PM91423: ABEND0C6 DUE TO DOUBLE USE OF IMS SAVEAREA BY ODBA THREAD. EOT PROCESSSING OCCURS PRIOR TO TERMINATING SYNCHPOINT PROCESS

A fix is available

Subscribe

You can track all active APARs for this component.

APAR status

  • Closed as program error.

Error description

  • ABEND0C6 receieved on ODBA ITASK while trying to return
    using a save area that is already marked as returned with
    the low order bit being set on.
    .
    This abend occurs during terminating synchpoint processing
    after the backout exit is driven. Backout processing completes,
    at which time EOT processing kicks off and completes prior
    to terminating synchpoint processing.  Terminating synchpoint
    cleanup receives the ABEND0C6.
    

Local fix

Problem summary

  • ****************************************************************
    * USERS AFFECTED: All users of IMS v11 ODBA                    *
    ****************************************************************
    * PROBLEM DESCRIPTION: Double use of savearea during           *
    *                      terminating syncpoint may lead to       *
    *                      ABEND0C6 or other symptoms.             *
    ****************************************************************
    * RECOMMENDATION: INSTALL CORRECTIVE SERVICE FOR APAR/PTF      *
    ****************************************************************
    In an ODBA environment with RRS enabled, there are two separate
    code paths that are driven if a thread or dependent region
    terminates abnormally.  There is a timing issue where these code
    paths can interfere with one another, causing the double use of
    a save area.  This double use of a save area can lead to a
    variety of external symptoms.  In the reported case, the
    customer encounters an ABEND0C6 when a module attempts to return
    to its caller using a return address for which the low order bit
    has already been set -- the branch target is invalid because it
    is not on a halfword boundary.
    
    Code path 1:  RRS detects that the dependent region address
    space is coming down and drives DFSRRSI0, which calls DFSAERN0
    to process a terminating syncpoint.  DFSAERN0 calls modules to
    process the syncpoint, then clean up the thread.  It then
    responds to RRS for post deferred UR processing.
    
    Code path 2:  DFSSDA20 is driven by a terminate identify AWE.
    This module performs serveral tasks needed to clean up dependent
    region control blocks in the event of abnormal termination.
    This includes cleaning up the thread.
    
    The problem comes as a result of both code paths attempting to
    clean up the thread.  There is a flag, LCRTODBA, defined in the
    LCRE that is intended to prevent this from happening.  Each code
    path sets the flag to indicate that it is handling thread
    cleanup, and will not drive thread cleanup if the other code
    path has already set the flag.  The defect exists because there
    is a way for the DFSRRSI0 path to drive thread cleanup without
    setting LCRTODBA.
    

Problem conclusion

  • GEN:
    KEYWORDS:
    
    *** END IMS KEYWORDS ***
    In the LCRE, a new flag is defined and an existing flag is
    moved so that the DFSRRSE0 and DFSSDA20 processing can be
    better coordinated.  DFSRRSI0 is changed because it uses the
    flag that is moved.  DFSSDA20 contains more extensive changes
    to the way it handles abnormal ODBA thread termination.
    
    Module changes
    --------------
    DFSLCRE - A new flag byte, LCREF5, is defined using existing
    reserved space in the control block.  The existing flag
    LCRTODBA is moved from LCREF1 to LCREF5.  A new flag,
    LCRTTRM1, is defined in LCREF5.  This new flag is set by
    DFSSDA20 the first time it is called to process a
    terminating syncpoint for the LCRE in question.  This
    allows DFSSDA20, when it is re-driven for PST cleanup, to
    detect that it has been called a second time.
    
    The reason LCRTODBA is moved is that during DFSSDA20
    terminating syncpoint processing, the flags in LCREF1
    through LCREF4 are cleared.  This creates a timing window
    where DFSRRSE0 can check LCRTODBA after it has already been
    cleared, resulting in DFSRRSE0 attempting to drive thread
    cleanup even though DFSSDA20 is already doing so.  LCREF5
    is not affected by this, so moving the flag removes this
    timing exposure.
    
    DFSRRSI0 - This module checks and sets LCRTODBA as part of
    terminating syncpoint processing.  This flag is moved from
    LCREF1 to LCREF5, so DFSRRSI0 is changed to reflect this.
    
    DFSSDA20 - Multiple changes are made for terminating
    syncpoint processing.  References to LCRTODBA are changed
    to use LCREF5 instead of LCREF1.  On the second pass through
    DFSSDA20, if LCRTTRM1 has been set, skip checking LCRTODBA
    and go directly to label SDA20550, where CLEANPST will be
    called.  The tracing call at SDA20550 is moved to label
    TMSETUP so that the trace will be written only once, during
    the first pass.  It will no longer be written a second time
    during the second pass through DFSSDA20.
    
    Additional keywords:  ABEND0C6
    

Temporary fix

  • *********
    * HIPER *
    *********
    

Comments

APAR Information

  • APAR number

    PM91423

  • Reported component name

    IMS V11

  • Reported component ID

    5635A0200

  • Reported release

    100

  • Status

    CLOSED PER

  • PE

    NoPE

  • HIPER

    YesHIPER

  • Special Attention

    NoSpecatt

  • Submitted date

    2013-06-19

  • Closed date

    2014-10-10

  • Last modified date

    2014-10-15

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

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

    PI13989 PI13990 UI22128

Modules/Macros

  •    DFSLCRE  DFSRRSI0 DFSSDA20
    

Fix information

  • Fixed component name

    IMS V11

  • Fixed component ID

    5635A0200

Applicable component levels

  • R100 PSY UI22128

       UP14/10/15 I 1000

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.



Rate this page:

(0 users)Average rating

Document information


More support for:

IMS

Software version:

100

Reference #:

PM91423

Modified date:

2014-10-15

Translate my page

Machine Translation

Content navigation