VM66778: ABEND PRG004 WHILE PROCESSING SSI ISFC SYNC MESSAGE

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• In extremely rare circumstances, a serialization problem in the
  ISFC TRANSMIT subroutine may lead to a dangling pointer which
  causes a PRG004 HARD ABEND at HCPKCD+2D96 during MBRBK cleanup.
  

Local fix

• N/A
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: Only customers using an SSI configuration    *
  *                 can encounter this error. A larger SSI       *
  *                 cluster seems more likely (compared to a     *
  *                 smaller cluster) to encounter the problem.   *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  ****************************************************************
  * RECOMMENDATION: APPLY PTF                                    *
  ****************************************************************
  The customer was running z/VM 7.3.0 in an SSI cluster. They
  reported Abend PRG004 taken on a production system during
  normal operations. The dump recorded an addressing exception
  in module HCPKCDTO entry point attempting to store into page 0
  while following the SCKBK queue of outbound messages waiting
  for completion (SCKOQUHD->MBRBK). Dump analysis revealed that
  the SCKOQUHD->MBRBK chain linkage had been damaged by an
  earlier error (later discovered in the TRANSMIT subroutine).
  

Problem conclusion

• HCPKCD TRANSMIT subroutine temporarily releases SCKLKWRD
  serialization while a message request (represented by MBRBK)
  is being processed by the Link I/O layer of ISFC. Upon
  completion of a request, SCKLKWRD should have been obtained
  earlier (to avoid a rare timing window).
  The Link I/O layer stacked a task to return to the original
  thread that sent the message, and continued to manipulate
  the MBRBK chain on SCKOQUHD (while holding SCKLKWRD).
  However, in rare cases, the stacked task can be dispatched
  on another processor while the inline code is removing the
  MBRBK. This caused the bad link pointers that eventually
  caused the PRG004 abend.
  The HCPKCD TRANSMIT subroutine is revised to obtain and
  release SCKLKWRD in code segments that originally did not
  appear to need SCKLKWRD serialization.
  

Temporary fix

• FOR RELEASE VM/ESA CP/ESA R730 :
  PREREQ: VM66693
  CO-REQ: NONE
  IF-REQ: NONE
  

Comments

APAR Information

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

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

  UM90458

Modules/Macros

• HCPKCD
  

Fix information

• Fixed component name

  VM CP CP

• Fixed component ID

  568411202

Applicable component levels

• R730 PSY UM90458

     UP25/06/10 P 2501

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.