IV21416: CRASH IN SNA_MUX

Communications Server for AIX, Version 6.4 -- Latest PTFs

APAR status

• Closed as program error.

Error description

• Communications Server for AIX
  -----------------------------
  Crash in sna_mux.
  

Local fix

Problem summary

• USERS AFFECTED: All using the back-level LU6.2 APIs.
  
  PROBLEM DESCRIPTION:
  Dump stack:
  CPU 0 CSA F000000030033600 at time of crash, error code for
  LEDs: 30000000
  pvthread+002300 STACK:
  ___memmove64+000024 ()
  sna_mux64:vba_buf_to_mblk_copy+000164 (F1000E0004615D00,
  000000C8000000C8, F1000E0006B1C11C, 0000000800000008)
  sna_mux64:tps_parsed_ltv+000204 (0008000000000008,
  0003000000000003, F1000E0006B1C11C, 0F00023A0F00023A)
  sna_mux64:tps_parse_ltv+0000D4 (F1000E0006B1C11C,
  0000006000000060, F1000000C078E090, 0F00023A0F00023A)
  sna_mux64:tps_rcv_query_dload_rsp+000234 (F1000E0006B1C100)
  sna_mux64:tps_receive_proc+0000EC (0000000000000000,
  0FFFFFFFF007F9A0, F1000000C06E5B58)
  sna_ddpmgr64:sna_p_vpm_rcv_datagram+000180 (F1000E0006C301D8,
  F1000E0006B1C100, 0FFFFFFFF007FA60)
  sna_ddpmgr64:vpm_rcv_msg+0000C8 (F1000E0006C30100)
  sna_ddpmgr64:sna_p_vpm_stream_r_svc+00002C (F1000E000425DC00)
  pse:sq_wrapper+000134 (??)pse:csq_run+00029C (??)
  pse:csq_lateral+00036C (??, ??, ??)
  pse:scheduled_run+000100 ()
  .hkey_legacy_gate+00004C ()
  Netintr+0002F8 ()
  netisr_thread+000020 ()
  threadentry+000094 (??, ??, ??, ??)
  
  PROBLEM SUMMARY:
  The crash occurs when loading a back level LU 6.2 application.
  The problem is introduced by a previous fix (LI75513) which was
  included, though not listed, in the 6.4.0.2 level. The TP Server
  code is shared between the user space TP Server daemon and APPC
  libraries and the kernel space back-level LU 6.2 code. The fix
  was not correct for the kernel code where a message was used
  twice. In most cases thise was harmless, but when an unitialized
  field matched a control block index, we tried to release a
  streams message before we had finished using it, and hence
  caused the crash.
  

Problem conclusion

• Code has been maltered to apply the previous fix to only the
  user space code, because there is no need for the extra error
  checking in the kernel space back-level code.
  

Temporary fix

• sna_mux64
  

Comments

APAR Information

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

  IV06518

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

Fix information

• Fixed component name

  CS AIX V6.X

• Fixed component ID

  5765E5100

Applicable component levels

• R640 PSY U851826

     12/09/11 I 1000