processing.
    (16486) The  generated the new option in the SYN packets in the
    case where fast local sockets could have been used for DRVIPA
    connections using OPTLOCAL. We should use fast local sockets in
    this case and not send the option. Changes are made in tcsndint
    and t6sndint to strip the option from the SYN packets if the
    tcb_DRVIPA_FLS flag is on.
    (16506) When we were stripping the option from the SYN/SYN-ACK
    flows we were not cancelling the connection setup timer. If this
    timer was still running when a FIN was received it caused
    EZBTCRD to call EZBSRUSC even though the connection was not
    using the new link. The timer will be cancelled when we strip
    the option from the SYN/SYN-ACK flows.
    (16500) The problem is that the smcrmbbr_conn_index_smccb@ field
    is pointing to a freed cb. I found two places in ezbscutl where
    we may free the cb without calling ezbsrusc to clear the
    smcrmbbr_conn_index_smccb@ field. Both of these places will
    occur when we are CSM constrained. I also restructured ezbscrcr
    so that it will not reference the  control blocks without
    holding the  lock.
    

Temporary fix

Comments

APAR Information

  • APAR number

    PM92185

  • Reported component name

    TCP/IP V3 MVS

  • Reported component ID

    5655HAL00

  • Reported release

    210

  • Status

    CLOSED UR1

  • PE

    NoPE

  • HIPER

    NoHIPER

  • Special Attention

    NoSpecatt

  • Submitted date

    2013-07-01

  • Closed date

    2013-07-12

  • Last modified date

    2013-08-19

  • APAR is sysrouted FROM one or more of the following:

  • APAR is sysrouted TO one or more of the following:

    UK95779

Modules/Macros

  • EZBSRWCX EZBCTFME EZBSRUTL EZBTCFWR EZBNMGTC EZBTCRD  EZBT6SND
    EZBTLFWR EZBTCICT EZBSCINB EZBSROUT EZBTCPTM EZBTCSND EZBSCUTL
    EZBSRINB EZBTCUTL EZBT6FWR EZBTCSTR EZBTSFWR EZBITSTO EZBSRLLC
    EZBIFSTC EZBT6LWR EZBTCRDG EZBTCFSH
    

Fix information

  • Fixed component name

    TCP/IP V3 MVS

  • Fixed component ID

    5655HAL00

Applicable component levels

  • R210 PSY UK95779

       UP13/07/26 P F307

Fix is available

  • Select the PTF appropriate for your component level. You will be required to sign in. Distribution on physical media is not available in all countries.

[{"Business Unit":{"code":"BU054","label":"Systems w\/TPS"},"Product":{"code":"SG19M","label":"APARs - z\/OS environment"},"Component":"","ARM Category":[],"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"210","Edition":"","Line of Business":{"code":"","label":""}},{"Business Unit":{"code":"BU054","label":"Systems w\/TPS"},"Product":{"code":"SSCY4DZ","label":"DO NOT USE"},"Component":"","ARM Category":[],"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"210","Edition":"","Line of Business":{"code":"","label":""}}]

Document Information

Modified date:
19 August 2013