PI97025: DFHMQ0370 DFHMQMON DFHMQERR REGISTER 13 CSA RSA

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• The customer received the following messages
  .
  DFHMQ0370 MQMONITORS ARE CURRENTLY RESTRICTED
          TO 60 MQGETS PER SECOND.
  DFHMN0115I CICS SERVER Z/OS WLM HEALTH PERCENTAGE IS NOW 25.
  DFHKE0201  ABOUT TO TAKE SDUMP.  DUMPCODE: KERNDUMP
  +.........EBDAA10....".0.".0.".0.".0.".0.".0
  BPXP018I ... ENDED WITHOUT BEING UNDUBBED
  .
  The BPX message indicates that CICS terminated abnormally
  without being able to shutdown any processes.
  .
    The KE Error Table in the dump showed the following error:
  .
  Error for KE_NUM: 0068  Task Num: 00037
  Error Code: 0C4/AKEA   Error Type: PROGRAM CHECK
  CICS Error Instruction Address in Module DFHKELRA +x'0232'
      Instruction Length: 6
      Interrupt Code: 0010
      Exception Address: 3E2B2400
  CICS Registers:
  R0:  1E1FE000  Trace Domain Anchor Block + X'0'
  R1:  211D0C00  KE STACK Entry
  R2:  1EC04788_1EC05788  Storage invalid
  R3:  211D04B0  KE STACK Entry
  R4:  1F6AD000  Kernel Task Entry  + X'0'
      for TaskNum=0066 TranNum=00039
  R5:  1ED29A5E  Module DFHZCX + X'55E'
  R6:  9ED29BA8  Module DFHZCX + X'6A8'
  R7:  211D04B0  KE STACK Entry
  R8:  1ED2AA5E  Module DFHZCX + X'155E'
  R9:  00000100
  R10: 00000000
  R11: 1ED2BA5E  Module DFHZCX + X'255E'
  R12: 1E3DE800  Task 00039 User TCA + X'0'
  R13: 00041800  Module DFHCSA + X'800'
  R14: 9ED2B6D4  Module DFHZCX + X'21D4'
  R15: 1F6AD018  Kernel Task Entry  + X'18'
      for TaskNum=0066 TranNum=00039
  .
    The error was for TASK#37 on KE#68,
  but the registers were for TASK#39 on KE#66
    While the program check was in module DFHKELRA
  the System Trace showed that the SVC13 (SVCD SVC D 13)
  was issued by module DFHZLOC.
    The CICS internal trace showed Task#37 in MQ processing.
  The return registers pointed into module DFHMQMON in
  proc(csErrMsgParm) where it called module DFHMQERR.
    Task #39 include terminal processing
  .
    DFHMQMON changes Register 13 to point at either the CSA
  or the DFHEISTG.  It enters the proc Process_mqmonitors
  which is new in 5.4, dealing with WLM Health.
    That proc switches R13 to point to the CSA
  and makes some domain calls.
    It then does the call to MsgRtn which calls DFHMQERR,
  with R13 pointing to the CSA. This is the problem.
    DFHMQERR saves his caller's registers
  using the caller's R13 - the CSA.
    It then acquires DFHEISTG storage
  and follows a similar pattern of swapping R13.
  Then it does a domain call,
  which leads to the task being suspended.
    Task 39 resumes processing and picks up Task 37's registers
  from the CSA.  DFHZLOC does a BASR to DFHKELRA,
  which program-checks because it is running
  with the wrong set of registers.
    While the problem was in DFHZLOC & DFHKELRA this time,
  the error could happen to any program that picks up the
  wrong registers.  So, the symptoms of this error
  could be unpredicable results.
  .
  Additional Symptom(s) Search Keyword(s): KIXREVxxx
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICS users                               *
  ****************************************************************
  * PROBLEM DESCRIPTION: Abend S0C4 due to invalid registers     *
  *                      restored from DFHMQMON                  *
  ****************************************************************
  CICS starts up with MQCONN=YES and begins to activate a
  connection to MQ.
  The code in DFHMQMON calls DFHMQERR with R13 pointing
  incorrectly to the CSA.
  This causes a misuse of the CSA's Register Save Area which can
  lead to unpredictable results if the CSA RSA is used by another
  task.
  It is possible that DFHMQMON will abend because it has picked
  up another task's registers.
  It is possible that any task which picks up DFHMQMON's
  registers will abend.
  Additional Keywords: STARTUP DFHMQ0370 MSGDFHMQ0370
                       Process_Mqmonitors Msg_Rtn WLM HEALTH
  

Problem conclusion

• DFHMQMON has been changed to point R13 to the correct
  register save area.
  

Temporary fix

Comments

APAR Information

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

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

  UI56839

Modules/Macros

• DFHMQMON
  

Fix information

• Fixed component name

  CICS TS Z/OS V5

• Fixed component ID

  5655Y0400

Applicable component levels

• R100 PSY UI56839

     UP18/06/27 P F806

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.