Note: There was a bad value `' for the field `Class'.
	It was set to the default value of `sw-bug'.

>Number:         30160
>Category:       kern
>Synopsis:       TCP SACK code is unstable and periodically causes a 'panic: m_copydata: m == 0, len 1'
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri May 06 23:13:00 +0000 2005
>Originator:     jmcneill@NetBSD.org
>Release:        NetBSD 3.99.3, NetBSD 2.99.16
>Organization:
Invisible
>Environment:
System: NetBSD triton.invisible.ca 3.99.3 NetBSD 3.99.3 (TRITON.MP) #0: Sun Apr 24 13:27:16 ADT 2005 jmcneill@triton.invisible.ca:/usr/obj/sys/arch/sparc/compile/TRITON.MP sparc
Architecture: sparc
Machine: sparc
System: NetBSD centauri.invisible.ca 2.99.16 NetBSD 2.99.16 (CENTAURI) #0: Mon Mar 14 07:57:22 AST 2005  jmcneill@centauri.invisible.ca:/u2/obj/sys/arch/alpha/compile/CENTAURI alpha
Architecture: alpha
Machine: alpha
>Description:
Periodically, two of my machines (running completely different workloads) will
die with 'panic: m_copydata: m == 0, len 1'. One is an MP sparc, the other is
a UP alpha.

Disabling sack makes the problem disappear:
  sysctl -w net.inet.tcp.sack=0
  sysctl -w net.inet6.tcp6.sack=0

This is a definite must-fix before the next release.
>How-To-Repeat:
Unknown.
>Fix:
Unknown. See 'Description' for workaround.

>Unformatted: