PM11552: LOOP FPQSTDMP WHEN PRINTING DIAGNOSTIC TRACE WITH NO ENTRIES & MSGFPQ8007I DUE TO CONN CONTROL BLOCK PREVIOUSLY FREED

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• Loop in FPQSTDMP when printing diagnostic trace with no entries.
  Loop occurs within about five instructions between about x'346'
  and x'418' into FPQSTDMP.
  
  Additional issue addressed:
  FPQ8007I error message due to CONN control block previously
  freed.
  
  This is a forward fit of ITKB V1.1 APAR's PM08799 and PM11551.
  
  Keywords: looping wait hang hung ITKB ISPF user hanging
            MSGFPQ8007I
  

Local fix

• A user can circumvent the looping problem by updating the
  FPQBPCFG member to add the line as follows:
  # DEFINITIONS FOR FPQ TRACES
  TRCLEV=(*,LOW,FPQ)          /* DEFAULT ALL TRACES TO LOW      */
  TRCLEV=(DIAG,HIGH,FPQ,PAGES=16)    <------------ Add this line
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: There are two errors addressed in this PTF:  *
  *                  Loop in FPOQSTDMP - affects any user that   *
  *                    issues the diagnostic command DUMPTRACE   *
  *                    with no entries in the trace table.       *
  *                  FPQ8007I -                                  *
  *                    This error can affect all users, however  *
  *                    it is very rare resulting from an ITKB    *
  *                    client address space being cancelled      *
  *                    while a request is being processed by     *
  *                    the server.                               *
  ****************************************************************
  * PROBLEM DESCRIPTION: Issuing a DUMPTRACE command when there  *
  *                      are no trace entries (as occurs when    *
  *                      the trace is off) results in a loop in  *
  *                      the server address space.               *
  *                      Note the distributed BPECONFIG sets     *
  *                      the trace off.                          *
  *                                                              *
  *                      If an ITKB client TCB (such as an IMS   *
  *                      Tool product that is writing to the     *
  *                      report repository) is abnormally        *
  *                      terminated while a request is           *
  *                      processing for that TCB in the server   *
  *                      address space, a race condition may     *
  *                      exist in the server address space such  *
  *                      that it attempts to cleanup the same    *
  *                      control block twice resulting in        *
  *                      message FPQ8007I being issued.  In some *
  *                      cases another client's processing may   *
  *                      be affected.  When this problem is      *
  *                      detected, a DUMPTRACE is issued         *
  *                      internally.  If there are no trace      *
  *                      entries, the server address space       *
  *                      loops.                                  *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  A loop occurs if the DUMPTRACE command executes when there
  are no trace entries.
  
  Cancelling an ITKB client way, in rare cases, result in
  attempting to free the same control block twice.  This will
  issue a DUMPTRACE internally resulting in a loop.
  

Problem conclusion

• The DUMPTRACE command processing is corrected to handle zero
  entries.
  
  The cleanup process (which is launched for lost clients)
  serialization was improved to ensure that it does not attempt to
  free the CONN when there is an active thread.  Lost client
  notification processing was moved into FPQSREQQ so that the
  chain scan can be performed with a shared latch.
  

Temporary fix

Comments

APAR Information

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

  PM08799

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

  UK57992

Modules/Macros

•    FPQSGSRB FPQSMSRB FPQSREQQ FPQSTCPL FPQSTCPQ
  FPQSTDMP FPQSXCFT
  

Fix information

• Fixed component name

  IMS TOOLS BASE

• Fixed component ID

  5655V9300

Applicable component levels

• R110 PSY UK57992

     UP10/06/17 P F006

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.