remote gdb (and firewire/dcons) not functioning as expected

Hi,

I'm trying to debug the problem introduced by the earlier commit to
remove mbufs from the stag (MT_TAG) over the weekend.

First of all, I tried to use dconschat to open a connection to the GDB
port over firewire. This didn't work. dconschat complained it couldn't
find the CROM entry on the system under test. Manual inspection of the
CROM entries via fwcontrol(8) revealed that it was indeed present, and
the EUI64 was typed correctly.

Now, when I try to use 9600 baud serial (over COM1 on both machines) as
a fallback, I get the following results (after earlier being able to
set a breakpoint in ip_output() before the box went multiuser):-

(kgdb) target remote /dev/cuaa0
Remote debugging using /dev/cuaa0
0xc0522d81 in ip_output (m=0xc10ecf, opt=0x0, ro=0x0, flags=0, imo=0x68000000, 
    inp=0x38c28d01) at ../../../netinet/ip_output.c:139
139             int hlen = sizeof (struct ip);
warning: Unable to find dynamic linker breakpoint function.
GDB will be unable to debug shared library initializers
and track explicitly loaded dynamic code.
warning: shared library handler failed to enable breakpoint
(kgdb) step
Can't send signals to this remote system.  SIGSEGV not sent.

Program received signal SIGSEGV, Segmentation fault.
0xc0522d81 in ip_output (m=0xc10ecf, opt=0x0, ro=0x0, flags=0, imo=0x68000000, 
    inp=0x38c28d01) at ../../../netinet/ip_output.c:139
139             int hlen = sizeof (struct ip);
(kgdb)

The target system and my laptop (where I run gdb) are both seeing the
same kernel image, which I NFS-mount from my build server.

All boxes are running fresh -CURRENT from Saturday.

Any ideas? This is most perplexing...

BMS