PK66562: ABEND0C4 IN DFHMQTRU WHEN USING A CSQCAPX CROSSING EXIT AND THE EXIT PROGRAM MAKES MQ API CALLS.

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• Abend0C4 in DFHMQTRU when using a CSQCAPX crossing exit and
  that exit makes MQ API calls.
     The abend0C4 happens just after the EXEC CICS LINK to
  CSQCAPX returns back to DFHMQTRU.  If trace is active the last
  trace entry prior to the abend0C4 *EXC* trace entry is
  EIP EXIT LINK.   The RET address on that trace entry will be
  at +X'2794' into DFHMQTRU (UK27574).  This is right after the
  call to DFHMQPRH.
     If you have the corresponding EIP ENTRY LINK trace, the RET
  on that will be pointing to +X'2E82' into DFHMQTRU.  This the
  point where DFHMQTRU does the EXEC CICS LINK to CSQCAPX.
     The abend0C4 can occur in different spots.  One spot is
  at +X'26D0' on a 58207024 because Reg7 is bad.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: ALL                                          *
  ****************************************************************
  * PROBLEM DESCRIPTION: DFHMQTRU abends when using the API      *
  *                      crossing exit CSQCAPX.                  *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  Customers report abends occurring in the MQ TRUE (DFHMQTRU)
  occurring during the execution of transactions that utilize the
  MQ API.
  
  These abends occur at various points within DFHMQTRU. In all
  cases the use of the API Crossing Exit, CSQCAPX, was evident.
  The problem was shown to be an error in the code, within
  DFHMQTRU, that which handled the calling of the API exit via an
  EXEC CICS LINK.
  
  It is possible for the MQ TRUE to be called recursively should
  the API exit itself make MQ API calls. To protect the contents
  of its working storage the TRUE adjust an address in an MQLOT,
  the new value addressing a "Second Level" working storage area.
  A section at the top of each of these areas is reserved as a
  register savearea. When the link to the API exit completes the
  adjustment is undone and the address in the MQLOT is returned to
  its original value.
  
  The EXEC CICS LINK call requires a savearea and one is provided
  by setting R13 to the adjusted working storage value. However
  this action is incorrect
  
  When the MQ TRUE is re-entered to process an MQ API call issued
  by the API exit the EXEC CICS LINK register save area is
  corrupted. The incoming MQ request requires the TRUE to call MQ
  interface code. This call requires a savearea and one is
  provided again by setting R13 to the address of the current
  working storage. As this is the same location where the EXEC
  CICS LINK registers are stored, they are now overwritten. When
  the EXEC CICS LINK to the API exit completes invalid data is
  restored into the registers, leading to abends at unpredictable
  locations.
  

Problem conclusion

• The setting up of the savearea address for the EXEC CICS LINK,
  to the API exit has been amended. DFHMQTRU will now use the
  working storage address before the adjustment carried out to
  handle recursive calls.
  

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:

  UK37779

Modules/Macros

•    DFHMQTRU
  

Fix information

• Fixed component name

  CICSTS V3 Z/OS

• Fixed component ID

  5655M1500

Applicable component levels

• R500 PSY UK37779

     UP08/07/03 P F807

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.