VM66280: VSWITCH OSA RDEV BLOCKED WITH POSSIBLE ABEND MCW002 OR MPC008

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• A stable VSWITCH network began to exhibit poor performance,
  connect failures, and/or timeouts after OSA microcode update.
  In one case, system performance was impacted and one processor
  was found to be looping through entry point HCPVQSEQ.
  
  Network traces revealed that a significant number of malformed
  packets were being generated, and being processed by the OSA
  Express. This appears to be a contributing factor to the
  problem. Special processing is required by the OSA Express, and
  by the host device driver (CP VSWITCH logic in this case). An
  error in this process leads to (1) the loss of additional,
  valid, packets when the malformed packet is discarded, and (2)
  the loop in HCPVQS if the unexpected buffer state is detected at
  the wrong point in processing.
  
  The loop in HCPVQS will disrupt connectivity through the
  affected OSA RDEV connection, and lead to an ABEND MCW002
  or ABEND MPC008.
  

Local fix

• Apply PTF
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: Systems using VSWITCH in a physical network  *
  *                 where malformed packets may be generated     *
  *                 could notice loss of valid packets and       *
  *                 (in rare cases) a system hang with a loop    *
  *                 in HCPVQSEQ.                                 *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  ****************************************************************
  * RECOMMENDATION: APPLY PTF                                    *
  ****************************************************************
  The OSA Express detected a malformed packet and marked the
  associated buffer segment with an error state (part of the
  protocol for that interface). A flaw in the OSA firmware
  resulted in a change to other buffer state indicators that
  was outside the defined protocol. The VSWITCH device driver
  processing inbound data (HCPVQSEQ) could not process that
  buffer state, but it also could not advance to the next
  buffer segment. That left HCPVQS with an infinite loop.
  This can lead to a hang, ABEND MCW002 or ABEND MPC008.
  
  Note that a firmware update has corrected the OSA Express
  flaw, but the VSWITCH driver needs to address the possibility
  that a protocol error is possible in some future firmware
  upgrade, and make provisions to avoid a CP failure.
  

Problem conclusion

• HCPVQS is updated to handle buffer state tests using a
  strategy that will detect an invalid state (by testing all
  valid states).  This includes logic to count a "bad" buffer
  as an error packet in the NICBK error count. When a protocol
  error (such as an unsupported buffer state) is encountered,
  HCPVQS will force a reset (CSCH) of the adapter.
  
  HELP HCP2832E is modified to include this section:
   Connection device for VSWITCH SYSTEM switchname is not active
   OSA Express interface error detected.
  
   Explanation:  The VSWITCH device driver in CP encountered an
   unexpected or undefined status associated with the OSA Express
   queue buffer.  This makes it impossible to continue using the
   interface without resetting it to a known state.
  

Temporary fix

• FOR RELEASE VM/ESA CP/ESA R640 :
  PREREQ: VM66160
  CO-REQ: NONE
  IF-REQ: NONE
  FOR RELEASE VM/ESACP/ESAR710 :
  PREREQ: VM66219 VM66283
  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:

  UM35538 UM35539 UM35540

Modules/Macros

• HCPMES   HCPMESA  HCPMESB  HCPMXRBK HCPVQS   HCP2832E
  

Fix information

• Fixed component name

  VM CP

• Fixed component ID

  568411202

Applicable component levels

• RA64 PSY UM35812

     UP21/02/17 I 1000 ¢

• R640 PSY UM35539

     UP19/09/30 P 2001 ¢

• R710 PSY UM35540

     UP19/09/30 P 2001 ¢

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.