amd64/148157: IPFW in kernel nat BUG found in FreeBSD 8.1-PRERELEASE

[Garrett Cooper]

The following reply was made to PR amd64/148157; it has been noted by GNATS.

From: Garrett Cooper <yanefbsd at gmail.com>
To: Shant Kassardjian <pookme at hotmail.com>
Cc: bug-followup at freebsd.org
Subject: Re: amd64/148157: IPFW in kernel nat BUG found in FreeBSD 
	8.1-PRERELEASE
Date: Sun, 27 Jun 2010 11:27:42 -0700

 On Sun, Jun 27, 2010 at 11:25 AM, Shant Kassardjian <pookme at hotmail.com> wrote:
 >
 >
 >> Date: Sun, 27 Jun 2010 00:04:16 -0700
 >> Subject: Re: amd64/148157: IPFW in kernel nat BUG found in FreeBSD
 >> 8.1-PRERELEASE
 >> From: yanefbsd at gmail.com
 >> To: pookme at hotmail.com
 >> CC: freebsd-gnats-submit at freebsd.org
 >>
 >> Hi Shant,
 >> Please bottom post from here on out.
 >>
 >> On Sat, Jun 26, 2010 at 11:47 PM, Shant Kassardjian <pookme at hotmail.com>
 >> wrote:
 >> > Hi Garrett,
 >> >
 >> >
 >> > I just tried to perform a kernel dump with dumpdev="YES" and had no
 >> > luck, I
 >> > keep getting:
 >>
 >> dumpdev="<blah>" always appears to fail to me as well (contrary to
 >> what others have claimed). Try doing the following after booting up:
 >>
 >> dumpon `awk '$3 == swap { print $1 }'`
 >>
 >> Then you'll be able to reproduce the problem and grab the resulting
 >> kernel core dump.
 >
 >
 > Hi Garrett,
 >
 >
 > I still can't get it to dump. Perhaps it's because my system does not have
 > any swap space allocated?
 >
 >
 > Here's what I added in my /etc/rc.conf to simulate an swap device:
 >
 >
 > dumpdev="YES"
 > dumpdir="/home/crash/"
 > swapfile="/home/crash/swap0"
 >
 > Then I ran:
 >
 > core# sh /tmp/ipfw_test ~
 > 00001 nat 100 ip from any to any via em0
 > ipfw nat 100 config ip 192.168.1.104 redirect_port tcp 172.25.1.1:22 22
 > core# dumpon `awk '$3 == swap { print $1 }'` ~
 >
 >
 > Fatal trap 12: page fault while in kernel mode
 > cpuid = 0; apic id = 00
 > fault virtual address = 0xc
 > fault code = supervisor write data, page not present
 > instruction pointer = 0x20:0xffffffff801d5cd6
 > stack pointer = 0x28:0xffffff8074fbc370
 > frame pointer = 0x28:0xffffff8074fbc620
 > code segment = base 0x0, limit 0xfffff, type 0x1b
 > = DPL 0, pres 1, long 1, def32 0, gran 1
 > processor eflags = interrupt enabled, resume, IOPL = 0
 > current process = 1818 (sshd)
 > trap number = 12
 > panic: page fault
 > cpuid = 0
 > Uptime: 2m36s
 > Cannot dump. Device not defined or unavailable.
 > Automatic reboot in 15 seconds - press a key on the console to abort
 >
 >
 > I'm still getting the "Cannot dump" error message, any idea what I can do
 > next?
 
 Yeah.. you need a swap device otherwise you're not going to be able to
 get a coredump...
 
 >> > Cannot dump. Device not defined or unavailable.
 >> >
 >> > my custom kernel is set to disable:
 >> >
 >> >
 >> > #options KTRACE # ktrace(1) support
 >> >
 >> > #options KDTRACE_FRAME # Ensure frames are compiled in
 >> > #options KDTRACE_HOOKS # Kernel DTrace hooks
 >> >
 >> > must recompile kernel to enable tracing?
 >>
 >> No.
 >>
 >> > I'm currently using the intel pro 1000 chipset / em0 driver, I've been
 >> > experiencing all sorts of network stability problems for a while
 >> > now(ever
 >> > since I upgrade to stable a month ago). It looks like the em0 driver for
 >> > amd64 needs alot of work however a couple of days ago when I recompiled
 >> > my
 >> > box to the latest stable 8.1-prerelease I saw alot of improvments and my
 >> > ipfw/dummynet firewall seems to be running stable with no
 >> > crashes/lockups so
 >> > far...
 >> >
 >> >
 >> > It is very easy for me to replicate the in nat kernel problem, i just
 >> > cant
 >> > get a dump to provide you the additional info.