[2025] in Release_7.7_team

home help back first fref pref prev next nref lref last post

Athena 8.3.24 patch release

daemon@ATHENA.MIT.EDU (Greg Hudson)
Mon Jan 10 11:20:50 2000

Date: Mon, 10 Jan 2000 11:20:25 -0500
Message-Id: <200001101620.LAA03488@egyptian-gods.mit.edu>
From: Greg Hudson <ghudson@MIT.EDU>
To: release-announce@MIT.EDU

The Athena 8.3.24 patch release is currently scheduled for the evening
of Monday January 17.

A warning: if you take the update by hand, or if you are watching
while the update happens automatically, you may see your machine panic
with the message "recursive mutex_enter".  Simply let the machine
reboot and (if you are updating by hand) restart the update, and all
should be well.  The Solaris bug which causes this panic is fixed in
the 8.3.24 patch release, so it shouldn't happen again.

Changes in this patch release include:

	* Solaris patch 105529-08 will be applied to prevent recursive
	  mutex locking panics.

	* IRIX patch 3800 will be applied to fix an NFS3 client bug.

	* A symlink for the cosmoplayer netscape plugin will be fixed
	  on IRIX.

	* The order of operations in the update has been changed so
	  that daemons will be killed off before the update touches
	  the disk.

	* There's a fix in krb5.conf for the CYGNUS.COm realm KDCs.

	* ftpd -C should work better with local accounts.

	* The lprng servers have a workaround for an HP printer
	  problem.

	* On Solaris, the sadmind service will be disabled in
	  /etc/inet/inetd.conf.  (We previously disabled the service
	  in a non-patch-release by forcing /usr/sbin/sadmind to be
	  zero-length; that continues to be the case.)

	* The zephyr syslogd has a patch to avoid core dumping when
	  reporting certain errors.

	* The masks file has an update for 18.101.

	* traceroute has some changes to prevent (unlikely) buffer
	  overflows and to make it less useful as a packet flooder.

	* attach will block several signals while writing out
	  attachtab files to prevent attachtab corruption.

	* attach -o won't free static memory.

	* attach won't lose the first character of def_options
	  strings.

	* detach -C won't detach lockers wanted only by the invoking
	  user.

	* A typo in the binhex-decoding rule in the MH MIME
	  configuration has been fixed.

home help back first fref pref prev next nref lref last post