Patch-ID# 106596-06
Keywords: AP alternate pathing BAD TRAP panic mether_rarp mutex_panic 
Synopsis: AP 2.1: BAD TRAP panic mether_rarp mutex_panic chksumming 
Date: Oct/05/00

Solaris Release: 2.6

SunOS release: 5.6

Unbundled Product: Alternate Pathing

Unbundled Release: 2.1

Topic: AP 2.1: BAD TRAP panic mether_rarp mutex_panic chksumming 

BugId's fixed with this patch: 4141438 4143514 4147674 4153152 4161396 4163270 4166620 4170818 4197600 4215821 4220646 4244664 4291115 4297640 4311971 4337444 4338346 4339219

Changes incorporated in this version: 4297640 4337444 4338346 4339219

Relevant Architectures: sparc.sun4u sparc.sun4u1

Patches accumulated and obsoleted by this patch: 106538-03

Patches which conflict with this patch: 

Patches required with this patch: 

Obsoleted by: 

Files included with this patch:

/kernel/drv/ap
/kernel/drv/mge
/kernel/drv/mge.conf
/kernel/drv/mhme
/kernel/drv/mle
/kernel/drv/mnf
/kernel/drv/mqe
/kernel/drv/mqfe
/kernel/drv/mvge
/sbin/ap_dbupgrade
/sbin/apboot
/sbin/apcheck
/sbin/apconfig
/sbin/apdb
/sbin/apdisk
/sbin/apinst
/sbin/apnet
/usr/lib/libap.so.1
/usr/sbin/ap_daemon
/usr/sbin/apconfig
/usr/sbin/apdb
/usr/sbin/apdisk
/usr/sbin/apnet

Problem Description:

4297640 BAD TRAP panic with mether_rarp with AP network switching
4337444 PANIC: "mutex_panic" with AP switching and heavy Gigabit traffic
4338346 AP patch 108719-01 for 4311971 regresses
4339219 AP has improper handling of HW chksumming (ICK_M_CTL_MAGIC mblks)

(from 106596-05)

4291115 switching of mether with IP traffic can cause panic
4311971 recursive mutex_enter panic on OS 2.6 running with AP 2.3 and Trunking/GEM

(from 106596-04)

4215821 first manual ap switch of ethernet metadevice loses communication for a while
		Due to the various configurations of network switches
		and routers, there may still be some period immediately
		following a switch on the AP path during which traffic
		will appear to be suspended.  This period can last up
		to 90 seconds, after which traffic will resume
		normally.  This is a side effect of the various media
		negotiation and topology "learning" used by some switch
		manufacturers.  It may be possible to reduce this delay
		by configuring the underlying device drivers (e.g. hme)
		and the switch with similiar link parameters (speed,
		duplex), and disabling the autoconfiguration features
		on your switch or router.  Consult the physical device
		driver and switch documentation for details on the
		specific steps.  Note that in no case should a delay of
		more than 90 seconds be experienced, and in no case
		should this cause TCP or RPC connections to be
		dropped.
4220646 AP patchID 106538-02 has incorrect dependancy on patches 106048-04 + 106284-02
4244664 SECURITY: meta drivers have world read/write permissions!

(from 106596-03)

4163270 netstat of network ap network meta interface shows no tallies

(from 106596-02)

4166620 snoop of AP  meta network interface stops snooping after switch
4170818 If you run the $<callouts nadb macro, or fm2's "callout ts" command, you see tho

(from 106596-01)

4141438 mhme interface hangs under heavy load
4143514 FDDI with AP hangs with heavy use of network
4147674 AP cuases ifconfig to hang on a mutex

(from 106538-03)

4197600 /sbin/apdb dumps core on bad arg

(from 106538-02)

4161396 AP 2.x needs capability to work with GEM

(from 106538-01)

4153152 apconfig works as an ordinary user

Patch Installation Instructions: 
-------------------------------- 
Refer to the Install.info file within the patch for instructions on
using the generic 'installpatch' and 'backoutpatch' scripts provided
with each patch.  Any other special or non-generic installation
instructions should be described below.

Special Install Instructions: 
----------------------------- 
none