PI61098: CICS CRASH WITH S0C4 ABEND BECAUSE OF TRACE TABLE OVERLAY.

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• Task has trouble in writing trace entry because of trace anchor
  block overlay and failed with abend 0C4.
  From the dump taken for this issue, KE domain shows many
  0C4 abends happen repeatedly under module DFHTRPX.
  In TR domain, INTERNAL TRACE TABLE can not check with message:
  "** INVALID FORWARD POINTER IN TRACE BLOCK"
  In reported case, TRACE ANCHOR BLOCK in TR domain is as below:
  TRA 266B0000 TRACE DOMAIN ANCHOR BLOCK
    0000  01146EC4 C6C8E3D9 C1D5C3C8 D6D94040
    0010  00000000 00000000 A65AC988 266B0000
    0020  00000047 AB23C5F5 00000000 56642A0B
          ******** *******
    0030  02710000 000003D8 000001FC 00027EAE
    0040  00000003 009F81A0 00000048 40000000
    0050  00000048 00000000 00000048 02710000
          ******** ******** ******** ********
  (20)  TRA_NAB         Next byte in internal table
  (50)  TRA_INTTAB_PTR  Address of start of table
  (58)  TRA_ENDTAB_PTR  Address of end of table
  .
  The trace table is held in storage between 48_00000000 and
  48_02710000. However the TRA_NAB in the trace table is
  x'47_AB23C5F5', which points outside the boundaries of the
  trace table because of overlay and cause CICS crash.
  
  Additional Symptom(s) Search Keyword(s): KIXREVxxx TRA_NAB
  S0C4 TRACE ANCHOR BLOCK overlay DFHTRPX
  DFHKE0303 RECURSIVE ABEND KERNEL DOMAIN.
  AKEA DFHTRPT 0999 DFHKERRI
  

Local fix

• No
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All.                                         *
  ****************************************************************
  * PROBLEM DESCRIPTION: The CICS region has taken a dump but    *
  *                      the trace domain section shows message  *
  *                      "** INVALID FORWARD POINTER IN TRACE    *
  *                      BLOCK" so the trace can not be          *
  *                      formatted.                              *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  The problem can manifest such that the CICS region will
  terminate with messages +DFHKE0303 A RECURSIVE ABEND HAS BEEN
  DETECTED BY THE KERNEL DOMAIN. and +DFHKE0201 ABOUT TO TAKE
  SDUMP. DUMPCODE: KERNDUMP, DUMPID: 0/0000. (MODULE DFHKESTX).
  The main symptom is that when looking at the dump, the Trace
  Domain (TR) section will not format with an explanation in the
  form of message "** INVALID FORWARD POINTER IN TRACE BLOCK".
  
  The trace domain will have been able to format the trace domain
  anchor block. At offset x'50' is the TRA_INTTAB_PTR and at
  offset x'58' is the tra_endtab_ptr, both these values are 64 bit
  addresses and state the beginning and end of the trace table
  respectively. At offset X'20' within the anchor block is the
  TRA_NAB, this value should show where the next available trace
  entry should start. In the dump, this will now show a value that
  is outside the range of the trace table. This will also be
  backed up by the value TRA_AVLEN at offset x'28'. This should
  show how much room is left in the trace block so the value
  should always be below X'1000'. The dump will show that this
  value is greater than that.
  
  The TRA_NAB value has been calculated based on the values of the
  trace entry data lengths. In this case a severe error trace
  point PG 190A 'PGCR  *EXC* - FREEMAIN_SET_STOR_ERROR - ' from
  module DFHPGCR contains two data areas however these data areas
  are 64 bit addresses but when the trace_put call was made only
  31 bit addresses were used. This means that residual data from
  the trace domain parameter list was used for the values of data
  area 2 rather than the actual 64 bit addresses. The residual
  data was interpreted as a large negative length which resulted
  in the TRA_NAB calculation being reduced to be outside the trace
  table range and also overwriting large parts of the trace table
  when the incorrect trace entry was written to the trace entry
  block as the length value was now being treated as a very large
  length.
  

Problem conclusion

• DFHPGCM has been altered to ensure that when a severe error is
  required, and the data provided is using 64 bit addresses, that
  those addresses are maintained during the TRACE_PUT call. This
  in turn will ensure that DFHPGCR is updated with this new
  severe_error64 routine.
  

Temporary fix

•             *********
              * HIPER *
              *********
  FIX AVAILABLE BY PTF ONLY
  

Comments

APAR Information

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

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

  PI64160 UI39875 UI39876

Modules/Macros

• DFHPGCR
  

Fix information

• Fixed component name

  CICS TS Z/OS V5

• Fixed component ID

  5655Y0400

Applicable component levels

• R800 PSY UI39875

     UP16/08/11 P F608 ¢

• R900 PSY UI39876

     UP16/08/11 P F608 ¢

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.