if_re Fatal trap 12: page fault while in kernel mode

[Yarema]

I'm running a pair of VIA C7 mini-ITX carp(4)ed 6.2-STABLE boxes with one 
if_vr and three if_re RealTek 8169S Single-chip Gigabit Ethernet NICs.  vr 
is used for pfsync and all re interfaces have carp.  I'm unable to run any 
kernel newer than April 11th, 2007 without getting something along the 
lines of:

Fatal trap 12: page fault while in kernel mode
fault virtual address	= 0x31
fault code		= supervisor read, page not present
instruction pointer		= 0x20:0xc0658773
stack pointer		= 0x28:0xe374bc28
frame pointer		= 0x28:0xe374bc7c
code segment		= base 0x0, limit 0xfffff, type 0x1b
			= DPL 0, pres 1, def32 1, gran 1
processor eflags		= interrupt enabled, resume, IOPL = 0
current process		= 11 (swi1: net)
trap number		= 12
panic: page fault
Uptime: 32s

usually within a minute of going multiuser.  Single user is stable and 
allows me to roll back to the good Apr 11th kernel.  I've tried building a 
newer kernel using /sys/dev/re/if_re.c Revision 1.46.2.26 committed on Apr 
11th with the same results.  Kernel panic within a minute of bootup, but 
rock solid in single user.

--On Friday, October 19, 2007 11:19:40 +0300 Vlad GALU <dudu at dudu.ro> wrote:

> I recall experiencing some crashes with ZERO_COPY_SOCKETS on
> RELENG_6 and it was related to the VM subsystem. Tor Egge may have a
> better idea about the problem and whether it was fixed in RELENG_7 or
> not.

I do have ZERO_COPY_SOCKETS enabled.  Could this be relevant?

I've put up the the dmesg.boot, kernel, vmcore, panic.txt and the KERNCONF 
files at http://yds.CoolRat.org/C787D3NG/

-- 
Yarema