PH61398: CICS REGION HUNG NON RESPONSIVE TASK IN LOOP ON QR TCB INQUIRE_TASK_STORAGE KERNDUMP

Obtain the fix for this APAR.

APAR status

• Closed as unreproducible in next release.

Error description

• You are running CICS TS 5.5 and the region suddenly becomes
  unresponsive.  The region is running with Fault Analyzer and
  happens to be processing transaction abend codes for
  dumpcode EC4120. In the joblog you see the following messages:
  +DFHKE0201 ABOUT TO TAKE SDUMP. DUMPCODE: KERNDUMP,
             DUMPID: 0/0000. (MODULE DFHKESTX).
  +DFHKE0201 ABOUT TO TAKE SDUMP. DUMPCODE: KERNDUMP,
             DUMPID: 0/0000. (MODULE DFHKESTX).
  +DFHKE0201 ABOUT TO TAKE SDUMP. DUMPCODE: KERNDUMP,
             DUMPID: 0/0000. (MODULE DFHKESTX).
  BPXP018I THREAD 1F72100000000000, IN PROCESS 1332, ENDED
  WITHOUT BEING UNDUBBED WITH COMPLETION CODE 940C4000
  
  +DFHME0118 An error has occurred when calling the Message User
             Exit for message DFHSJ0904
  +DFHME0118 An error has occurred when calling the Message User
             Exit for message DFHAP0001
  BPXP018I THREAD 1F9308000010331E, IN PROCESS 16777931, ENDED
  WITHOUT BEING UNDUBBED WITH COMPLETION CODE 0033E000
  
  +IDI0095W Unexpected condition found in IDI:347: STATIC
            INVOCATION OF EXIT NOT SUPPORTED
  
  Looking at a KERNDUMP, found a task, running on the QR TCB,
  looping overlaying CICS DSA storage above the line.
  The task had abended, and Fault Anayzer was in in control
  gathering information about the abend.  CICS internal
  trace showed it issued and INQUIRE_TASK_STORAGE call to CICS.
  CICS keeps track of
  task storage for each individual task in a SMX control block,
  anchored off the Storage Manager Anchor Block.
  FA was inquiring on a task that Just recently ended.
  
  Here is that trace:
  
  QR SM 0E01 SMMC2 ENTRY INQUIRE_TASK_STORAGE
        TRANSACTION_NUMBER(0085979C)
        ELEMENT_BUFFER(27100008 , 0012E4F0  , 0012E4F0)
        LENGTH_BUFFER(28500008 , 0012E4F0 , 0012E4F0)
  
  Task 85979 recently terminated, but Storage manager had not
  completely finished its
  clean up of control blocks for this task. This caused code in
  DFHSMMC2 to reference
  low core - address 0000000, thinking it was chaning through CICS
  control blocks,
  This was updating the buffer storage - 27100008 with information
  from low core, which
  happened to be the CVT address.  This overlaid many CICS control
  blocks, as well as
  causing storage violations.
  
  Here is an example of this overlay, which happens to be the
  address of the CVT- 00FD6380:
  27100000   E4F0F0F8   F5F9F2F9   00FD6380   00FD6380
  27100010   00FD6380   00FD6380   00FD6380   00FD6380
  27100020.:2743D43F. LENGTH(X'33D420')--Same as above
  Additional Symptoms: KIXREVDAM
  quickcell SM0102 DFHSM0102 DFHEC4120
  

Local fix

• disable Fault Analyzer
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICS users.                              *
  ****************************************************************
  * PROBLEM DESCRIPTION: Corruption of storage during            *
  *                      DFHSMMCX XPI (INQUIRE_TASK_STORAGE)     *
  *                      processing.                             *
  ****************************************************************
  The DFHSMMCX (INQUIRE_TASK_STORAGE) TRANSACTION_NUMBER()
  is called to inquire upon a task which has not got
  any task storage subpools (either because they have not
  yet been allocated, or they have been de-allocated).
  The INQUIRE_TASK_STORAGE mechanism returns invalid pointers
  to the caller, corrupting storage.
  In the reported case, many instances of a fullword which
  contained the address of the CVT were returned.
  

Problem conclusion

Temporary fix

Comments

• DFHSMMCX (INQUIRE_TASK_STORAGE) has been changed to
  ignore subpools which are not allocated.
  .
  IBM Publications for CICS, will be updated at
  section:"The INQUIRE_TASK_STORAGE call",
  operand: NUMBER_OF_ELEMENTS
  to state:
  "NUMBER_OF_ELEMENTS can be zero."
  

APAR Information

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

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

  UI97311 UI97312

Modules/Macros

• DFHSMMC2
  

Fix information

• Fixed component name

  CICS TS Z/OS V5

• Fixed component ID

  5655Y0400

Applicable component levels

• R200 PSY UI97312

     UP24/06/14 P F406

• R300 PSY UI97311

     UP24/06/14 P F406

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.