ipchains / ipmasq still broken in 2.1.130 [Re: Oops in 2.1.129]

Brian Ristuccia (brianr@osiris.ml.org)
Thu, 26 Nov 1998 22:39:42 -0500

On Tue, Nov 24, 1998 at 09:29:24PM -0300, Juanjo Ciarlante wrote:
> Hi Paul!
> On Tue, Nov 24, 1998 at 12:33:52PM +1130, you wrote:
> > In message <19981119212856.A447@osiris.ml.org> you write:
> > > 2.1.129 get a Null Pointer Dereference followed by an oops, followed by a
> > > panic when I enable ip_forwarding under 2.1.129 with masquarading ipchains
> > > rules present.
> > >
> This is due to a bug in ip_masq.c:proto_doff() handling of th->doff.
> Its fixed in pre-2.1.130-2.
>
> Anyway, although now it correctly checks th->doff bounds, it DOESN'T
> change bogus ip_masq behavior when handling fragments (only iff
> CONFIG_IP_ALWAYS_DEFRAG=n).
>

Whatever the problem was, it's still broken in 2.1.130. Null pointer
dereference, oops, kernel panic inside interrupt handler, no sync. Here's my
backtrace:

c016ee14 t cleanup (EIP)
c016ee84 t ip_fw_check
c014fd74 t rtnetlink_fill_ifinfo
c0170448 T ipfw_forward_check
c010bc8c t read_long
c0106000 T get_options
c0100000 t startup_32
c0107310 T cpu_idle
c01075f4 T kernel_thread

pr440fx motherboard, 2.1.130 SMP kernel. Using the onboard Intel
Etherexpress Pro/100 and a 3c59x card. I can swap out the 3c59x if it will
help me work around the problem, but the eepro/100 is built onto the
motherboard. 

-- 
Brian Ristuccia
brianr@osiris.ml.org
bristucc@baynetworks.com
bristucc@cs.uml.edu


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/