IV96352: LOGIN HANGS TRYING TO WRITE TO /ETC/UTMP

APAR status

• Closed as program error.

Error description

• **************************************************************
  * USERS AFFECTED:
  * Systems running the AIX 7100-03 Technology Level
  * with bos.rte.libc at the 7.1.3.49 level.
  **************************************************************
  * ERROR DESCRIPTION:
  * In the latest SPs, all logins to AIX or VIOS LPARs via
  * telnet, ssh, or console may hang.  This can happen because a
  * lock on /etc/utmp is permanently held by a running process.
  *
  * This has been seen commonly with the 'screen' or 'init'
  * processes holding the lock.
  *
  * Killing all the processes that opened the file /etc/utmp ,
  * can free the lock and allow logins.
  * For example:
  *
  * # fuser -k /etc/utmp
  *
  * If the process cannot be killed (such as 'init'), then the
  * file can be renamed, allowing new logins to create and use
  * a new /etc/utmp file and succeed.
  * For example:
  *
  * # mv /etc/utmp /etc/utmp.bak
  *
  * Note: this will cause some commands such as 'who' and 'w'
  * to report incorrect information about users logged in to
  * the system.
  *
  * Note: These are only temporary workarounds, as the issue
  * can reoccur until the fix is applied and the system is
  * rebooted for the fix to take effect in all processes.
  **************************************************************
  * RECOMMENDATION:
  * An interim fix is available from either
  * ftp://aix.software.ibm.com/aix/ifixes/iv96352/
  * https://aix.software.ibm.com/aix/ifixes/iv96352/
  * Installation of the ifix requires a reboot.
  **************************************************************
  

Local fix

Problem summary

• login sessions start to hang. In customer scenario init
  process while writing to /etc/utmp has taken a write lock
  and it is never released. Hence login sessions hang as they
  do not get lock to write /etc/utmp. This is caused because
  an uninitialized size is assigned to len value in lock
  structure
  

Problem conclusion

• Initialize a size variable to proper value i.e size of lock
  structure before assigning this value to len variable
  in lock structure.
  

Temporary fix

•   *********
    * HIPER *
    *********
  

Comments

APAR Information

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

  IV96177

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

Fix information

• Fixed component name

  AIX V7.1

• Fixed component ID

  5765H4000

Applicable component levels