PH20975: REGION WITH TRANISO=YES INCURS 0C4 ABENDS IN DFHKESTX & DFHKERCD THEN HANGS AFTER PURGE OF TASK EXECUTING DB2 RE

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• You have a task issuing DB2 requests that gets purged, resulting
  in an expected AD2R abend however the  RTM2WA shows 0C4 abends
  that occurred in DFHKERCD and DFHKESTX.
  
  DFHKERCD records kernel error data during percolation of the
  abend AD2R and abends when trying to locate the module header
  for DFHD2EX1 who was loaded near the end of an extent. DFHD2EX1
  uses LIFO stacks and LIFO PLISTs and there is half word offset
  value in the LIFO PLIST that is declared as a signed half word
  instead of an unsigned half word. This value is treated as a
  negative value.
  
  DFHKERCD uses this negative value to do a subtraction
  which ends up adding a positive value. This resulting address
  is in a different extent which is owned by the TCP task.
  TRANISO=YES so the task that abended cannot access KEY9 storage
  owned by the TCP task, resulting in the 0C4 in DFHKERCD. This
  program check drives the CICS ESTAE exit (DFHKESTX) who tries
  to access this same address then program checks for the same
  reason.
  
  This is catastrophic as CICS is marked as 'in recovery' by the
  operating system at the time of the program checks. This causes
  the operating system to pull the rug away from our attempt to
  recover and the QR TCB ends up in operating system abend
  processing which eventually leads to a hang.
  
  Additional Symptom(s) Search Keyword(s): KIXREVxxx
  

Local fix

• N/A
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICS Users                               *
  ****************************************************************
  * PROBLEM DESCRIPTION: Purging a DB2 task causes 0C4 abends to *
  *                      occur in DFHKERCD and DFHKESTX, which   *
  *                      then leads to the CICS region hanging.  *
  ****************************************************************
  After a CICS task running DB2 has been purged, abend 0C4s then
  follow in DFHKERCD and DFHKESTX. This is due to a bad address
  being generated in DFHKERCD when it tries to gather kernel error
  data from DFHD2EX1. The kernel modules use a signed halfword in
  DFHD2EX1's LIFO structure to address the module name and this
  leads to a bad address beyond the end of the module. Normally
  this would cause undefined data to be reported as the module
  name but in this case DFHD2EX1 was at the end of a storage
  extent and the following extent was for Key9 storage. This led
  to the program checks occurring.
  
  Additional keywords: TRANISO
  

Problem conclusion

• The definition of field LF_PLIST_MODULE_OFFSET in copybook
  DFHKEMHD has been changed to be BIN(16) rather than the
  previously incorrect BIN(15).
  

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:

  UI67713 UI67714

Modules/Macros

• DFHD2TM  DFHDUDU  DFHISIC  DFHKEDD  DFHKEDS  DFHKEDUF DFHKEEDA
  DFHKEGD  DFHKEIN  DFHKERCD DFHKERER DFHKERET DFHKERKE DFHKERPC
  DFHKERRI DFHKESIP DFHKESTX DFHKETA  DFHKETCB DFHLDDM  DFHLDDMI
  DFHLDLD1 DFHMEBM  DFHMEBU  DFHMPIB  DFHMQTM  DFHPCXDF DFHPGDM
  DFHPISC  DFHSODUF DFHSUME  DFHTDA   DFHTRDM  DFHW2DUF DFHWBBMS
  DFHWBDUF DFHWBTL  DFHWUDUF DFHXMXD  DFHXSSA  DFHXSSH  DFHZGAI
  

Fix information

• Fixed component name

  CICS TS Z/OS V5

• Fixed component ID

  5655Y0400

Applicable component levels

• R100 PSY UI67713

     UP20/02/18 P F002 ¢

• R200 PSY UI67714

     UP20/02/12 P F002 ¢

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.