OA67537: TCPIP LOOP RUNNING INBOUND SRB WHEN ISTLLCIE RETURNS TO INCORRECT CALLER

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• Inbound packet processing running under a SRB in the TCPIP
  address space abends.  The ISTZRM01 FRR gets control and
  releases the CPU lock.  The R13 register save area (RSA) address
  is extracted and retry is done to IERETRY in ISTLLCIE.  The
  IERETRY routine uses the extracted RSA to obtain the caller's
  RSA address and restores the caller's registers and branches to
  R14 to return to the caller.
  When the inbound SRB started it obtained the CPU lock and used
  the DWA associated with the CPU number as its RSA.  When the FRR
  released the CPU lock the SRB it became interruptible.  Before
  IERETRY obtained the caller's RSA address an interrupt occurred
  on the CPU for a VTAM managed network device, causing the DWA to
  be used for the disabled interrupt code path.  This caused the
  RSA to be updated with the registers at the time the VTAM module
  was called to process the interrupt.  The RSA in the DWA now has
  an R14 value that points back to the SLIH (second level
  interrupt handler) and the unit of work that was interrupted and
  the SLIH will return control to is the original inbound SRB that
  abended.
  The resulting symptoms from using the DWA RSA after releasing
  the CPU lock are unpredictable.  The reporting problem resulted
  in an enabled loop where the inbound SRB branched back to the
  SLIH R14 and the SLIH restored the interrupted PSW so the code
  was stuck loading the interrupted PSW and giving control which
  loaded the interrupted PSW.  Interrupts were able to run because
  the SRB was enabled at this point, but every time an interrupt
  occurred it was always this same PSW address so the loop resumed
  as soon as the interrupt exited.
  
  VERIFICATION STEPS:
  If the loop occurs a dump that includes the system trace table
  will show an enabled loop for an SRB with TCPIP as the home
  address space and the PSW in IEAVEIO.
  
  ADDITIONAL SYMPTOMS:
  Work may be delayed if the LPAR is capped due to the loop.
  Omproute delayed sending HELLO and loss of adjacency due to
  dispatching delay causing routes to be lost.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * All users of the IBM Communications Server for z/OS 2.5 and  *
  * 3.1                                                          *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * Loop in the second level interrupt handler occurs when an    *
  * abend processing an OSA incorrectly releases the CPU lock.   *
  ****************************************************************
  * RECOMMENDATION:                                              *
  * Apply PTF                                                    *
  ****************************************************************
  ISTLLCIE is running for inbound OSA packet processing and an
  abend occurs.  The recovery routine, ISTZRM01, gets control and
  releases the CPU lock and passes the R13 value back to the retry
  routine in ISTLLCIE.  The retry routine accesses the R13 value
  that is only valid while the CPU lock is held and uses this
  value to restore the caller's registers.  At the time the R13 is
  accessed that storage area has been reused by the second level
  interrupt handler (SLI), resulting in it getting control with
  the CPU lock released.
  

Problem conclusion

• ISTZRM01 has been amended to extract the caller's R13 address
  before releasing the CPU lock and passing the extracted value to
  the retry routine in ISTLLCIE.
  

Temporary fix

• *********
  * HIPER *
  *********
  

Comments

APAR Information

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

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

  UJ97105 UJ97106 OA68080

Modules/Macros

• ISTZRM01 ISTLLCFB ISTLLCIE ISTS2RMQ ISTSRRMQ IUTLLRTP ISTLLCAD
  ISTLLCOO ISTLLCRB ISTSRIS8 ISTTSCCN
  

Fix information

• Fixed component name

  VTAM MVS/ESA

• Fixed component ID

  569511701

Applicable component levels

• R310 PSY UJ97106

     UP25/05/28 P F505 ¢

• R250 PSY UJ97105

     UP25/05/28 P F505 ¢

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.