IV03534: CRASH IN SNA_V5ROUTER

Personal Communications 6.0.5
Communications Server for AIX, Version 6.3 -- Last PTFs

APAR status

• Closed as program error.

Error description

• Communications Server for AIX and Linux
  ---------------------------------------
  Crash in sna_v5router
  

Local fix

Problem summary

• USERS AFFECTED: All.
  
  PROBLEM DESCRIPTION:
  CRASH INFORMATION:CPU 2 CSA F0047BE0 at time of crash, error
  code for LEDs: 70000000
  pvthread+000C00 STACK:
  [032E0F0C]sna_v5router:nrm_ps_to_rm_rec+000190 (7A2C136C,
  372FFCFC)
  [032E1988]sna_v5router:nrm_queue_handler+00009C (7A2C136C,
  00040004, 372FFCFC)
  [0311A51C]sna_v5router:nba_dispatch_input+00027C (38DEE498,
  372FFCFC)
  [0311ABE4]sna_v5router:nba_dispatch_process+0000C4 (38DEE498,
  372FFCFC)
  [0311B3D8]sna_v5router:nba_scheduler+0001FC ()
  [0310D51C]sna_v5router:vpr_stream_uw_drive_scheduler+000028 ()
  [0310DA08]sna_v5router:vpr_stream_uw_svc+000120 (726C5494)
  [028E23B8]pse:sq_wrapper+000164 (??)
  [028CEB44]pse:csq_run+000524 (??)
  [028CD8B4]pse:csq_lateral+00009C (??, ??, ??)
  [028E2C80]pse:scheduled_run+0000B8 ()
  [001D4F18]Netintr+0001A0 ()
  [001D4D34]netisr_thread+000014 ()
  [00152534]threadentry+000014 (??, ??, ??, ??)
  
  PROBLEM SUMMARY:
  There is a very small window when two transactions (one on each
  end of the session) are initiated at the same time. If the
  session is a contention loser on the CSAIX end, and the remote
  end sends a complete conversation (BB and CEB) at the same time
  that CSAIX sends a BID to initiate a conversation, those can
  cross and be valid. If those two events are processed in the
  node under the same callback interrupt, then the processing for
  them gets interleaved, causing a crash.
  
  The window of opportunity for this crash is very, very small.
  

Problem conclusion

• Internal code has been modified to properly handle this
  situation. Due to differences in DLC code between AIX and Linux,
  it is believed that this problem can not occur on Linux, but the
  fix will be applied to CS Linux, anyway.
  

Temporary fix

• sna_v5router
  

Comments

APAR Information

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

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

  IV21415 LI76744 LI76745 LI76746

Fix information

• Fixed component name

  CS AIX V6.X

• Fixed component ID

  5765E5100

Applicable component levels

• R631 PSY U848721

     UP11/11/18 I 1000

• R640 PSY

     UP