OA60230: MASTER HANG DETECTED ABEND30D BPXLKKW BPXPRTRM EOM

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• An address space is going through EOM processing under the
  Master address space. A hang is detected by RTM while z/OS UNIX
  termination manager is attempting to cancel a waiting thread,
  resulting in aS30D abend.
  
  
  ANALYSIS:
  RTM under the Master address space invoked the z/OS UNIX
  resource recovery termination manager (BPXRRTRM) for EOM
  processing. BPXRRTRM next gave control to process termination
  (BPXPRTRM) using the syscall interface.  The stack at time
  of abend shows the following call sequence:
  
    BPXJCPC->BPXPRTRM(UA99597)->BPXLKKW(UJ02237)
  
  z/OS UNIX Process termination found a waiting thread and
  attempted to cancel the kernel wait by invoking
  BPXLKKW(#cancel). The issue is that the wait was entered while
  running in the address space that is going through EOM
  processing. BPXLKKW checks the Kserascb field to determine
  under what address space the wait was issued.  However, BPXPRTRM
  changes the Kserascb field to the currently running address
  space before invoking BPXLKKW thus causing BPXLKKW to
  incorrectly enter the wait.
  
  VERFICATION STEPS:
  1) Abend30D under the Master Address space
  2) wait was issued by BPXLKKW
     review the Linkage Stacks under the TCB
     that suffered the S30D abend and locate
     the linkage stack that issued the PC 030D
     R14=030D and the TARG = 00000000 0000030D
     The PSWE field should be an address within
     BPXLKKW
  3) Verify that BPXLKKW was called by BPXPRTRM
     with parm #CANCEL.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * All users of z/OS UNIX System Services for HBB77B0 and       *
  * HBB77C0                                                      *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * A process going through EOM and running in the master        *
  * address space is hung, waiting for a POST.                   *
  * This results in ABEND30D - END OF MEMORY RESOURCE MANAGER    *
  * HANG DETECTED.                                               *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  During EOM for a process, USS resource recovery termination
  manager, BPXRRTRM gets control, running in the master address
  space.  BPXPRTRM is called to cleanup Kernel resources.   If
  there is an outstanding WAIT for a thread, BPXLKKW is called to
  cancel the WAIT.  Before calling BPXLKKW, the KSERASCB field is
  changed from the ASCB for the processes in EOM to the master
  address space.  When BPXLKKW CANCEL processing sees an
  outstanding WAIT, it checks the KSERASCB field and since it is
  the master address space (which is active), it continues
  waiting.   Since any POST is directed to the process going
  though EOM and not the master ASID, the master ASID hangs and
  eventually receives an ABEND30D.
  

Problem conclusion

• In BPXPRTRM, the ASID of the process going through EOM is saved
  in the KSER.   When BPXLKKW is called to CANCEL the WAIT, it
  will see that the thread is in EOM, and the current active ASID
  (the Master ASID) is different than the original ASID.  Since
  the original ASID is no longer active,  the WAIT is cancelled.
  

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:

  UJ04698 UJ04699

Modules/Macros

• BPXPRTRM BPXLKKW
  

Fix information

• Fixed component name

  OPENMVS SYS SRV

• Fixed component ID

  5695SCPX1

Applicable component levels

• R7B0 PSY UJ04698

     UP21/01/12 P F101

• R7C0 PSY UJ04699

     UP21/01/12 P F101

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.