PK76043: TCP TASK LOOPS AT SHUTDOWN WITH KERNEL STACK POINTING TO ITSELF.DFHMQDSC CALLS DFHMQERR WITH REG13 REG SAVE AREA ADDRESSING CSA

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• CICS region loops at shutdown. The looping task is the TCP task
  due to Kernel stacks that end up pointing to themselves. The
  problem is in DFHMQDSC due to Reg13 addressing the CSA on a call
  to the MQ Error message builder module, DFHMQERR.
  .
  The 2 tasks involved are TCP and the CKAM task. DFHMQDSC is
  running under the CKAM task. In the DELETE_OPEN_TCBS proc, Reg13
  is set up to address the CSA (previously Reg13 was addressing
  DFHEISTG storage). The message routine is called to put out
  message:
  .
  DFHMQ0334 Adapter shutdown successful
  .
  This involves a call to DFHMQERR to put out the message. On the
  call to DFHMQERR, R13 is still addressing the CSA. DFHMQERR then
  saves the registers in the CSA register save area. However,
  during processing of the DFHMQ0334 message, CKAM is suspended on
  the TDIPLOCK. While CKAM is suspended, the TCP task runs and the
  CSA register save area is used on behalf of the TCP task.
  .
  When the CKAM task is resumed, it completes it's message
  processing and attempts to return to DFHMQDSC. However, as the
  registers are now associated with the TCP task, processing now
  goes awry.
  
  Additional Symptom(s) Search Keywords(s): KIXREVRER
  .
  DFHZRAC LIFO stack back pointer points to itself.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: All                                          *
  ****************************************************************
  * PROBLEM DESCRIPTION: The TCP task loops at shutdown, with    *
  *                      the kernel stack pointing to itself.    *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  DFHMQDSC routines delete_open_tcbs and
  write_uss_stats_to_smf_and_reset fail to reset R13 to address
  the EISTG upon exit. When DFHMQDSC calls DFHMQERR to issue
  message DFHMQ0334, register 13 is still addressing the CSA.
  DFHMQERR saves its registers into the CSA save area CSAOSRSA
  because of this. During processing of the DFHMQ0334 message,
  the CKAM task is suspended on a wait for the intrapartition TD
  TDIPLOCK. While CKAM is suspended, the TCP task runs and the
  CSA register save area is used on behalf of the TCP task.
    When the CKAM task is resumed, it completes its message
  processing and attempts to return to DFHMQDSC. However, as the
  registers are now associated with the TCP task, processing now
  goes awry. The result is a looping TCP task.
  KEYWORDS: msgdfhmq0334 mq0334 0334 334 mq wmq
  

Problem conclusion

• DFHMQDSC has been changed to correctly restore the eistg address
  into R13 upon exit from delete_open_tcbs and
  write_uss_stats_to_smf_and_reset.
  

Temporary fix

• FIX AVAILABLE BY PTF ONLY
  

Comments

APAR Information

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

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

  UK44273

Modules/Macros

•    DFHMQDSC
  

Fix information

• Fixed component name

  CICSTS V3 Z/OS

• Fixed component ID

  5655M1500

Applicable component levels

• R500 PSY UK44273

     UP09/02/26 P F902

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.