PK69023: RECEIVING ABEND 0C4 IN DFHTRPX AT OFFSET X'306' WHEN ISSUE CEMT SET FECONNECTION RELEASED COMMAND

APAR status

• Closed as returned (APAR cannot be resolved without additional info from IBM or customer.)

Error description

• You issue a CEMT Set FECONNECTION RELEASED command,
  receive the following messages in the log,
  DFHTR0112 Bad data passed for tracing to
            module DFHTRPT.
  DFHME0116 (Module:DFHMEME) CICS symptom string for message
            DFHTR0112 is  PIDS/5655M1500 LVLS/640 MS/DFHTR0112
            RIDS/DFHTRPT PTFS/HCI6400
  DFHDU0201 ABOUT TO TAKE SDUMP. DUMPCODE: TR0112
  .
  DFHAP0001 An abend (code 0C4/AKEA) has occurred at
            offset X'FFFFFFFF' in module DFHSZRDP.
  DFHPC0409 Abends ASRA and ASRA have been issued while processing
            abend ASRA for the same task, transaction CSZI.
  .
  KE domain shows tranid CSZI is running with error
  ==KE: KE Domain Error Table Summary
  RROR TYPE    ERR_CODE  MODULE    OFFSET
  =========    ========  ======    ======
  ROGRAM_CHECK 0C4/AKEA  DFHTRPX 00000306
  .
  The stacks are as follows:
        DFHKETA
        DFHDSKE
        DFHXMTA
        UNKNOWN
        UNKNOWN
        DFHSZSIP
        DFHSZRDP
  *YES* DFHTRPT
        DFHMEME
        SEND
        CONTINUE_SEND
        TAKE_A_DUMP_FOR_CALLER
        DFHDUDU
        SYSTEM_DUMP
        TAKE_SYSTEM_DUMP
  .
   Error happened in program DFHTRPX  at offset 00000306
   Error happened under the CICS RB.
   CICS Registers and PSW.
     PSW:  078D3000 9C13B416   Instruction Length:  2
                               Interrupt Code:  11
                               Exception Address:  00797000
     Execution key at Program Check/Abend: 8
     Space at Program Check/Abend: Basespace
     Branch Event Address: 1C13B38A
    r0 - r3   1C30BF30 1E050980 000001CB 9C13B110
    r4 - r7   1F6E3500 1C30BF30 1CC19B60 0000061E
    r8 - r11  00000023 00000453 1CC19B8A 00000023
    r12 - r15 1CEC8080 1DFD30B0 00797FD6 00000023
  .
  The failing instruction is 0EAE/doing MVCL r10,r14
  which fails when it encounters a "fetch protect" accessing data
  at 00797FD6 (key '88' storage).
  .
  The trace shows we received a BIND before it was abended,
  1476 SZ2SB ENTRY FEPIRM-2SB SLU2 Bind proc
  0102 TRPT  *EXC* - INVALID-DATA-PASSED : CALLING DOMAIN WAS AP :
                     PARAMETER LIST WAS FUNCTION(TRACE_PUT)
                     POINT_ID(1476) DATA1(1F6E3500 , 0000017C)
                     DATA2(00797FD6 , 00000023)
  0102 TRPT  *EXC* - INVALID-DATA-PASSED : CALLING DOMAIN WAS AP :
                     PARAMETER LIST WAS FUNCTION(TRACE_PUT)
                     POINT_ID(1476) DATA1(1F6E3500 , 0000017C)
                     DATA2(00797FD6 , 00000023)
  The problem seems to be related to the receipt of an externally
  generated BIND request from the target system around the same
  time as the FEPI connection is being manually released using
  CEMT.
  It could be that the target system is trying to automatically
  reconnect the session to FEPI when it detects that it has been
  lost.
  CICS log shows the following messages for this FEPI node,
  DFHSZ4201 I date time applid FEPI node nnnnnnnn now has
              status GOINGOUT, RELEASING.
  DFHZC3464 I date time applid termid tranid Node netname
              released by MT Operator/LU Services Manager.
  DFHZC3462 I date time applid termid tranid Node netname
              session terminated.((2) Module name: DFHZCLS)
  DFHSZ4201 I date time applid FEPI node nnnnnnnn now has
              status INSERVICE, RELEASING.
  5 minutes later,
  DFHZC2403 E date time applid termid CSNE Bind failure.
             VTAM RETURN CODE 1001 SENSE CODE 0801 0000
             ((1) Module       DFHZSYX)
  DFHZC3437 I date time applid termid CSNE Node nnnnnnnn
              action taken: NOCREATE CLSDST ABTASK ABSEND
              ABRECV ((name: DFHZNAC)
  DFHZC3462 I date time applid termid CSNE Node nnnnnnnn
              session terminated.  ((2) Module name: DFHZCLS)
  
  DFHZC3437 I date time applid termid CSNE Node nnnnnnnn
              action taken: NOCREATE CLSDST ABTASK ABSEND
              ABRECV ((1) Mod name: DFHZNAC)
  DFHZC3462 I date time applid termid CSNE Node nnnnnnnn
              session terminated.  ((2) Module name: DFHZCLS)
  DFHSZ4155 W date time applid session in pool pppppppp to
              target tttttttt and node nnnnnnnn ended with
              a reason of X'00000000'.
  Additional Symptom(s) Search Keyword(s): KIXREVRJS
  

Local fix

Problem summary

Problem conclusion

Temporary fix

Comments

• This APAR is being returned because the supplied documentation
  is insufficient to resolve the problem.
  
  The reported problem is an ABEND 0C4 in DFHTRPX at offset X'306'
  after a CEMT SET FECONNECTION RELEASED COMMAND. Internally FEPI
  issued a TERMSESS OPTCD=UNBIND command out of DFHSZ2TE to
  release the connection. However at the same time FEPI was
  processing an externally generated BIND command for the same
  connection which resulted in SZD_CD_BINDR and SZD_CD_PARMSESS
  being set.
  After processing the TERMSESS the FEPI session controller
  DFHSZ2SC found SZD_CD_BINDR set and called DFHSZ2SB to process
  the incoming BIND. However the TERMSESS had already unbound the
  session and SZD_CD_PARMSESS was no longer valid.
  The trace shows three consecutive attempts to write out an
  AP 1476 SZ2SB ENTRY FEPIRM-2SB SLU2 Bind proc trace record, each
  of which attempted to trace out the storage pointed to by
  SZD_CD_PARMSESS. On the third attempt the storage was no longer
  accessible resulting in the reported ABEND 0C4.
  The supplied documentation was insufficient to fully diagnose
  the problem, specifically:
  1. How it was possible for an externally generated BIND request
  to be received when FEPI was still in the process of UNBINDING
  the existing session.
  2. How the AP 1476 SZ2SB ENTRY trace was repeated without any
  other intervening trace entries.
  
  A new APAR with the following documentation should be raised:-
  
  All error dumps generated for the ABEND.
  For the period from before the  CEMT SET FECONNECTION RELEASED C
  to the ABEND.
  - CICS auxiliary trace (SZ component tracing must be active)
  - CICS Message logs.
  - VTAM buffer trace for the session between one of the failing
    FEPI nodes and the target system.
  
  APAR PK71890 corrected FEPI TERMSESS OPTCD=UNBIND processing and
  should be applied before any documentation is collected.
  
  Please refer to information APAR II02709 for guidelines on how
  to submit CICS APAR documentation on tape.
  

APAR Information

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

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

Fix information

Applicable component levels